Two of KPMG International’s most senior leaders arrived in Australia this week. Global chairman Bill Thomas and global COO Gary Wingrove are meeting partners and clients. The stated purpose: resetting “its approach to governance and culture.”

The second name is the one that should draw attention.

Gary Wingrove was CEO of KPMG Australia from 2013 to 2021. KPMG’s own announcement of his election as the next global chairman and CEO, effective October 1, credits him with “nearly doubling the firm’s revenue, profitability and headcount while reshaping its culture” during that period. He is, by the firm’s own account, the architect of modern KPMG Australia.

The culture he reshaped is the one now under ASIC investigation. Revenue he doubled is contracting under a nationwide government ban. His firm is preparing to cut a thousand jobs and slash partner pay by a fifth.

Wingrove left Sydney for London in late 2021, was appointed global COO, and in March was elected to succeed Thomas as global chairman and CEO. He takes over in October. Between now and then, the firm he built needs rescuing, and KPMG International has sent him to help provide it. The timing is extraordinary: the incoming leader of the entire KPMG network is flying into the jurisdiction where the most serious crisis in the network’s recent history is unfolding, and the crisis grew in the soil he tended.

To be clear: we do not know what Wingrove knew, when he knew it, or whether any of the alleged misconduct occurred with his awareness. What we know is that the systems of oversight, the whistleblower processes, the internal investigation protocols, and the partner accountability mechanisms that failed so comprehensively were built or operated during his eight-year tenure. KPMG’s own language for what he did is “reshaping the culture.” When a firm’s culture is under investigation, the person who shaped it owes an accounting of what was shaped and why.

What has happened since the CEO appointment

For readers following the KPMG Australia story, the pace of developments since we last wrote has been relentless. Here is what has changed.

The firm expelled former chief operating officer Eileen Hoggett from the partnership last week. Six weeks earlier, KPMG had announced that Hoggett would “step aside” and retire on her own terms. Then an external investigation by Allens confirmed that she had stored printed copies of confidential Lendlease board papers in a locker at KPMG’s Sydney office, shared them with colleagues, and lied about it when asked. New CEO John Sams wrote to the parliamentary committee acknowledging the whistleblower’s persistence and apologising for a “serious breach of trust.”

The locker allegation was one of the whistleblower’s original claims. KPMG ran three internal investigations into his allegations. All three “failed to substantiate any wrongdoing.” An external law firm, retained only after the CEO resigned and the chairman departed, found the evidence in weeks. We have written before about the distinction between an investigation that fails and an investigation designed to fail. KPMG’s own CEO has now confirmed which category these belonged to: “It is unacceptable that it has taken this long for the truth to come out, and that it has come to light in this manner.”

Victoria this week became the latest jurisdiction to bar KPMG from new government work, joining the Commonwealth, New South Wales, Queensland, Western Australia, and the ACT. The firm has signed a deed of undertaking not to bid until September 30. KPMG’s disclosed Victorian contracts total A$24 million; broader consortia involving the firm account for A$195 million more. The nationwide lockout is now complete. For a firm whose consulting revenue depends heavily on the public sector, client revenue has been switched off, state by state, over six weeks.

Macquarie Group used its annual general meeting last week to announce a formal inquiry into KPMG’s capacity to deliver its audit. Chair Glenn Stevens, who at last year’s AGM had dismissed questions about the KPMG appointment as “silly talk,” announced reviews of both KPMG’s ongoing capability and the integrity of its involvement in Macquarie’s audit tender process. A shareholder noted that Deloitte and EY had been eliminated early from the tender, leaving KPMG as the sole remaining candidate. That audit is reportedly worth approximately A$30 million per year. Westpac’s mandate, at roughly A$32 million, is also under review. Lendlease ended a 68-year audit relationship in June.

Inside the partnership, the numbers tell a consistent story. Partner distributions for FY26 face cuts of up to 20 per cent, with the AFR reporting average losses of A$717,000 per partner. Up to 1,000 staff may be cut from a workforce of roughly 9,000. Dozens of Canberra-based partners, whose practices depended on the government work now frozen, are reported to be looking for the exit. KPMG International has reportedly intervened to slow those departures, concerned that a wave of partner exits during audit season would leave the firm unable to meet its obligations to major clients.

The firm carries approximately A$557 million in debt with revenue covenants, and revenue has already declined from A$2.55 billion in FY23 to A$2.32 billion in FY25. If the government bans persist, the major audit mandates are pulled, and the partnership continues to haemorrhage experienced people, the revenue covenant question becomes a live one.

The KPMG International problem

Wingrove’s visit carries a complication that the firm’s reset narrative cannot accommodate.

The whistleblower did not only raise concerns with KPMG Australia. He contacted Bill Thomas and KPMG International’s Sydney-based global general counsel, Anne Collins, in May 2025, approximately a year after first raising them internally. KPMG International initially decided not to investigate. It retained Freshfields, and the magic-circle law firm endorsed the position that the global body could not launch a probe due to insufficient information from the whistleblower. KPMG International maintained throughout that its role was “oversight, not control” of member firms.

We have heard this before. When KPMG South Africa’s Gupta scandal erupted in 2017, KPMG International maintained the same line for months before eventually sending in executives and cutting hundreds of jobs. Australia is following the South African playbook with a lag of weeks rather than months. Under the independent-member-firm structure, accountability disappears into the gap between global and local: the global body sets standards but disclaims responsibility; the local firm follows its own governance but operates under the global brand.

A parliamentary hearing is scheduled for 14 August. KPMG International figures, including Collins and Wingrove, could be called to give evidence. The Treasury consultation on whether to break up the Big Four’s audit and consulting arms, or cap partnerships at 400, closes on 12 August.

The October question

The question nobody at KPMG International appears to be asking publicly is whether the global CEO appointment itself needs re-examining.

Wingrove was put forward by the global board in early March 2026. The whistleblower’s allegations were read into the Australian parliament on March 24. He was formally elected by the global council in mid-March. Either the board knew the Australian scandal was about to become public and elected him regardless, or it did not know what was happening inside the firm whose culture its candidate had spent eight years building. Neither answer is comfortable.

October 1 is two months away. Every week between now and then adds new confirmed evidence of failures that occurred during Wingrove’s tenure. If the parliamentary committee calls him on August 14, he will be asked what he knew about the information-sharing practices, the investigation protocols, and the partner accountability mechanisms at the firm he ran for eight years. KPMG International will need an answer that has been prepared before the question is asked, and that answer cannot be “I was in London by then.”

We are not suggesting Wingrove is personally implicated in the misconduct. We have said clearly that we do not know what he knew. But consider the standard KPMG would apply to its own clients. If a company were about to install a CEO whose eight-year tenure at a subsidiary coincided with the period now under regulatory investigation, KPMG’s own governance advisory practice would recommend an independent review of that tenure before the appointment took effect. The standard the firm sells to clients is the minimum it should apply to itself.

For the Australian partners watching this visit, the signal is hard to misread. Their income is dropping by A$717,000. Their colleagues in Canberra are looking for exits. Their former COO has been expelled for conduct that three internal investigations failed to find. And the person arriving to reset the culture is being promoted to lead the global network on the strength of a tenure that produced the culture under investigation. The system that created the failure is the same system that promotes from it.

Gary Wingrove may be the right person to lead KPMG globally. But “may be” is not a standard that survives the scrutiny this firm is under. Before October 1, KPMG International’s global board should commission an independent examination of his Australian tenure, conducted to the standard the firm would recommend to any client in the same position. If KPMG’s own governance logic does not demand that, the logic is broken.

Wingrove’s tenure at KPMG Australia (2013-2021) produced a firm that nearly doubled in size. The misconduct, the investigation failures, and the whistleblower retaliation all occurred during or shortly after that period. Should KPMG International’s global board commission an independent review of his tenure before he takes over on October 1? If a client asked KPMG the same question about their own incoming CEO, what would KPMG recommend?

One of our contributors once caught a polling company fabricating respondents.

The firm had commissioned its annual industry survey, the same one it had published for years, the same one journalists and analysts quoted as if it were peer-reviewed research. The polling company, a reputable outfit with a long client list, claimed to have surveyed a specific number of C-suite executives in a specific sector in a specific geography. Our contributor knew that sector and geography well enough to count the actual population. The numbers did not work. There were not enough people matching the claimed demographics in the claimed location for the sample to exist.

So our contributor did something nobody was supposed to do. He contacted several of the people who should have been in that sample, people he knew personally, people who fit the exact profile. None of them had been contacted. None had received an outreach. None knew the survey existed.

When confronted, the polling company retreated behind professional discretion. Their methodology was proprietary. Their respondent lists were confidential. They could not disclose who they had spoken to or how they had assembled the panel. They did not deny the accusation directly. They simply declined to address it.

Our contributor raised this with the firm’s marketing leadership. The response was three words: leave it alone.

The survey published on schedule. The press release went out. Journalists cited it. Clients referenced it in board papers. An analyst at a bank used it in a sector report. The numbers entered the information supply chain and became, for all practical purposes, true.

We wish this were unusual. Across 170 combined years inside these firms, not one of us can identify a single survey or report in the last fifteen years that followed every step of the process it claimed to follow. Not one.

How the sausage actually gets made

The public imagines that a Big 4 or MBB report begins with a question and arrives at an answer through research. The actual sequence runs in the opposite direction.

A partner or a marketing team decides what the report should say. The conclusion comes first. It is usually a market-sizing figure that positions the firm’s service line favourably, or a trend narrative that creates urgency for the type of engagement the practice sells. The research is then commissioned to produce evidence that supports the predetermined conclusion.

This is not a secret within the firms. The partners who lead these projects would describe it differently. They would say they are “testing a hypothesis” or “validating a market perspective.” The distinction between testing a hypothesis and confirming a conclusion is supposed to be methodological rigour. In practice, the rigour is absent.

We watched it work the same way across multiple firms, multiple geographies, multiple decades. The pattern was the same everywhere.

Start with the sample. When survey responses came back and the data contradicted the messaging, the team would identify “outliers” and remove them. The justification was always ad hoc and always arrived after the inconvenient data appeared. Nobody decided in advance which responses would constitute outliers. The decision was made retroactively, by the people who needed the data to say something specific.

Then the mathematics got creative. One of the most common abuses we witnessed was averaging percentages. If 30% of respondents in one segment and 50% in another reported a particular behaviour, the report would state that “40% of executives” exhibited that behaviour, regardless of segment size. When one of our contributors flagged this as statistically invalid (you cannot average percentages across groups of different sizes without weighting), the response was genuine confusion. The people assembling the report did not understand why it was wrong. They had been doing it for years.

The questions did the rest. Survey questions were structured to produce the desired result. Leading phrasing, anchoring effects, forced-choice options that excluded inconvenient answers. A question like “How significant is AI to your organisation’s strategy?” with options ranging from “significant” to “extremely significant” will reliably produce a headline claiming that “95% of executives consider AI significant to their strategy.” The option of “not significant at all” was either absent or buried.

And through all of this, the teams went through the motions of analytical process. They ran models. They produced charts. They wrote methodology sections. What they produced was, in most cases, no better than fabrication wearing a lab coat. But the appearance of rigour served a purpose: it gave the marketing team something to point to when anyone asked how the numbers were derived.

The supply chain of manufactured confidence

The polling company incident our contributor described was not an aberration. It sits inside a supply chain that has financial incentives at every link to produce the answer the client wants.

Consider how the commissioning works in practice. Firms hire external polling companies. Those polling companies know that repeat business depends on delivering results the client can use. A polling company that consistently returns data contradicting the firm’s preferred narrative does not get hired again. A polling company that reliably delivers clean, quotable numbers supporting the firm’s thesis gets a multi-year contract.

This incentive structure is not unique to consulting. But consulting adds a complication: the firms commission the research, control the questions, shape the sample criteria, and then publish the results under their own brand with no external review. The polling company provides plausible deniability. The firm can say the research was “conducted by an independent third party” while having determined the outcome before the first question was asked.

In April 2025, the U.S. Department of Justice indicted eight people connected to market research firms Op4G and Slice for a decade-long scheme to fabricate survey data worth $10 million. The defendants recruited “ants” who posed as legitimate survey respondents, were coached on how to answer screener questions, and used VPNs to conceal their locations. Their clients included Google and Seattle Children’s Hospital. Healthcare decisions may have been based on entirely fabricated data.

That case is the criminal extreme. But the academic evidence suggests the everyday reality is not dramatically better. Researchers Noble Kuriakose and Michael Robbins analysed over 1,000 public datasets from international surveys and found that roughly one in five failed a statistical test for fabricated data. NORC at the University of Chicago reported in early 2026 that 40 percent of nonprobability survey interviews in 2025 were likely fraudulent, driven by click farms, bots, and professional survey takers. The Insights Association has called survey fraud an “existential threat” to the industry.

The CDC bleach study illustrates what happens at the other end of this pipeline. In 2020, the CDC reported that Americans were ingesting household cleaners to prevent COVID-19. The finding was reported by over 150 news outlets. When researchers replicated the study with proper quality screening, they found that 100% of the reported bleach ingestion came from problematic respondents: bots, inattentive clickers, and professional survey takers who answered “yes” to questions like “have you ever suffered a fatal heart attack?” The alarming public health finding was entirely an artefact of bad data from an unvetted panel.

Nobody at the CDC fabricated anything. They contracted a market research vendor, the vendor sourced from a common panel supplier, and nobody checked whether the respondents were real. The same supply chain that fills Big 4 survey panels.

The reports get worse. The citations keep coming.

In 2024, accounting professors Jeremiah Green and John Hand published a replication study of McKinsey’s “Diversity Wins” research in Econ Journal Watch. McKinsey’s series of reports, published in 2015, 2018, and 2020, claimed a correlation between executive diversity and financial outperformance. The studies were cited by BlackRock, the Nasdaq, and U.S. government agencies. Green and Hand could not reproduce the claimed correlations. They identified a reverse-causality flaw in McKinsey’s design: McKinsey measured diversity after the performance period, not before. McKinsey refused to share its datasets. When asked, the firm said it stood by its findings.

The diversity studies are worth examining not because diversity is unimportant (it may well drive outperformance through mechanisms McKinsey did not test), but because they illustrate how consulting research enters the policy bloodstream. A non-peer-reviewed study, produced by a firm with a commercial interest in diversity consulting, using a methodology that independent academics could not replicate, influenced board composition rules at major stock exchanges and investment criteria at the world’s largest asset manager. At no point did anyone with the authority to mandate these changes require the underlying data.

McKinsey’s market-sizing projections follow a similar pattern. The firm’s Quantum Technology Monitor projected in April 2026 that quantum computing could generate $400 to $600 billion in value for financial services alone by 2035. A detailed critique published at postquantum.com (which we came across independently and recommend reading in full) showed that the arithmetic on McKinsey’s own slide cannot be reproduced from the assumptions displayed on it, that the impact estimates measure quantum computing and AI together without allocating between them, and that the finance chapter ignores peer-reviewed resource estimates, including two papers from Goldman Sachs’s own researchers, showing the hardware requirements are orders of magnitude beyond what the industry expects to deliver in that timeframe. The author’s conclusion: a headline figure whose published inputs produce a different figure is a graphic, not a forecast, and nobody should commit capital on its basis.

These are “value-at-stake” estimates, a construction McKinsey uses frequently. The phrase sounds technical. What it means is: “this is the total amount of money that could theoretically be affected if every assumption we made turns out to be correct.” It is not a forecast. It is not a projection. It is a ceiling so high that the actual outcome will always fall somewhere beneath it, and the firm can always claim the estimate was directionally correct.

The consulting firms produce these numbers for industries where they sell advisory services. McKinsey publishes AI market-sizing reports while selling AI consulting engagements. The Big 4 publish cybersecurity threat reports while selling cybersecurity services. The conflict of interest is so consistent and so visible that the only explanation for its persistence is that nobody with enough influence cares to challenge it.

Then came the AI reports

If the pre-AI era involved humans cutting corners on research methodology, the AI era has introduced a new efficiency: letting the machine fabricate the citations while the humans fabricate the conclusions.

In 2025 and 2026, GPTZero and academic researchers exposed AI-fabricated content in reports from three of the four Big 4 firms. Deloitte Australia refunded part of a A$440,000 government contract after its welfare-compliance report was found to contain fabricated citations, non-existent academic papers, and a made-up Federal Court quote. Deloitte Canada’s $1.6 million healthcare review for Newfoundland and Labrador contained false academic citations and researchers misattributed to studies they never wrote. EY Canada withdrew a cybersecurity report after GPTZero found that 16 of 27 cited sources were fabricated, including a citation to a McKinsey report that does not exist. KPMG pulled a customer-experience report in which only 5 of 45 citations were accurate.

The irony is almost too precise. These firms sell AI transformation to their clients. They charge millions for advice on how to implement AI responsibly. And they cannot manage the integrity of their own AI-generated content.

But here is the part the coverage misses: the AI reports are not a new category of failure. They are the old failure accelerated. The methodology was already broken. The citations were already decorative. AI simply made the decoration faster and the fabrication more obvious. A report that starts from its conclusion and works backwards does not become honest because a human, rather than a language model, selects the supporting references. It just becomes harder to catch.

Goodhart’s Law with a consulting fee

A reader named Claudine Cassar recently shared an account that captures a parallel phenomenon. KPMG’s U.S. advisory business reportedly told staff to register AI use on 75% of business days. The predictable result: employees started typing random questions into the system, summarising emails they had already read, and automating prompts simply to meet the usage target. The dashboard showed AI “usage” rising. Press releases followed. The metric created the appearance of adoption while revealing nothing about whether the technology improved anyone’s work.

This is the same mechanism applied to different content. The survey that starts from its conclusion. The market-sizing estimate that defines its terms broadly enough to guarantee an impressive number. The AI adoption metric that measures keystrokes instead of outcomes. EY’s definition of “AI-related revenue” is instructive here: the firm counts everything from enterprise-wide transformations to AI governance frameworks, a category broad enough that adding a chatbot module to a broader programme makes the entire engagement “AI-led.” When you have spent $2.4 billion and need to show a return, every practice leader has an incentive to relabel existing work as AI-adjacent.

The reported numbers are often technically defensible and practically meaningless. “90% of EY professionals have completed foundational AI training” sounds impressive until you learn the metric is “awarded or initiated,” not completed. “161,000 AI badges” could mean 161,000 people who clicked through an online module. We sat in the partner meetings where these metrics get constructed. The pressure to demonstrate ROI on AI investment is real. The easiest way to demonstrate ROI is to redefine what counts as AI revenue until the growth rate looks impressive enough for the press release.

What to do when you read the next one

We are not suggesting that every number in every Big 4 or MBB report is fabricated. Some of the underlying research is competent. Some of the analysis is worth reading. The problem is that there is no reliable way for an outside reader to distinguish the competent research from the manufactured kind, because the same brand, the same formatting, and the same authoritative tone are applied to both.

These reports are, in library-science terms, “grey literature,” material produced outside traditional academic or commercial publishing that is not subject to peer review. They acquire authority through repetition: a McKinsey estimate gets cited by a journalist, picked up by an analyst, referenced in an academic paper, and trained into a language model. By the time it reaches a board presentation, the number has been laundered through enough credible-seeming intermediaries that nobody traces it back to a marketing team that decided what the report should say before the research began.

So treat them accordingly.

When a Big 4 or MBB report sizes a market, check whether the firm sells services into that market. If it does, the estimate is advertising. Read the methodology section, if one exists, and check whether the stated assumptions actually produce the headline number. (In the McKinsey quantum case, they do not.) When a report claims survey results, look for the sample size, the sampling method, the response rate, and the exact question wording. If any of these are missing, the results are decoration.

And if you see a statistic from a consulting firm cited in an academic paper, a news article, or a board pack, ask one question: has anyone outside the firm that produced it attempted to reproduce the finding? If the answer is no, the number has never been tested. It has only been repeated.

We built these reports. We sat in the rooms where the conclusions were written before the research was commissioned. We watched polling companies deliver impossible samples and marketing teams suppress inconvenient data. We averaged the percentages. We wrote the methodology sections that nobody read. We knew, and we said nothing, because the reports served their purpose and the purpose was never truth.

The purpose was always the same: give the market a number it could cite, attach our brand to it, and wait for the phone to ring.

What is the worst example of a manufactured number you have seen in a Big 4 or MBB report? If you work at one of these firms now, what does the process actually look like from the inside? The comments are open, and so is the inbox.

The question arrives at dinner parties, always from the person seated to your left, usually between the starter and the main, and always with the same subtext: I want to know if it was worth it. The short answer is that we do not know. The honest answer takes longer, and it has no LinkedIn gloss on it.

The day

You wake at 4:30. Not because you are a morning person. Because the window between 6 and 8:30am, before the meetings begin, is the only uninterrupted block you will get, and your inbox has accumulated 40 emails since you last checked it at 11pm. Some of these require decisions. Several are marked urgent by people whose threshold for urgency you have learned not to trust. You clear as many as possible. One is from a client whose subject line reads “Quick question” and whose body contains seven paragraphs, two of which contradict each other. You flag it for the mythical “later” that does not exist.

At 8:30 the meetings begin. They do not stop.

Your calendar today contains seventeen accepted meetings. That does not count the ones your PA or you already rejected. These are the ones that survived the cull. For some of them you will wait until the last possible minute to cancel or postpone, buying yourself fifteen minutes you will spend not thinking but triaging. For others, you will dial into two simultaneously, one in each ear. These are the meetings whose organisers do not need your input; they need your name on the attendance list so they can say that you were present when the decision was made. You listen to both with one ear each, hoping to catch the moment anyone addresses you directly, while using your hands to respond to emails. You are also hoping, with the quiet dread of experience, that you have not implicitly agreed to something by failing to object to it in real time. This has happened before. It will happen again. You will discover what you agreed to in approximately three weeks, when someone references “the decision the leadership endorsed” in a meeting you do not remember attending because you were also in another meeting at the time.

The practice leadership call runs ninety minutes and covers six items, four of which could have been an email and two of which will generate a follow-up meeting to discuss what was discussed. But this is not your only leadership meeting. You also sit on the country leadership, the regional leadership, and three industry leadership groups because your practice spans sectors. Financial services wants your pipeline numbers. Energy wants your pipeline numbers. Government wants your pipeline numbers. You have twelve senior partners, each of whom considers themselves your primary reporting line, each of whom wants to see your financials, your forecast, and your narrative for why the forecast has not improved since last week. The narrative has a half-life of seven days, after which the next review cycle requires a fresh one.

Somewhere in the mid-morning, if you are lucky, you get a client meeting. This is the part of the job that actually resembles what you told yourself the job would be. The client has a problem. You have relevant experience. For ninety minutes you are a consultant in the way the word was meant: engaged, useful, specific. These minutes are the best minutes. They are a decreasing proportion of your week, and you are aware of this while they are happening, which makes them worse.

By the afternoon, your inbox has refilled. Your teams are sending messages marked URGENT because a proposal needs your sign-off by 1pm, a deliverable needs your approval before it ships to the client, a commercial agreement needs your signature before the procurement deadline. You are supposed to read these documents. As a partner and practice leader, your name on them carries legal and commercial weight. Even after years of doing this, you still harbour a faint hope that this will be the deliverable you actually read before signing. It never is. Your senior manager finally catches you in the corridor between meetings — or, more accurately, outside the washroom at 12:57pm, the first time you have managed to go to the washroom while walking between meeting rooms — and you sign on your phone without reading past the executive summary. The executive summary was also written by someone who did not read the full document.

The 7pm call with the offshore team runs long because it always runs long. You carve out an hour with your spouse somewhere between 8 and 9. The 9pm call with the client who “just wants five minutes” runs fifty. At 11pm you reopen the laptop and apply the zero-inbox protocol. A few emails get immediate responses. Several get forwarded to team members with a one-line instruction that you hope is sufficient because you do not have time to check whether it is. The remainder you triage by sender: key clients first, then your twelve reporting lines, then everyone else. "Everyone else" is a large category. Their emails, the ones you have not responded to and cannot respond to and will not have time to respond to tomorrow because tomorrow's calendar already contains nineteen accepted meetings, you delete. Not with malice. With arithmetic. Your inbox is now at zero. The operating theory, which you have never stated aloud but which governs your evenings, is that if someone genuinely needed your input, they will send the same email again in a few days. Most do not. You choose not to examine what that implies about how necessary your input was in the first place.

The inbox does not get quiet at 5pm. The inbox does not get quiet.

The parts they do not put on the recruiting site

The meeting paradox is the one we want to name, because it reveals the operating logic of the partnership better than any org chart.

Whenever you most need time to do actual work, you get the least of it. A large opportunity appears: you should be at the client’s office, understanding their problem, building the proposal. Instead, every senior partner who is even tangentially connected to the client, the sector, or the capability wants a meeting. They want to contribute. More precisely, they want to be seen to have contributed, because if the bid succeeds, they would like their name attached to the win, and if the bid fails, they would like to have been in a meeting where they offered wise counsel that was regrettably not followed.

One of us counted: on a single major bid, fourteen partners held fourteen separate meetings about the opportunity. Two of them had met the client. The others contributed by offering strategic observations unrelated to the client’s actual problem, scheduling follow-up meetings to discuss their unrelated observations, and subsequently claiming partial credit for the win. Some were shameless enough to tell the practice leadership that they had sold it, on the basis that they had attended one of fourteen meetings. (The client, for the record, chose us because of the proposal we wrote at midnight after the fourteenth meeting.)

This was politics, but not the dramatic kind people imagine. No shouting matches, no sabotage. Calendar warfare. Credit allocation. Positioning. You never directly offended a peer. They might become your boss, or you might need their vote when you wanted to become theirs. So you never offended them. You screwed them over in much more subtle ways, and they did the same to you, and everyone understood the rules. Some partners did nothing except play this game, and they were not punished for it, because the system could not distinguish between a partner who attended twelve meetings and contributed nothing and a partner who attended twelve meetings and shaped the outcome. It measured attendance. We will give the politics their own piece in this series; they have earned the space.

The unit statement landed quarterly, and it deserves its own piece in this publication (it will get one). You could not decode it. A doctorate, or the professional-services equivalent, did not help. You could not reverse-engineer your own compensation from the document your firm provided. Ask a more senior partner to explain and they could not either, but they had been receiving the statement long enough that they had stopped asking.

People management consumed hours every week and resolved nothing. Team conflicts, underperformers, hiring disputes, two senior managers who could not occupy the same room. All of it escalated to the partner. And the partner, no matter how clear the situation, could say nothing of substance, because one wrong sentence would produce an HR investigation or a legal letter. So you delivered the same empty phrases you remembered from the last HR training, and the problem persisted, and the next conflict escalated to you the following week. Every time, the person sitting across from you carried the same implicit hope: that this time, against all historical evidence, the partner would take a side and resolve it decisively. You never did. You could not afford to.

The thought-leadership machine was its own species of time destruction. Marketing wanted whitepapers, conference appearances, social media engagement. Occasionally you got excited about a paper and drafted something over several nights, carving out midnight to 1am, that you were proud of. Then the review process began. Every partner whose practice, industry, or technology area was even remotely adjacent wanted input. Local partners. Global partners. Legal. Marketing. The writer hired by marketing to rewrite what you wrote. The social media team planning the promotion. One of us counted (we are people who count things): one whitepaper, nine months, 130 meetings. The finished paper, after every partner had sanded off every edge, said the equivalent of “AI is coming and you should prepare.” It used considerably more words than that, but the intellectual content could fit in a tweet. One of us declined to put their name on the final version. It did not stop the other contributors from promoting it on LinkedIn as an amazing paper, written by an amazing team, at an amazing firm. Three amazings. That is the minimum for a whitepaper. Two amazings would suggest you had reservations.

The firm also ran a permanent rotation of initiatives. Women in tech. A technical career ladder. Improving the knowledge database (for the estimated 150th time in the firm’s history). New office lighting that would show clients how technologically advanced the firm was and would improve team performance by 50%. Every director or partner who led one of these treated it as a pet project and held weekly showcases. Not quarterly. Weekly. And you attended, and you said it was excellent work, because the person behind the initiative might one day hold a position that affected your career, and you never offended a peer if there existed a subtler alternative. Which there always did.

With all of this (the pipeline theatre, the leadership calls, the credit-allocation politics, the initiative attendance, the sign-offs you could not read, the people problems you could not solve, the talent calibrations, the fire-fighting, the firm committees you volunteered for just to remain visible), you did not have time to do the things the partnership was supposed to be for: becoming a genuine expert, building real client relationships, developing an industry presence, thinking. The partnership track was designed to produce those capabilities. The daily reality of the role left no room for them.

The parts that were genuinely good

Honesty cuts both ways, and if this section were missing, the rest would be dishonest.

The calibre of your colleagues at their best was extraordinary. A senior manager who produced a 40-page analysis overnight because the client changed the question at 5pm, and the analysis was good. The partner across the hall who knew more about pharmaceutical supply chains than anyone in the industry, including the people running them, and who would spend hours walking you through it if you asked. A junior who pushed back on your recommendation in front of the client because they had found a flaw in your logic, and you let them, because the flaw was real and catching it in the room saved the engagement. The system that used to produce these people was superb; we have written about what happened to it.

The access was real. You sat in rooms where decisions were made that you read about in the Financial Times the following week. When a client’s CEO asked what you thought, they meant it.

The moments the work mattered. Not every engagement, not even most, but enough of them. A restructuring that saved 2,000 jobs because the alternative proposal would have closed two plants. A compliance programme that caught a problem before the regulator did. The client who called you two years later to say that the thing you recommended had worked. Those calls arrived rarely enough that each one stayed with you, and frequently enough that they kept you from leaving when the Sunday-evening dread suggested you should.

What the gap reveals

The question from your dinner companion, the one about whether it was worth it, has no single answer because the job contained two jobs.

One was the consulting work: the clients, the problems, the colleagues, the access. That job was frequently superb, and if it had comprised 80% of the role, most of us would still be doing it.

The other was the apparatus: the seventeen daily meetings, the dual-dialling, the documents you signed without reading, the pipeline narratives, the credit-allocation politics, the thought-leadership machine that produced no thoughts, the people management you could not manage, the initiative showcases you attended weekly and called “amazing.” That job expanded every year we were in the role. The balance tipped. By the time we left, the apparatus was the job, and the consulting happened in the gaps between meetings, if it happened at all.

We do not say this with bitterness. We did this job for a long time and chose to stay for most of it. Some of us miss it, or miss the version of it that existed when the ratio between the two jobs was different. All of us recognise that we are describing a period that is not coming back, which is why we left and why this publication exists.

Current partners: what is the biggest gap between what you thought the job was and what it turned out to be? The comments are open, and so is the inbox.

The partner track, at its best, was one of the most effective professional development systems ever designed. We say this as people who went through it.

You started as an analyst or associate, doing the technical work: building models, checking data, learning the grammar of a client’s business from the ground up. After a few years you moved into client management, running workstreams, translating between what the partner promised and what the team could deliver. Then came business development: learning to listen for the problem behind the problem, to earn trust over years rather than pitch decks, to build a network of relationships that would eventually, if you were good enough and patient enough, produce revenue. And finally, if you survived all of that, you reached the leadership layer: setting direction for a practice, mentoring the next generation, making judgment calls with incomplete information under real pressure.

Each stage took two to four years. Each had genuine feedback mechanisms, brutal ones sometimes. Partners who had spent decades in the same practice area sat across from you during your admission interview and asked questions designed to expose the gap between what you knew and what you claimed to know. You could not bluff a panel that had seen every variation of bluff the profession could produce. (You could, however, watch someone try. Those twenty minutes felt longer from every seat in the room.)

The partners this system created were exceptional across multiple dimensions. Technical depth. Client instinct. Commercial awareness. The ability to walk into a boardroom and add value that the board could not generate internally, because the partner had spent fifteen years building a perspective the board’s own executives did not have the vantage point to develop. We witnessed this. We participated in it. We benefited from it.

That is not nostalgia. That is what happened.

What also happened

The degradation of this system over the past fifteen years has been thorough and specific, and we want to describe it precisely, because the lazy explanation (”the firms got too big”) obscures the actual mechanism.

The firms did get big. But size alone does not destroy a development track. Law firms got big. Investment banks got big. Medical training programmes scaled enormously. Some of those institutions maintained rigorous professional development standards through periods of rapid growth. The Big 4 and MBB did not, and the reason has less to do with headcount than with a shift in what the market was willing to pay for.

Until roughly 2010, the partner had to bring deep domain knowledge. Not the kind you can acquire from three analyst reports and a briefing note on the flight over. The kind that takes a decade to build. You needed a genuine network of contacts who could help solve client problems, and not just your contacts: the regulator who would take your call because you had earned credibility with their office over years of honest engagement. The industry CEO who would give you a reference check on a potential hire because you had given them the same courtesy when they needed it. The academic who had published the definitive paper on a niche technical question and would walk you through the implications over dinner because you had sent them three referrals for interesting consulting engagements last year.

These relationships could not be manufactured on a pipeline. You developed them over years, sometimes a full decade, before they produced a single dollar of revenue. Try explaining that timeline to a modern practice leader with quarterly origination targets. You had to be genuinely knowledgeable about your clients’ industries because the client was buying you: your judgment, your experience, your capacity to see around corners they could not see around themselves.

When buyers stopped buying partners

The incentives shifted when buyers increasingly started purchasing the Big 4 for brand rather than for the individual partner.

The mechanism is worth tracing. For decades, a client hired McKinsey or Deloitte because a specific partner had demonstrated specific expertise on a specific problem. The partner’s reputation was the sales channel. The firm’s brand mattered, but it mattered as a credentialing wrapper around the individual. “Sarah at PwC” was the pitch. The firm’s logo was the envelope.

Sometime around 2008 to 2012, this began to invert. Procurement departments professionalised. Panel arrangements replaced relationship-based hiring. Clients started buying “Deloitte” or “McKinsey” as an institutional product rather than buying a specific partner’s judgment. The reasons were rational: corporate governance was tightening, audit committees wanted defensible vendor selection processes, and boards were asking tougher questions about how advisory mandates were awarded. And, candidly, there was an insurance dimension: “we hired McKinsey” protects a CEO’s career if the strategy fails in ways that “we hired Sarah” does not.

(We will examine that insurance dynamic in detail in a later piece. It is one of the most corrosive forces in the industry.)

Once the buyer was purchasing the brand rather than the partner, the partner’s job description changed. Deep knowledge became less relevant, because the client was not paying for depth. The decade-long relationship network became less relevant, because procurement was selecting from panels, not Rolodexes. The patient cultivation of industry expertise became less relevant, because the pitch was now about the firm’s “proprietary methodology” and “global reach,” not about the partner’s personal track record in chemicals or financial services or healthcare.

What became relevant was the ability to hustle.

The new partner

We use the term descriptively, not disparagingly. We all know them. Some of us trained them. Some of us became them in our later years, or at least adopted enough of the new playbook to remain competitive.

The partner who succeeds in the post-2012 model needs a different skill set. Aggressive sales capability: the willingness and the talent to chase pipeline relentlessly, to respond to RFPs within hours, to work the internal system to get their proposals prioritised over competing bids from colleagues in other practice areas. (At one firm, we watched a partner submit fourteen proposals in a single quarter. He won two. The metrics system counted it as strong origination activity.) Comfort with the leverage model: the ability to scope engagements so that the firm’s margin targets are hit, which in practice means staffing with the most junior people the client will tolerate. Speed: the market rewards partners who can close quickly, not partners who spend six months understanding a client’s industry before proposing a solution. And an instinct for internal politics — the navigation of partner elections, the management of alliances, the careful positioning during governance reviews — that determines survival as much as any commercial metric.

(We have described what that daily reality looks like in practice. Seventeen accepted meetings, one washroom break, and a zero-inbox protocol that relies on the assumption that if someone genuinely needed your input, they will send the email again.)

Almost all of the skills that defined partnership excellence a generation ago became irrelevant or less relevant once the market stopped paying for them. The depth. The patience. The intellectual curiosity that drove a partner to read an academic paper on a Saturday morning because it might inform a client conversation three months later. The willingness to say “I do not know, but I know someone who does” instead of projecting confidence on a topic you studied for twenty minutes in the taxi.

The shift was not instantaneous. It played out over a decade, unevenly across firms and practices. Audit partnerships resisted it longer than advisory ones, partly because the technical knowledge requirements in audit are harder to fake (regulators will catch you) and partly because the audit partner’s client relationship is anchored in a statutory requirement rather than a discretionary purchase. Advisory and consulting partnerships adopted the new model faster because advisory revenue is purely discretionary, and discretionary buyers are more susceptible to brand-based selling.

But by 2020, the transformation was largely complete across all the major firms. The partner admission criteria at most Big 4 and MBB firms now weight origination (the ability to sell) more heavily than any other factor. Technical excellence, client satisfaction, team development, intellectual contribution: all still present on the scorecard, all subordinate to the revenue number.

What we see now

We can describe the before and the after with precision, and what we see now in the partnership cohorts at these firms is a generation of leaders who are superb at selling and managing upward, and who lack the technical depth, the client intimacy, and the intellectual curiosity that used to be the minimum entry requirement for the role.

This is a pattern we recognise, not an accusation we throw from a distance. Among the contributors to this publication, some were “old school” partners who built practices on expertise and relationships. Some adapted to the new model and competed on origination. All of us were complicit in the transition to some degree. We sat in partner admission committees and voted for candidates who were strong sellers even when we had reservations about their technical depth. We watched the criteria shift and adjusted our own behaviour to match. We saw it happening and we stayed.

The complicity is worth naming because it explains something important about why the degradation was so thorough. There was no villain. There was no memo from the managing partner announcing that deep expertise no longer counted. (If there had been, at least one of us would have kept a copy. We kept copies of everything.) There was a series of incremental decisions, each individually defensible, that collectively rewired the system. A partner admission committee looks at two candidates: one with deep sector knowledge and a modest pipeline, one with shallow sector knowledge and a strong pipeline. Revenue is down. The practice needs to show growth at the next partner meeting. The second candidate gets the votes. Nobody feels they have made a bad decision; in the moment, it is the obviously correct one. Multiply that choice across every practice, every office, every admission cycle for fifteen years, and you have rebuilt the partnership without ever deciding to rebuild it.

The signal the market is sending back

The people coming through can see what the role has become, and they are responding rationally.

Only 34% of senior consultants now consider making partner a strong motivator. Twenty years ago, that number would have been unintelligible. Partnership was the prize. You oriented a decade of your career around it the way a medieval knight oriented around a quest, except the grail was a capital account and a reserved car park space. Today, a significant majority of the people best positioned to make partner are actively choosing not to pursue it, or are leaving for platforms where the role looks different.

The firms’ own behaviour confirms the diagnosis. UK Big Four partner promotions fell to 179 in 2025, down from a peak of 276 three years earlier. Deloitte UK promoted 60, down from 124 in 2022. KPMG promoted almost nobody to equity partner between 2021 and 2023. Across the four firms in the UK, total equity partner numbers fell for the first time in five years, to approximately 3,050.

Some of this is cyclical. Advisory revenues are down; fewer partners are needed to service the work. But some of it is the firms admitting through their actions what they cannot admit in their recruiting materials: the path to partnership has become narrower, less rewarding, and less connected to the skills that make someone genuinely good at the job.

Look at the compensation data and you can see the perverse outcome. Average partner payouts at KPMG UK hit a record £816,000. Deloitte UK partners exceeded £1 million on average. Profits per partner are rising because the denominator is shrinking: fewer partners, dividing the same (or slightly smaller) pool, produces larger individual numbers. Firms present this as evidence of a thriving partnership, presumably with a straight face. What it actually represents is a smaller, more sales-oriented group capturing a larger share, while the people who would have diversified the partnership’s capabilities a decade ago are building their careers elsewhere.

What this says about everything else

The partner track is a diagnostic, and this is why we gave this piece its subtitle. When you understand what happened to the track, you can see the same mechanism operating across every dimension of these firms.

The leverage model runs on the same logic. Engagement teams used to be staffed with a mix of experienced and junior professionals calibrated to the complexity of the work. Today, the staffing decision is driven by margin targets, and the margin is maximised by putting the cheapest bodies on the engagement and billing at rates that imply seniority the team does not have. The buyers tolerate this because they are buying the brand, not the team. The same shift from substance to brand that changed the partner track also changed the staffing model, because a buyer who selects on brand has no reason to scrutinise whether the analyst on their engagement has six months of experience or six years.

Quality erosion follows the same causal chain, and the data shows it clearly. PCAOB deficiency rates at Big Four firms rose from 12% in 2020 to 26% in 2022-23 before improving to 20% in 2024. Those rates are not produced by incompetent individuals. They are produced by a system that has deprioritised technical excellence relative to commercial performance, from the bottom of the pyramid to the top. When the partners running engagements were selected for their sales ability rather than their technical judgment, the quality of the work those engagements produce deteriorated accordingly.

The talent exodus is the mirror image. When the people who valued depth and expertise and patience look at what the partnership has become, they leave. They go to Alvarez & Marsal, to boutique firms, to industry roles, to anything that does not require them to subordinate every professional instinct to a revenue target. The firms respond with retention bonuses and hardened forfeiture provisions — paying people more to tolerate the same dysfunction rather than fixing the dysfunction. We will examine that exodus in detail in a forthcoming piece.

Every one of these patterns traces back to the same root cause: the market stopped paying for expertise and started paying for brand, and the firms adapted to what the market rewarded. Rational adaptation, corrosive consequences. And the partner track is where the corrosion is most visible, because it is where the firms’ values are most explicitly encoded. When you change what it takes to become a partner, you change what the firm is.

The question we cannot answer

We have described a system that degraded because incentive structures shifted. We have described our own complicity in the shift. What we have not described is a plausible path back, because we are not certain one exists within the current model.

The firms cannot unilaterally decide to weight technical depth over origination in their partner admissions. If one firm did, its revenues would decline relative to competitors in the short term, because the deeply knowledgeable partners would be outpaced by the aggressive sellers at rival firms. The shift happened because it was individually rational for each firm to follow the market, and the same logic prevents any single firm from reversing course.

The change would have to come from the buy side. If corporate buyers started selecting advisory firms on the basis of the individual partner’s expertise again, rather than the firm’s brand, the firms would adapt overnight. They are extraordinarily responsive to buyer behaviour; that is, after all, how they got here. But the trend in procurement is moving in the opposite direction: toward more standardised vendor selection, more panel-based purchasing, more institutional decision-making. Nobody in procurement has ever been fired for ticking the box marked “Big Four.” The conditions that produced the old partner track are unlikely to return.

Which leaves the question we put to you, because we do not have the answer ourselves.

Current or former partners: when you compare the partner cohort today to the one from fifteen years ago, what is the biggest difference? Is it skills, values, depth, or something else entirely? And if you had to design the partner admission process from scratch, knowing what you know now, what would you change?

In August 2019, KPMG’s engagement partner on Lojas Americanas and B2W (then separate listed companies, merged into the present Americanas structure only in 2021) sent both companies a formal internal control letter. It set out deficiencies in their controls over verbas de propaganda cooperada, the cooperative advertising allowances that suppliers pay retailers for in-store promotion. The firm had already raised the point with the finance directors. No formal response came back. So it escalated to a letter, which is what the letter is for.

Six days later, management terminated the engagement. The stated reason was commercial.

Carla Bellangero, the KPMG partner responsible, described that sequence to a Brazilian congressional inquiry on 1 August 2023. By then Americanas had disclosed accounting inconsistencies initially estimated at R$20 billion and entered bankruptcy protection with roughly R$43 billion of debt. Its own independent investigation later concluded that prior management had cumulatively inflated reported results by approximately R$25 billion and understated gross financial debt by approximately R$20.6 billion. Those cooperative advertising allowances were one of the two principal mechanisms.

Almost nothing has been written about this outside Brazil, which is a shame, because the file documents something the profession discusses far less readily than a missed balance. Wirecard, Evergrande and Carillion all cover the ground of a firm failing to find what was there. Americanas covers what happens to a firm that starts asking about the right accounts, and how quickly the architecture built to protect auditor independence can be turned into the exit route.

Two legs, one ledger

The fraud had a debt leg and a margin leg.

The debt leg was risco sacado, the Brazilian term for reverse factoring. A bank pays the retailer’s supplier early. The supplier gets its cash, the bank gets a fee, and the retailer now owes the bank rather than the supplier. Under IFRS, the question at that point is whether the obligation still has the nature and function of a trade payable or has become financing that belongs elsewhere on the balance sheet. Extended payment terms, an explicit interest charge and a bank counterparty all point towards financing. Americanas kept it in accounts payable regardless. The company later said that R$18.4 billion of purchase financing and R$2.2 billion of working-capital financing had been inadequately recorded in supplier accounts. Gross financial debt was understated by approximately R$20.6 billion. Depending on presentation, supplier finance can also make operating cash flow look stronger by moving what is economically a financing flow through working capital.

This is not exotic. Similar opacity around supplier-finance obligations was central to Carillion’s collapse in 2018 and became a factor in subsequent failures including NMC Health and Greensill. It is the reason the SEC has spent years pressing US filers on supply chain finance disclosure.

The margin leg was the cooperative advertising allowances. Suppliers genuinely do pay retailers to promote their products, and the credit does reduce cost of goods sold. Americanas is alleged to have booked allowances it never received. That flatters gross margin directly, and it has the property every accounting fraud eventually discovers: the fiction has to grow each year to cover the prior year’s fiction, because last year’s fake credit is this year’s baseline.

Put the two together and you get a company that appeared to be less indebted, more cash-generative and more profitable than it was, across every one of those dimensions at once, for years.

The rotation sequence

Brazil has had mandatory audit firm rotation since 1999. Five consecutive financial years with the same firm, extended to ten where the company maintains a statutory audit committee, with a minimum three-year gap before the firm can be re-engaged. It is the reform that UK and US commentators have spent two decades pointing at as the answer to auditor capture. Brazil built it. The Supreme Federal Court upheld its constitutionality in 2020.

Here is how it ran at Americanas, according to the report of the independent committee the company itself commissioned:

PwC audited the group from 2011 to 2015. KPMG took over for 2016. In the third quarter of 2019, with KPMG able to continue to at least 2020, the engagement was terminated early. PwC came back.

PwC’s three-year cooling-off period had, by then, just elapsed.

We are not going to claim that anyone sat in a room and designed that calendar. What we will say is that the sequence is one we have seen from the inside, more than once, at more than one firm. An early termination, six days after a control letter, followed by the return of the firm that signed the accounts before the questions started. Nobody has to conspire for this to work. Rotation permanently supplies a legitimate-looking reason to change firms at any moment, and a legitimate-looking pool of firms to change to.

The independent committee added a detail that says more about Brazilian governance than any of the fraud mechanics. It found no record of anyone on the boards or the fiscal councils of Lojas Americanas or B2W questioning the early switch. The board has disputed that account. We would note only that an early auditor change amid unresolved control concerns is exactly the kind of event a board, fiscal council and audit committee are supposed to scrutinise and document, and that under the same CVM rules, having a qualifying statutory audit committee is what buys a company an extra five years with the same firm.

Do not make KPMG the hero

The tidy version of this story is that one firm asked the right question and got fired for it. The record contradicts it in two places, and we would rather deal with the version the record supports.

KPMG audited the group for three full years and part of a fourth. It signed unmodified opinions throughout. It did not find the fraud. Bellangero told the inquiry that the firm found no evidence of fraud at either Lojas Americanas or B2W across 2016 to 2019, and that it identified control deficiencies, which is a materially different statement.

There is worse. In June 2023, the chief executive who had taken over in February (Leonardo Coelho Pereira, brought in after Sérgio Rial’s nine-day tenure) gave the congressional inquiry internal correspondence which, he said, showed the wording of the internal controls letter being changed after requests from the retailer. The change reclassified the cooperative advertising item from recommendations requiring management’s attention to other recommendations, with the effect that it no longer counted as a significant deficiency. Americanas confirmed that reclassification in its own statement. KPMG and PwC both told the inquiry the communications had been taken out of context, and no finding has been made against either firm.

Keep both facts in view at once. The audit firm raised the correct accounts. The classification of what it raised moved in the client’s direction. Then the firm sent the escalation letter anyway, and was gone within a week.

Neither the heroic reading nor the corrupt one survives that combination. What is left is an engagement team applying professional judgement under commercial gravity (which is the condition every audit partner reading this has worked in for an entire career). The pressure is always in the same direction. Nobody instructs you. You find that the version of the finding which keeps the relationship intact is also, on reflection, the version you can defend to the file reviewer, and the two facts arrive in that order more often than any of us would put in writing.

What the incoming firm is required to do

Under the profession’s own ethics framework, an incoming auditor has to approach the outgoing one before accepting the appointment and ask whether there is anything it should know.

Americanas has stated that the 2019 substitution was made with KPMG’s own consent. We have no reason to doubt that. We would only observe what that consent is worth as a control.

Any partner who has taken on a listed engagement knows the general shape of the handover conversation. In our experience it is short, professionally courteous, and documented in a file note. It tends to happen after the proposal has been won and the fee agreed, at a point where the incoming partner has already told their firm’s leadership what they have landed. Retrospectively unwinding a Novo Mercado-listed engagement backed by the shareholder group behind AB InBev, Burger King and Kraft Heinz, on the strength of a predecessor’s control letter about advertising rebates, is not a decision the acceptance process is designed to produce. The public record does not disclose what PwC asked KPMG, what KPMG communicated, or whether the August 2019 control letters were shared.

PwC then signed unmodified opinions on the 2019, 2020 and 2021 financial statements. None referred to the reverse factoring exposure. When PwC’s audit leader appeared before the inquiry alongside Bellangero, his position was that the scheme was sophisticated and built specifically to defeat detection.

That defence deserves to be taken seriously rather than waved away. In January 2026 the CVM opened additional investigations involving banks, their administrators and other intermediaries connected with the reverse factoring operations, including the transparency of those arrangements towards the auditors. Those inquiries may clarify what information the banks generated and shared. If the funding banks were party to a structure that concealed the true nature of the obligations from the auditors, that would constitute genuine mitigation.

It is not complete mitigation, for one reason. The reverse factoring exposure was not hidden from the auditors as a category. KPMG had reported to the company on risco sacado in 2019. The question was raised, inside the file, by name, and then the engagement changed hands.

The standard-setters responded with disclosure

The most quoted line in the online write-ups of this case is that no accounting standard requires you to break reverse factoring out of trade payables. That was true when the fraud was running. It is no longer quite true, and the reason it changed says as much about the system as the fraud itself.

The IASB’s IFRIC had discussed the classification of reverse factoring under existing standards in December 2020, and a formal project was already underway before Americanas collapsed. The FASB issued ASU 2022-04 in September 2022, effective for financial years beginning after 15 December 2022. The IASB issued its Supplier Finance Arrangements amendments to IAS 7 and IFRS 7 on 25 May 2023, effective from January 2024.

Put the dates next to the collapses. Carillion failed in January 2018. NMC Health in 2020. Greensill in March 2021. Americanas in January 2023. The IASB work predated the last of those, but it took the accumulation of all of them before the amendments were issued.

Both the FASB and the IASB were explicit that the new requirements do not change recognition, measurement or presentation. They added targeted supplier-finance disclosures but did not impose a universal balance-sheet classification rule. Considerable judgement therefore remains over whether a particular arrangement belongs within trade payables, other financial liabilities or a separate line item. That is a weakness worth debating. It is not, of course, permission to conceal financing or falsify the underlying entries, which is what Americanas is alleged to have done. But the judgement-based framework that created the space for the misclassification of R$20.6 billion of financial debt is still the framework.

We have watched standard-setters respond to failure this way for thirty years and we have advised clients on how to satisfy the resulting requirements, so this is complicity talking rather than commentary. A disclosure remedy is what a system produces when it wants to be seen responding without disturbing anybody’s balance sheet.

Three years, two tracks

The Brazilian criminal system has moved with real speed. Former chief executive Miguel Gutierrez was arrested in Madrid in June 2024. Federal prosecutors filed criminal charges against thirteen former executives in March 2025. On 25 June 2026 the Federal Police launched the second phase of Operation Disclosure and executed nine search warrants across Rio de Janeiro and São Paulo. A federal court authorised precautionary asset freezes up to a ceiling of R$54 billion, roughly $10.4 billion. Brazilian media identified among the targets reference shareholder Carlos Alberto Sicupira and Paulo Lemann, a former board member and son of Jorge Paulo Lemann, as well as executives at private banks, on the theory that the lenders understood the structure and continued to fund it. None of those named has been charged.

The auditor track has moved differently.

Brazil’s securities regulator, the CVM, opened its administrative proceeding against KPMG and Bellangero on 26 November 2024. The proceeding covers the 2017 and 2018 financial years. Citation was served in February 2025. Three and a half years after the disclosure, the case has not been heard, and the charging document is not public. The parallel inquiry into PwC’s conduct across 2019, 2020 and 2021 was consolidated into another proceeding that remains under review by the regulator’s accounting and audit superintendency. KPMG and its former engagement partner face a pending administrative accusation. PwC’s conduct remains under regulatory review. No final sanction has been issued against either firm.

Some of this is capacity rather than deference. Valor Econômico reported in late 2025 that the CVM had been operating with two of its five board seats vacant, with judgments declining and a backlog approaching 860 administrative cases.

What the firms argued in court

The clearest statement of the profession’s position came not from a firm’s press office but from its lawyers.

A minority shareholder sued PwC and KPMG in Rio de Janeiro for negligence, claiming the collapse in the share price was a direct consequence of the auditors’ failure. Both firms argued that the shareholder had no standing to sue them at all. Any breach of duty by the auditor, on their case, damaged the company; shareholders were harmed only reflexively, through the company, and could not bring a claim in their own name.

On 4 February 2025, the 14th Chamber of Private Law of the Rio de Janeiro state court rejected that argument unanimously and reinstated the claim.

The defence was not a throwaway. It was correct as a general proposition of Brazilian company law and advanced by serious counsel. It is also the position the profession takes everywhere when it is sued by investors, which is the part that matters for this discussion.

It is also, stated plainly, the argument that the audit does not run for the benefit of the people who rely on it. An auditor can in principle accept that investors are intended users while arguing that a specific claimed loss is legally derivative. But the practical effect is identical: every prospectus, every investor presentation, every defence of the statutory audit’s existence rests on the proposition that the opinion exists so that outside investors can trust the numbers. When the numbers turn out to be wrong and the investors turn up in court, the first line of defence is that their loss belongs to the company, not to them. The profession has arranged its affairs so that it never has to reconcile those two propositions in a forum where the answer costs money.

The pattern, again

Strip out the Portuguese and the mechanics are generic.

An engagement team asks about a specific account. The classification of the finding softens, allegedly under client pressure. A formal letter goes out anyway. The client exercises an entirely lawful right to change firms. The client invokes a rotation rule that exists to prevent capture, and it provides the mechanism and the cover. The incoming firm performs a handover enquiry it has every commercial reason to keep brief. Three unmodified opinions follow. The collapse comes four years later, the criminal process moves quickly against individuals, the administrative process against the firms begins but has not been heard, and the standard-setter responds with a disclosure requirement.

Nobody in that chain has to be corrupt for the outcome to arrive, which is why the outcome keeps arriving, and why every reform aimed at the corruption case leaves the rest of the chain intact. Rotation was the reform. Here it supplied the exit.

Americanas filed to exit bankruptcy protection in March 2026. Its stores are open. Of the two firms that audited it through the fraud, KPMG faces a pending CVM accusation, PwC remains under regulatory review, and no final sanction has been issued against either.

A reader in São Paulo raised this case with us and walked through the mechanics in detail. The analysis above is ours; the prompt was his.

One question for the comments. KPMG raised the right accounts, the classification was allegedly softened at the client’s request, and the letter went out anyway, six days before the termination. If you had been the engagement partner in August 2019, what would you have done differently that would not also have ended the engagement? We are interested in the specific alternative, not the principle.

Every RFP we advise on now asks the same question in some form: “Which members of your team have actually done this work in an operating environment?”

Not modelled it. Not benchmarked it. Done it. Managed a P&L through a downturn. Run a post-merger integration where the acquiring company’s ERP system didn’t talk to the target’s. Restructured a supply chain when the second-largest supplier went bankrupt during the engagement. Buyers want consultants who have made consequential decisions under genuine uncertainty, not people whose experience consists of recommending that other people make those decisions.

The firms know this. They have known it for a decade. They recruit senior industry leaders with budget, fanfare, and a partner title. Those leaders arrive in January with a client list and two decades of credibility. By March of their second year, they are being asked why their pipeline is thin. By their third year, most of them are gone.

One of our contributors watched this cycle repeat four times in six years at the same firm, same practice group. Four industry hires. Each one brought a genuine skill the practice lacked. Each one won pitches the practice would have lost without them. Each one left. The firm classified every departure as “cultural fit.”

The firm’s own selection mechanism pushed them out.

The origination filter

The partnership model at most major consulting firms selects for one capability above all others: the ability to sell work. Origination is the gateway to partnership, the determinant of compensation, the basis for internal status, and the metric that decides who survives the next restructuring. A partner who originates $8 million in annual revenue and delivers mediocre work will outlast a partner who originates $2 million and delivers exceptional work. This is not a secret. Every senior manager considering the partnership track knows it. The economics are explicit.

Origination is a specific skill. It requires the ability to identify a client’s latent anxiety before the client has fully articulated it, to frame that anxiety as a problem the firm can solve, and to do this repeatedly across a portfolio of accounts. It requires comfort with the mechanics of consulting sales: constructing proposals that promise enough to win the work without promising so much that the firm bears risk. And it requires internal navigation: securing the right team for the pitch, negotiating with other partners for access to their accounts.

None of this has anything to do with running a manufacturing plant.

The operator who spent twenty years turning around distressed industrial businesses brings something the firm does not have: judgment born from making decisions where the consequences were real and the data was incomplete. That judgment is exactly what clients are asking for when they write “operational experience required” into their selection criteria. The partner whose experience is a decade of selling and overseeing consulting engagements has a different kind of judgment, useful in different circumstances, but it is not what the buyer specified.

Everyone involved knows the operator is more valuable to the client. The firm recruits them anyway, and then its own selection mechanism destroys the hire.

The eighteen-month clock

What happens next is consistent enough to describe with precision. It operates on a cycle between eighteen months and three years, and the stages are predictable.

Months one through six: the operator is deployed as a credential in pitches. “Our team includes a former COO of [Fortune 500 company]” goes into the proposal. The operator sits in the pitch meeting, answers the client’s operational questions with the specificity that only firsthand experience produces, and the firm wins the engagement. The partner who brought the operator to the pitch takes origination credit.

Months six through twelve: the operator delivers the engagement. The client is satisfied because they are getting what they asked for. The operator’s utilization is high. Internal perception is positive. Nobody asks hard questions about their pipeline.

Months twelve through eighteen: the internal review process begins. The operator has delivered well but has not generated new revenue independently. They have no pipeline of their own because they have been delivering work, and also because they do not know how to sell consulting. They know how to solve operational problems. Selling consulting is a different activity, involving a different vocabulary, a different set of relationships, and a different rhythm of engagement. Their annual review is polite but pointed: “We need to see more origination.”

Months eighteen through thirty: the operator tries to originate. They call former colleagues. Some calls produce meetings. Fewer produce proposals. Fewer still produce signed engagements, because the operator’s network consists of other operators, not consulting buyers with procurement authority and a budget for a $3 million engagement. Meanwhile, utilization drops because they are spending time on business development that is not converting. The economics now show a partner with declining utilization and minimal origination. A net cost.

Months thirty through thirty-six: the operator leaves. Sometimes voluntarily. Sometimes through the quiet mechanisms firms use: reduced compensation, reassignment to less desirable accounts, exclusion from the partner meetings where strategy is discussed. The firm records the departure, updates the alumni database, and begins recruiting the next industry leader.

We are not speculating. We watched this happen. In several cases, we participated in the review committees that produced exactly the outcomes we are describing. The committee would acknowledge, in the same meeting, that the operator had won pitches the firm would otherwise have lost, that the client feedback was exceptional, and that the origination numbers were below threshold. The first two observations produced sympathetic nods. The third produced action.

Why A&M and AlixPartners cracked this (mostly)

A reader named Vince, responding to our post on the three-identical-pitches problem, pointed out that turnaround firms like Alvarez & Marsal and AlixPartners seem to retain operators successfully where the Big 4 cannot. He is largely right, and the reason is worth examining because it clarifies exactly which part of the Big 4 model is incompatible with operational talent. (A caveat before we proceed: A&M is no soft landing. Recent reporting describes an intensely commercial culture with its own disputes over mandate ownership and senior turnover. The difference is not that A&M eliminated the origination pressure. It is that A&M made operational execution a credible path to generating revenue, rather than treating it as a support function for someone else’s sales.)

A&M and AlixPartners differ from the Big 4 on three dimensions that matter for this question.

First, compensation is tied more directly to origination-linked performance and, in many cases, includes real equity or equity-like instruments. An operator who brings relationships and credibility, even if those relationships produce revenue through referrals and introductions rather than through the traditional BD pipeline, is compensated for that contribution. At many Big 4 firms, the compensation model allocates partner profit shares through committee-driven processes that heavily weight origination volume, and cannot easily make this distinction. Revenue that arrives because a client hired the firm on the strength of the operator’s reputation is often credited to whoever signed the engagement letter, which is usually not the operator.

Second, the partner groups are smaller. A&M has approximately 800 managing directors globally. AlixPartners has fewer than 500 managing directors. Deloitte has more than 6,000 partners, principals, and managing directors in the US alone. In a smaller partnership, an operator with a strong industry network can generate revenue through relationships and reputation in ways that register as contribution. In a partnership of thousands, the same contribution is invisible in the aggregate numbers. The operator’s value is legible at one scale and illegible at another.

Third, these firms were built around operators. A&M’s founding model placed experienced operators inside distressed companies to run them, not to advise them. The firm’s culture and its definition of what constitutes good work derive from operational execution rather than from advisory recommendation. An operator joining A&M is joining a firm that already values what they do. An operator joining a Big 4 firm is joining a firm that values what they do when it appears in a pitch deck, and then asks them to become someone else.

Follow the talent and you see the verdict. A&M has become one of the primary destinations for departing Big 4 senior talent. European recruiter data from 2026 tracked roughly eight times as many directors and partners moving from Big 4 firms to A&M as in the reverse direction, and the pattern holds across markets. Bryan Marsal described his firm as “a hungry dog outside a butcher’s shop window,” and the Big 4 have been obliging enough to keep pushing talent through the door.

The irreconcilable tension

Vince raised a further point that deserves direct engagement because it contains a genuine difficulty: a firm needs both originators and delivery specialists to succeed. You need people who can sell the work and people who can do the work. The problem is that having both types at the partner level means splitting the profit pool more ways.

The Big 4 model has been moving in the opposite direction for twenty years. The trend is toward a smaller, more origination-focused partnership. Partners who deliver but do not sell are being converted from equity to salaried positions, or eased out entirely. At KPMG UK, equity partners who hold significant profit-share units but bring in little client revenue have acquired an informal label: Huncs. High Units, No Clients. The label says everything about how the model values delivery.

If you wanted to retain operators, you would need to expand the partnership to include people whose primary contribution is delivery quality and client impact rather than revenue generation. That means more partners splitting the same revenue. Every current equity partner would take a smaller share. No partnership governance system we have ever seen would vote for this voluntarily. The partners who control the vote are the ones whose share would shrink.

The obvious counterargument: firms can create alternative tracks. Salaried partner roles, “expert partner” designations, senior advisor titles with different scorecards. Several Big 4 firms have tried exactly this. The problem is that these tracks carry less authority, lower compensation, and visibly second-class status within the partnership. Clients notice when the operator in the room cannot sign the engagement letter, cannot set the fee, and cannot commit the firm’s resources without checking with a “real” partner. The operator notices too. Creating a formal tier for people who deliver but do not sell is an admission of the hierarchy, not a resolution of it.

So the tension is genuine, and within the current governance structure, it has proved functionally irreconcilable. The firm cannot simultaneously optimise for origination (which keeps the per-partner economics high) and for operational quality (which requires more partners, or at least more partner-level compensation for non-originating contributors). Every firm that has tried has ended up defaulting back to origination as the binding constraint, because origination is what the compensation committee measures, and the compensation committee is composed of originators.

The client is noticing

The consequence is visible in the market. A&M has expanded predominantly through direct hiring of exactly the operators the Big 4 could not keep. AlixPartners has grown from roughly 2,900 employees in 2023 to over 3,500 today. The boutique and specialist advisory market is filling with former Big 4 partners who left precisely because the model would not accommodate what they were good at.

And yet the Big 4 response has been to recruit more operators, faster, without changing the model in any way that matters. The recruiting budget goes up. The attrition rate stays the same. The exit interviews say the same things they have said for a decade. And the firms continue to classify the departures as cultural fit, because admitting the alternative would require acknowledging that the partnership model itself is the problem.

We are taking a position here, and we want to be explicit about it: the origination-first partnership model produces a firm that is optimised for selling and constitutionally incapable of retaining the people who would improve what it sells. The model worked when clients were buying the brand and the partner’s rolodex. It is failing now because clients are buying expertise, and expertise means operational experience, and operational experience is the one thing the model cannot accommodate.

A&M and AlixPartners did not discover a clever HR strategy. They built firms where operators are the product, not the accessory. The Big 4 are trying to bolt operators onto a machine designed to reject them, and then expressing surprise when the bolts do not hold.

We want to hear from readers on this. Specifically:

Has your firm successfully retained an operator hire for more than three years at the partnership level? What made the difference: compensation structure, cultural accommodation, or something else? And for the operators who left: was there a specific moment when you realised the model would never make room for what you do, or did the firm find a way to push you out without ever saying so directly?

In late October 2025, McKinsey gathered thousands of its people in Chicago to kick off the firm’s centennial celebrations. Bob Sternfels, the global managing partner, delivered a rallying speech about the second century. There were, one imagines, balloon arches. Possibly a cake shaped like a 2×2 matrix. Away from the festivities, managers in some non-client-facing functions were being told to prepare for a leaner organisation.

The firm that taught corporate America how to restructure, re-engineer, and rightsize entered its centennial year applying the playbook to itself. The workforce had already fallen from more than 45,000 at the end of 2023 to about 40,000. McKinsey named 224 new partners in November 2025, roughly 44% fewer than the class of 2022. Bloomberg reported discussions about reducing headcount in non-client-facing departments by a further 10% over the following 18 to 24 months.

McKinsey attributed much of the reduction to attrition and tougher performance reviews. I’ve sat through enough partner meetings to recognise the language. When a client told us their 10% headcount reduction was “natural attrition,” we’d nod politely and then put a slide in the deck about organisational honesty. (The slide was titled “The Attrition Myth: A Framework for Candid Workforce Planning.” It had a little iceberg graphic. The client loved it. They then fired 800 people and called it a “talent refresh.”)

The numbers McKinsey doesn’t volunteer

The Economist recently asked whether McKinsey has “fallen below fighting weight.” Their business editor, Tom Lee Devlin, laid out the basics: revenue growth of about 2% in 2024, BCG growing at roughly 10%, the gap closing fast. He was being diplomatic.

In 2012, McKinsey was roughly twice BCG’s size. McKinsey reported $16 billion in revenue for 2023; more recent reporting has placed its annual revenue at roughly $15 billion to $16 billion. BCG, by contrast, reported $13.5 billion in 2024 and $14.4 billion in 2025. Whatever the exact private-company figure, the gap has narrowed sharply.

For most companies, being the largest in a market by a shrinking margin is manageable. For McKinsey, it is an identity problem. The firm’s value proposition to clients, to recruits, and to the partnership itself rests on being the default choice when the stakes are highest. “We hired McKinsey” carries weight in a boardroom precisely because McKinsey is understood to be number one. That sentence loses force when BCG is growing three to five times faster and the crossover is a matter of arithmetic. (A simple extrapolation of the 2024 growth rates implied BCG could overtake McKinsey around 2027, though BCG’s growth slowed to 7% in 2025. The gap is still closing; the timeline is less certain.)

The revenue stagnation follows a decade of expansion that has now clearly overshot. McKinsey went from 17,000 employees in 2012 to 45,000 by late 2023, acquiring a succession of analytics, design, cloud, and digital-product businesses and pushing hard into implementation work. The firm was trying to compete with Accenture and the Big Four on delivery, which meant hiring thousands of specialists in software engineering and cloud implementation where McKinsey had no traditional strength.

When the growth slowed, those specialists went first. In 2023, the firm cut roughly 1,400 back-office roles. In 2024, McKinsey cut approximately 400 more in areas such as data engineering and software.

The firm cut its data engineers and software developers while intensifying its pitch as an AI-first firm.

BCG did what McKinsey tells its clients to do

What makes the comparison painful is that BCG succeeded by following the advice McKinsey sells: invest in capability before you need it, retain your specialists, commit to the transformation rather than hedge. BCG essentially walked into McKinsey’s own library, checked out The McKinsey Way, and actually did the homework. McKinsey wrote the book, framed the cover, and then fired the people who could implement chapter four.

BCG played a similar expansion game, building out its digital and technology offerings and standing up BCG X with more than 3,000 technologists. The difference, according to the Economist and multiple other reports, is that BCG did a better job retaining the specialists who make technical advisory work credible. BCG says AI- and technology-focused services represented more than 40% of total revenue in 2025, with AI services growing 25% year over year.

McKinsey’s equivalent is QuantumBlack, with approximately 1,700 people. McKinsey describes a broader technology community of roughly 6,000, plus another 4,000 to 5,000 consultants leading technical work from the business side. Those numbers matter, because McKinsey’s own AI leadership has said that the 4,000-to-5,000 group does not necessarily possess deep technical expertise. That admission is more revealing than any outside criticism could be. It’s like a restaurant announcing, “We have 12 chefs, plus 40 people who have watched a lot of cooking shows and can describe what a béarnaise sauce should taste like.”

QuantumBlack is not a vanity project. Its engineers build data pipelines, deploy models into production, and integrate systems into client operations. The firm has published credible deployed work in pharma, telecommunications, and manufacturing. But the gap between QuantumBlack’s genuine capability and how the broader partnership sells “AI transformation” to clients is wide enough to notice from the outside. From the inside, it is impossible to miss. You can smell it. It smells like a partner who learned the word “vector database” on the flight to the client meeting.

The talent problem nobody in the partnership will say out loud

The Economist covered the financials well. What it cannot cover, because its reporters don’t sit in the partner meetings, is the talent problem underneath them.

The best AI talent in the world is not going to consulting firms. The competition for people who can build, deploy, and scale AI systems is the most intense talent market in a generation. ManpowerGroup’s 2026 survey of 39,000 employers across 41 countries found that AI skills are now the single hardest capability for employers to find, topping every other skill category for the first time. The people who have those skills can work at OpenAI, Anthropic, Google DeepMind, Palantir, or any of hundreds of well-funded startups. They can start their own companies. They can name their compensation.

Why would they go to McKinsey?

The consulting model pays well at the junior level but is not competitive with what top AI engineers earn at technology companies. At the top end of the US AI market, senior individual contributors receive total compensation above $500,000 without entering a management track. More importantly, the work is different. A Palantir forward-deployed engineer builds production systems embedded in client operations. A McKinsey engagement, even one with QuantumBlack involvement, typically produces recommendations, architectures, and pilot implementations that the client’s own teams (or a technology vendor) must then scale. The career paths diverge further up: in tech, a senior ML engineer can stay technical indefinitely. In consulting, the path to partner-level compensation requires becoming a salesperson, managing client relationships, and surviving partnership politics. Your reward for being the best engineer on the team is that you never get to engineer anything again. Your new job is to play golf with a man named Greg and pretend to care about his boat.

The result is a firm that has genuine AI capability in QuantumBlack and the broader tech community, but whose AI-labelled work across the wider partnership relies heavily on generalist consultants whose technical depth does not match how the work is sold. I have seen this configuration firsthand. Partners who were selling “digital transformation” in 2019, “ESG transformation” in 2021, and “AI transformation” in 2024, with the same skill set and a new cover page on the same deck.

I want to be specific about this because it goes to the core of the AI credibility problem. The consulting partnership model produces people who are exceptionally good at learning a domain’s vocabulary in two weeks and presenting confidently on it in three. That skill is valuable for many types of advisory work. It is inadequate for AI, because AI clients are increasingly technical buyers who can tell within fifteen minutes whether the person across the table has ever trained a model, debugged a data pipeline, or shipped a feature. The partner who pivoted from ESG to AI cannot pass that test. They know it. The engagement manager staffing the team knows it. The only person who might not know it is the CEO who approved the statement of work, and even that gap is closing as in-house technical leadership gets involved in vendor selection. (The CEO will find out eventually. Usually at 11 p.m. on a Sunday, when the CTO sends a Slack message that begins, “So I just got off the call with McKinsey, and I have questions.”)

McKinsey’s own AI leadership has said that QuantumBlack “drives” all the firm’s AI initiatives and that AI-related work accounts for roughly 40% of what McKinsey does. If that is true, the question becomes whether QuantumBlack’s specialists are consistently attached to the engagements sold under the AI banner, or whether the firm’s commercial machine is selling AI work faster than the specialist bench can staff it. The public headcount arithmetic suggests the latter. This is a restaurant selling 400 steaks a night with a kitchen staffed for 80. At some point, someone is getting a very creative interpretation of “medium rare.”

Sternfels gave the game away in early 2026 when he described McKinsey’s workforce as “roughly 60,000,” including about 25,000 AI agents alongside roughly 40,000 humans. The headline agent is Lilli, an internal generative chatbot that lets consultants search the firm’s knowledge base and generate slides from prompts. McKinsey says about 72% of employees use Lilli and that it saves up to 30% of time spent searching and synthesising knowledge. That may well be true. But an internal chatbot that helps your own people produce slides faster is a productivity tool, not evidence that your workforce includes 25,000 additional “employees.” Every major consulting firm has deployed similar tools. Every client has access to the same foundation models. Describing your agent count as headcount is the kind of framing McKinsey would tell a client never to use publicly.

The competition McKinsey can’t acquire its way out of

The competitive picture is more complicated than a simple “builders versus advisors” split, but the direction of travel is clear.

Palantir grew its revenue by 48% year over year in Q2 2025, and by 85% in Q1 2026. Its forward-deployed engineers embed with clients, build production systems, and iterate on working software. A CFO who has paid McKinsey $3 million for an AI strategy that recommends hiring specialists to do the actual implementation has paid for a very expensive shopping list. A beautiful shopping list, with a proprietary framework and a memorable acronym. But a shopping list.

OpenAI launched its Deployment Company in 2026, staffing it with forward-deployed engineers to embed in enterprises. Interestingly, OpenAI also named McKinsey and BCG as Frontier Alliance partners for strategy, operating-model design, and change management. That alliance complicates the binary: McKinsey is not simply being displaced by technology companies; it is being repositioned as one layer in a stack where the technology partner owns the engineering. Whether that layer remains premium-priced is the central question. It is a little like being told you are still the most important person at the restaurant, but your new title is “the man who suggests which wine goes with the meal someone else cooked.”

Accenture announced its acquisition of Faculty, a British AI company, in January 2026, adding more than 400 AI professionals. Bain expanded its Palantir partnership in March 2026 rather than try to build AI delivery capability from scratch. (Bain had first partnered with Palantir in May 2025; the expansion itself tells you something about how the build option looked after six months of examining it.)

Enterprise AI delivery is becoming an alliance market. Frontier-model companies, platform vendors, engineering specialists, systems integrators, and management consultancies are combining capabilities. McKinsey’s challenge is to prove that its part of the stack produces enough measurable value to remain premium-priced, and that its genuine technical capability reaches the engagements sold under the AI banner.

The Economist’s Devlin made the useful observation that strategy projects bundle together two things: deep cogitation and a huge amount of grunt work (crunching numbers, preparing slides) that is usually done by the most junior people. AI compresses the grunt work, which is where the junior leverage margin lives. McKinsey is reportedly moving about a quarter of its engagements to outcome-based pricing, which suggests some recognition that the hours-for-dollars model is under pressure. Whether that transition is fast enough, or genuine enough, is the kind of question McKinsey would ask its clients. It has been less willing to answer it about itself.

The recommendation McKinsey would give itself

If McKinsey were its own client, the engagement would write itself. Any competent strategy team (including McKinsey’s own) would identify the same issues within the first two weeks of diagnostic interviews.

The firm expanded aggressively into implementation and delivery work, then cut the technical specialists it had hired for that expansion while intensifying its AI positioning. Its primary competitor invested more effectively in the same capabilities and has been growing three to five times faster. Technology companies are entering its market with delivery capabilities it cannot match independently, and McKinsey is being absorbed into alliance models where it occupies the advisory layer rather than the engineering layer.

The reputational bill is substantial. McKinsey had agreed to $1.6 billion in opioid-related settlements and resolutions before an additional $125 million Purdue agreement was approved in April 2026. McKinsey Africa separately agreed to a $122.85 million criminal penalty over bribery in South Africa. The opioid fallout damaged McKinsey’s federal position, freezing it out of FDA work and contributing to a sharp decline in prime-contract revenue. Saudi Arabia, which Bloomberg-derived reporting described as paying McKinsey at least $500 million annually in the decade leading up to 2024, has reportedly pared back consulting payments. Republican lawmakers have alleged that McKinsey failed to disclose work for Chinese government and state-controlled entities while holding more than $480 million in Defence Department contracts; McKinsey has said it does not work for the Chinese Communist Party or China’s central government. Each of these carries cost, reputational drag, or both, and they accumulated during the same period that commercial growth stalled.

And the internal incentive structure rewards partners for selling rather than building, which prevents the firm from making the capability investments its own analysis would recommend.

The recommendation would be obvious. Stop rebranding existing partners as AI experts and invest in hiring and retaining people with genuine technical depth, even if that means a different compensation structure. Accept that the leverage model (billing clients for junior hours while the partner shows up for the kickoff and the final presentation) needs rethinking when AI can do what the juniors used to do. Engage with the reputational drag directly rather than waiting for it to fade, because it is not fading. And restructure the partnership incentives so that building long-term capability is rewarded as highly as short-term revenue origination.

McKinsey’s partners would nod along to every one of those recommendations. They would commission a beautiful deck summarising them. They would agree in principle and then table the implementation discussion until the next partner meeting, where it would be tabled again, because every one of those recommendations threatens someone’s current compensation or status. The politics that block change at McKinsey’s clients are the same politics that block change at McKinsey. The deck would be gorgeous, though. The font would be impeccable. The recommendations would be numbered, prioritised, and colour-coded by urgency. And then the deck would go into a folder called “Final_v7_ACTUAL_FINAL(2).pptx” and never be opened again.

I spent nearly two decades in large consulting firms telling clients they needed to transform or die. The irony of watching these institutions resist the same advice has a flavour you don’t forget. It tastes like a $400-per-night hotel bar, a cold club sandwich, and the slow dawning realisation that you are billing someone for the privilege of watching them not listen to you.

Happy birthday. Maybe call a consultant.

McKinsey entering its second century is a firm that can still produce a brilliant report on why every other company needs to change. The partners are, individually, among the sharpest people in professional services. The institution’s problem is the one McKinsey diagnoses in other institutions every week: the incentive structure prevents the organisation from acting on what it knows.

McKinsey’s problem is not that it failed to acquire AI capability. QuantumBlack has real builders doing real work. The problem is that the institution may be unable to let that capability remake the partnership at the speed its own advice demands. The firm kicked off its centennial celebrations while quietly planning the largest headcount reduction in its history, counted its AI agents as employees, cut its data engineers while repositioning as AI-first, and watched its closest competitor close the revenue gap using the strategy McKinsey would have recommended to a paying client.

Somewhere, in a McKinsey office, there is a partner working on a case study about a century-old institution that failed to adapt to technological disruption despite having all the information it needed. I hope someone tells them to check the mirror.

Current or former McKinsey people: when did you first notice the gap between how the firm positions its AI capability and how that capability is staffed on actual engagements? And for clients who’ve hired McKinsey for AI work: was QuantumBlack on the team, or was it generalist consultants with a new vocabulary and a very confident handshake?

It took five days.

We published our first post on a Sunday. By Friday morning, someone had filed the charge sheet: retired partners criticising the industry while still collecting retirement from the firms they left. Still holding equity. Still using the old titles to win advisory work. Still dining out on the brand they claim is broken.

The accusation stung. We gathered on a conference call and sat in uncomfortable silence, confronting the contradictions of our position. Then someone said “maybe we should shut it down,” and the rest of us nodded slowly, chastened by the weight of our own complicity. Mike started sobbing uncontrollably. Bob put on his sunglasses, inside, like a man in a witness-protection programme who has already been identified. Anna started mumbling something about having to change her will and leaving everything to a charity. Gerald, who had not spoken in eleven minutes, whispered “I’ll tell my wife,” and no one asked what he meant.

We got caught, and the burden was just too much.

Nah. Just pulling your leg.

Mike forwarded the email to the group chat and someone replied , “Only five days? I had three in the pool.” Bob then sent a photograph of his quarterly retirement statement with a red circle around the direct-deposit line and the caption “blood money.” Anna, a former senior partner at a firm we have already been rude about, responded with: “lol.” Gerald sent a thumbs-up emoji, which, from a man who once billed a client for “stakeholder alignment,” felt about right.

We owe the accusation a serious answer, and we intend to give one. But first, in the spirit of the transparency we keep demanding from the firms, a full and voluntary disclosure of the conflicts of interest. Think of it as a prospectus, except the risk factors are all true and none of them are in eight-point font.

The inventory

Between us, the contributors to this publication currently receive retirement payments, deferred compensation, or both from our former firms. The quarterly statements arrive on schedule. We have not, to date, returned any of them. We have not considered returning any of them. John reports that he considered donating his to charity for about forty-five seconds before remembering he still has a mortgage on his fourth chalet. He wishes us to note that it is a “modest” chalet. It has only one wine cellar.

Several of us still hold equity or equity-equivalent interests in the firms we left. The value of that equity is directly tied to the profitability of the firms we are now publicly arguing have a broken business model. If we are right, we are making ourselves poorer. We have decided to accept that trade-off with a courage born of the knowledge that our firms would first financially hurt all the worker bees, then the junior partners, before they would consider impacting the compensation of senior partners still influential in their respective industries. In other words, we are brave in the specific and limited sense that the bill will land on someone else’s desk before it reaches ours.

Our LinkedIn profiles still say “Former Partner at [Firm].” We have left them that way because removing the line would cut inbound advisory inquiries roughly in half. When a prospective client calls and says “I see you were a partner at [Firm],” we do not reply “yes, but I now believe the firm’s operating model is structurally flawed and its quality has been declining for a decade.” We say “yes,” pause for the appropriate beat of gravitas, and quote our day rate. We have, on occasion, quoted it in the same breath. Efficiency is a value we still respect.

Bob advised a client last month on whether to hire the firm in question for a transformation engagement. He reviewed the proposal, identified the usual problems (too many juniors, vague milestones, a “proprietary framework” that was a 2×2 matrix with new labels and a gradient fill), and recommended the client negotiate a 30% fee reduction and demand named senior resources. The client followed the advice. Bob billed twelve hours for it. The invoice has been paid. Bob does not feel conflicted about this, though he concedes he probably should. He has conceded this every quarter for six years. The concession has not yet affected his billing rate.

Several of us sit on advisory boards or hold board seats that we obtained partly because of where we used to work. Jack is on the advisory board of a company that is currently a significant client of a firm we published a case study about last week. That case study was not complimentary. The next advisory board meeting is in September. We anticipate polite but pointed questions over the pastries. Jack has already rehearsed an answer. It involves the phrase “I’m sure it’s more nuanced than that” and a strategic reach for a croissant.

We attend industry conferences where our name badges still carry the old firm logos, because the conference organisers know which logos sell tickets. We give guest lectures at business schools where we open with “when I was at [Firm]” because that is the sentence that makes the room pay attention. We attend dinners where half the table currently works at firms we spent Tuesday criticising, and we enjoy the wine and the conversation, and nobody mentions the Substack, and we do not bring it up either. The wine, for the record, is always very good. This is, we suspect, not a coincidence. The firms have always understood that a well-chosen Barolo is cheaper than a well-chosen argument.

The branded merchandise deserves its own paragraph, though not a long one. Several of us still own firm-logo golf shirts. John uses an EY coffee mug every morning. It has a chip on the rim. He has had it for fourteen years. When asked why he does not replace it, he said the firm gave him the mug in lieu of a meaningful retention conversation in 2012, and he intends to get his money’s worth. He was then asked whether drinking from a branded mug while writing a publication critical of the brand constituted some form of conflict of interest. He said the coffee tastes the same regardless of what is printed on the outside, and that this was, if anything, a metaphor for the firm’s consulting output. He then drank from it, at length, while maintaining eye contact with the person who had asked the question. He is, even in retirement, very good at billable silence.

For the record, we tried to put a picture of the mug at the top of this post. We ran it through an AI image tool to beautify it and de-identify the chip and the location. We suppose this now qualifies us as AI consultants. We considered using a stock photo of a gavel, or a suited figure staring pensively out of a window, or a handshake over a mahogany table—the visual vocabulary of every thought-leadership PDF since 2009. The AI-updated mug won because it cost nothing extra, which is consistent with our general approach to expenditure and, coincidentally, with the approach our former firms take to junior staffing.

The admission

There it is. That is the list, or at least the portion we can disclose without our lawyers sending emails. The lawyers, incidentally, also used to work at the firms. The conflicts nest. It is very efficient. Every item is true. None of it is a secret. Some of it is genuinely funny, if you have spent enough years inside the system to recognise the absurdity of a man drinking from a chipped corporate mug while writing about the corporation’s decline.

And we are going to keep writing anyway.

The accusation rests on a premise that sounds intuitive but is wrong: that personal benefit and honest criticism cancel each other out. They do not. A doctor who smokes can still read a chest X-ray. A retired general collecting a military pension can still identify a flawed battle plan. A former partner receiving deferred compensation can still count the number of times PCAOB deficiency rates doubled in three years. (It was once. From 12% to 26%. The deferred compensation did not obscure the arithmetic. Neither, we note, did the complimentary espresso in the partner lounge.)

If the billing economics we described in The Leverage Trap are wrong, show us the numbers. If the scandal timeline contains errors, name them and we will publish a correction the same day. If our assessment of AI readiness in $10 Billion in AI Investment overstates the problem, produce the evidence. These are standing offers. We mean them. We even mean them on the days the retirement deposit clears.

But “you still get retirement payments” is not a rebuttal to any of that. It is a change of subject. We recognise the move because we perfected it. We have a plaque. It is in a drawer. We do not display it, but we do not throw it away either. See: mug.

We know this move

Twenty-plus years inside these firms taught us one reliable pattern: when someone raises an uncomfortable question about the work, the first instinct is never to address the work. The first instinct is to question the person. “They don’t really understand the methodology.” “The client isn’t technical enough to evaluate this.” “She has her own agenda.” Redirect attention from the substance to the source, and the substance never has to be defended. It is the oldest trick in the deck, and the deck is branded, and the deck was billed at £4,200 a day.

The hypocrisy accusation is that play, run from the other side of the table. Do not engage with the utilisation maths. Do not argue the conflict-of-interest analysis. Do not explain why four consecutive years of admitted audit failure at a single client produced a fine that rounds to zero against firm revenue. Just point out that the people asking the questions used to benefit from the system, and declare the questions invalid. Applaud the redirect. Order another round. The pastries are very good at these meetings. They are always very good.

We deployed that move in client meetings, partner retreats, and press briefings for two decades. We know exactly how effective it is: very, right up until someone decides to ignore it and ask the question again.

So: we are hypocrites. Stipulated. Entered into the record. Noted with the same forensic enthusiasm we usually reserve for audit deficiency rates. We have initialled every page. We have added a cover sheet. The cover sheet has a disclaimer.

The maths, however, do not care who presents them. And the maths are correct.

If you disagree with the analysis, we would genuinely enjoy hearing why. We have a comment section. We have an email address. We have a standing offer to publish corrections, and we have enough professional pride to take a well-argued rebuttal seriously.

If, on the other hand, your strongest objection is that we still drink from the mugs, we would respectfully suggest that the argument deserves a better opponent. And possibly a better mug. Ours is chipped. Yours, we suspect, is branded.

We will be here. Writing, collecting retirement, and drinking coffee from a chipped mug we got for free in 2012. Some habits, like some business models, outlast the logic that created them. The difference is that the mug still works.

This is a genuine invitation. If you think the hypocrisy disqualifies the analysis, pick any post we have published, identify the specific claim you believe is wrong, and make your case. We will engage with it publicly and we will do so without charging our day rate.

In late 2013, PwC partner Peter Collins joined Treasury’s BEPS Tax Advisory Group, one of several private-sector experts consulted on international tax reform. Collins, PwC Australia’s head of international tax, was a credible choice. He signed a confidentiality undertaking on 11 December 2013. The domestic measure that became the Multinational Anti-Avoidance Law came later, with an exposure draft released in May 2015. The MAAL was designed to close the loopholes that let multinationals shift profits out of Australia and into low-tax jurisdictions. Collins signed further confidentiality undertakings in April 2016 and February 2018.

Three separate undertakings, each one making explicit that the information Collins received was confidential, that it was not to be shared outside the consultation process, and that it was certainly not to be used for commercial advantage. Collins agreed to all of this. Then he did the opposite, systematically, for years.

Confidential OECD material had circulated internally during 2014. By mid-2015, the emails show a more explicitly commercial operation forming around the MAAL: confidential policy intelligence was being used to inform client approaches and restructuring work. The firm assembled what the Senate inquiry later described as a “global team” to win new business on the back of confidential government policy. PwC’s first tranche of the effort, dubbed Project North America, contacted at least 14 US multinational companies and advised them on restructuring their tax arrangements to sidestep the very law that Collins was helping to write. In August 2015, a PwC colleague emailed a Google employee to confirm the MAAL’s likely start date, information that came directly from confidential government briefings. The email did not mention that the information was confidential.

The revenue PwC generated from this effort was approximately A$2.5 million. Remember that number, because it will become important later.

The Distribution Network

The detail that matters most in this case is not what Collins did. One partner breaching a confidentiality undertaking is serious but containable. Firms deal with rogue individuals. They investigate, they terminate, they move on. What makes the PwC Australia scandal different is how many people received the information and what they did with it.

When the Senate Economics Legislation Committee tabled 144 pages of internal PwC emails in May 2023, the scale of the distribution became visible. At least 53 redacted PwC email addresses, spanning offices in Australia, the United States, the United Kingdom, and Ireland. PwC’s own internal review subsequently identified a broader set of recipients across the full chain of correspondence. Partners. Directors. Tax practice leads. People in positions of seniority who would have understood immediately what they were looking at and where it came from.

One of those emails, sent by partner Paul McNab to then-head of tax Tom Seymour in January 2016, captured the culture in a single sentence. McNab wrote that PwC had been “aggressive in telling these relationships they needed to act early,” and noted that this was “heavily helped by the accuracy of the intelligence that Peter Collins was able to supply us.” Seymour, who would later become CEO, replied the next day and copied in additional PwC personnel.

“The accuracy of the intelligence.” Not “information.” Not “insight.” Intelligence. The word choice tells you everything about how the participants understood what they were doing. They were running an intelligence operation inside a professional services firm, using confidential government policy as the raw material and multinational tax advisory as the product.

Anyone who has worked in a Big 4 tax practice knows how this would have played out operationally. The partners receiving Collins’s intelligence did not sit on it and agonise over its provenance. They picked up the phone. They called their relationship leads on US multinational accounts. They scheduled meetings. They developed restructuring proposals, complete with the specific detail that no competitor could match, because no competitor had a partner sitting inside the government’s policy development process. The value proposition was speed: act before the legislation takes effect, and we can show you how to avoid it. Speed was the product. The intelligence was the input.

Nothing in the disclosed chain shows anyone stopping the initiative or escalating the confidentiality concern. Not the Office of General Counsel. Not the risk team. Not Tom Seymour, who received the email, replied to it, and widened the distribution. Nobody in the available record read the words “intelligence that Peter Collins was able to supply us” and asked whether that intelligence was supposed to leave Collins’s desk. In a partnership culture where revenue generation is the primary measure of contribution, the partner who shares information that wins business is a hero. The partner who flags a compliance concern that kills a revenue opportunity is a nuisance.

How the Wall Was Supposed to Work

Every Big 4 firm will tell you it has controls. Confidentiality walls. Conflict-of-interest registers. Independence protocols. Annual sign-off requirements. Training modules with quizzes at the end. PwC Australia had all of these. The firm’s own later review noted that at no point did anyone identify “the potential conflict of interest that arose from having client-facing partners participating in confidential Government consultations,” nor did anyone alert the Office of General Counsel or the risk team.

PwC’s own subsequent findings make clear that the failure was in design as well as enforcement. The firm acknowledged that the inherent conflict created by having client-facing partners participate in confidential government consultations was never recognised, that no central mechanism recorded the undertakings and notified affected teams, and that commercial incentives operated without an independent counterweight. The wall was a policy document and a set of procedures, the kind that sounds rigorous when described in a pitch and looks thorough when printed in an annual transparency report. In practice, it was a partition in an open-plan office. Everyone walked around it when they had a reason to.

The reason, in this case, was commercial opportunity. Collins had access to information that gave PwC’s tax advisory practice a competitive advantage. Sharing that information meant PwC could approach multinational clients before their competitors did, with specific knowledge of what the legislation would contain and when it would take effect. The commercial incentive to share was direct and measurable. The compliance incentive to stay quiet was diffuse and, as events proved, unenforced.

We have all seen this dynamic. Every former Big 4 partner reading this has sat in a meeting where information that should have stayed behind a wall was shared, casually, because the partner sharing it was trying to win a pitch and the information made the pitch stronger. The difference between PwC Australia and every other instance is not that the wall failed. The wall fails routinely. The difference is that someone, eventually, reported it.

The Decade of Delay

The ATO began scrutinising MAAL-related structures during 2016. By October 2017, it had identified that Collins might have shared information covered by Treasury confidentiality obligations. Suspicion fell on PwC.

What followed was one of the most frustrating institutional failures in Australian regulatory history. The ATO tried to investigate. PwC responded by claiming legal professional privilege over tens of thousands of documents. The ATO’s second commissioner later told Senate estimates that the investigation was “frustrated through false LPP claims.” The ATO shared information with the AFP in 2018. The AFP concluded it did not have enough evidence to pursue a formal investigation, partly because privacy rules prevented the ATO from sharing the material the AFP would have needed.

So the ATO referred Collins to the Tax Practitioners Board in July 2020. The TPB, a small regulator without the resources or the mandate for a case of this complexity, took until the end of 2022 to reach its decision. And that decision, when it came, was modest: Collins was deregistered as a tax agent and barred from reapplying for two years. The TPB chair expressed “concern” about tax practitioners who “abuse their positions of trust.”

Two years. For a systematic breach of three confidentiality undertakings that fed a multinational intelligence-sharing operation across four countries. The TPB did also impose remedial orders on PwC for failing to manage conflicts of interest, requiring improved confidentiality registers, reporting processes, and training. Those measures were real, but they fell well short of accountability across the wider network that had commercialised the information. The system had processed the breach the way a body processes a low-grade infection: just enough immune response to keep functioning, not enough to clear it.

If the story had ended there, it would have disappeared. One paragraph in a regulatory bulletin. A footnote in compliance training. The system would have absorbed the breach the way it absorbs most breaches: quietly, without consequence, without change.

The Senate Blew the Doors Off

The Australian Financial Review broke the story in January 2023, drawing on the TPB’s published findings. That caught the attention of the Senate, which was already running a broader inquiry into consulting services. On 2 May 2023, the Senate Economics Legislation Committee tabled 144 pages of internal PwC emails that the TPB had provided as answers to questions on notice.

Those emails changed everything. They showed the scale of internal distribution, the language partners used when discussing the leaked intelligence, and the degree to which senior leadership was informed. They also showed that PwC had known about the problem for years and had chosen containment over disclosure.

Tom Seymour’s response in March 2023, before the emails were released, was revealing. At the Financial Review’s Business Summit, Seymour described evidence that 20 to 30 partners and staff had received confidential information as a “perception problem.” Six weeks later, when the full email cache made the scope undeniable, he resigned.

Nine partners were stood down. Four were named publicly: Collins, McNab, Bersten, and Fuller. The acting CEO, Kristin Stubbins, issued an open letter apologising on behalf of the firm, while simultaneously resisting calls to release the names of all individuals involved. The Treasury referred the matter to the AFP for criminal investigation. Federal agencies reviewed and in many cases paused PwC engagements; by April 2024, PwC and the Department of Finance had agreed that PwC would temporarily refrain from bidding for new Commonwealth work.

The Price of A$2.5 Million

The commercial consequences landed fast and hard. PwC entered an exclusivity agreement to sell its government consulting practice to private equity firm Allegro Funds for A$1. The business, rebranded as Scyne Advisory, had been valued at over A$100 million. PwC sold it for a single dollar because no other outcome was politically survivable. As long as PwC retained its government consulting arm, every government contract the firm held or bid for would carry the taint of the tax leak. The only way to cauterise the wound was amputation.

More than 1,400 partners and staff transferred into Scyne. Others departed through attrition, retirement, or redundancy. PwC was also embroiled in a separate but related dispute over expansive legal professional privilege claims. The ATO initially assessed penalties over 170 claims and settled for approximately A$642,000. Senate committees argued that PwC’s approach to privilege had frustrated regulatory scrutiny more broadly, though the ATO said the settlement was separate from the Collins confidentiality breach itself.

In its FY2023-24 transparency report, PwC Australia disclosed a 24 per cent decline in profit and an A$820 million drop in revenue, bringing total revenues down to approximately A$2.5 billion. Not all of that contraction was caused directly by the scandal; the Scyne divestment removed a large business from PwC’s reported revenue base. But the asymmetry between cause and consequence is unmistakable.

The first tranche of Project North America produced at least A$2.5 million in fees. The consequences cannot be reduced to an equally precise counter-number, but they include a government consulting business transferred at a nominal A$1 sale price, years of regulatory and parliamentary scrutiny, and a collapse of trust that no financial statement can fully isolate. Kevin Burrowes, installed as the new CEO, disclosed an Australian salary of A$2.8 million but later revealed a further annual payment of approximately A$1.2 million from PwC International that had not been communicated to the wider Australian partnership. That tells you something about the culture of disclosure even after the crisis that was supposed to have reformed it.

What the Firm Actually Did Wrong

The popular narrative is that Peter Collins went rogue. That is the narrative PwC preferred, and for a while, the regulatory system accommodated it. The ATO’s initial referral to the TPB named Collins alone. When the TPB expanded its inquiry to examine PwC’s institutional role, the ATO, citing secrecy provisions, refused to provide further documents, frustrating the TPB’s broader investigation. Later proceedings extended to other senior figures: in September 2025, the TPB terminated former CEO Tom Seymour’s tax-agent registration and imposed a four-year bar.

The one-bad-apple framing is comforting. It implies the system works and this was an aberration. But look at what actually happened after Collins shared the intelligence. He did not have to persuade reluctant colleagues to use the information. He did not face internal resistance or whistleblower reports from partners who received the emails and recognised the breach. The information flowed through the firm’s global tax network the way all useful commercial information flows through a Big 4 partnership: quickly, widely, and with enthusiasm. Partners did not push back. They asked how to monetise it.

The firm’s response, once the ATO started asking questions, was not transparency. It was legal professional privilege claims, 170 of which were later found to be false. PwC used the legal architecture designed to protect legitimate client communications to obstruct a regulator investigating the firm’s own misconduct. That is not one bad apple. That is institutional behaviour.

And that institutional behaviour has a logic to it that anyone who has been a partner at one of these firms will recognise. The first priority in a crisis is containment. Limit the exposure. Identify the smallest possible number of individuals who can take the blame. Invoke every procedural shield available. Wait for the news cycle to move on. The calculus is simple: the cost of transparency (partner departures, client losses, regulatory escalation) is immediate and quantifiable. The cost of concealment is uncertain and may never arrive. Partners optimise for the certain cost, every time.

The Pattern Repeats

Eighteen months after the PwC Australia tax leak became front-page news, PwC faced a second crisis on a different continent. In September 2024, Chinese regulators imposed a six-month business suspension on PwC’s mainland auditing arm, PwC Zhong Tian, plus fines totalling 441 million yuan (approximately US$62 million) over its audits of collapsed property developer Evergrande. The China Securities Regulatory Commission said PwC had “turned a blind eye” to and “even condoned” Evergrande’s fraud, finding that 88 per cent of PwC’s real-estate site-observation records were inauthentic and that Evergrande had inflated revenues by approximately US$78 billion across 2018 to 2020. PwC had audited Evergrande for nearly 14 years. In April 2026, Hong Kong’s Accounting and Financial Reporting Council imposed a further HK$300 million fine over the same audits.

More than 50 Chinese companies dropped PwC or cancelled plans to appoint it, including Bank of China, which switched to EY. The client exodus triggered layoffs across PwC’s China practice before the mainland ban even formally began.

Then, in February 2025, Saudi Arabia’s Public Investment Fund restricted PwC from competing for certain new advisory assignments across PIF and its subsidiaries, cutting the firm out of Vision 2030 mega-projects that had been a growth engine for its Middle East practice. The Financial Times linked the dispute to PwC’s attempt to recruit NEOM’s chief internal audit officer, though neither the precise rationale nor the full scope of the restriction was publicly established. PwC cut approximately 1,500 staff and 60 partners across the region. The restriction was lifted in January 2026.

Three crises in three jurisdictions in less than two years. Australia, China, Saudi Arabia. Different circumstances, different regulators, different triggers. But the same firm, exhibiting the same institutional reflexes: protect the revenue, contain the damage, minimise the number of people held accountable, invoke procedural defences, and hope the next quarter’s numbers are strong enough to make everyone forget.

The connecting thread is not recklessness or corruption in the way those words are normally understood. Nobody at PwC woke up and decided to enable fraud or betray government confidences for the thrill of it. The connecting thread is a governance model in which the financial incentives to cut corners consistently overwhelm the compliance incentives to maintain standards, and in which the partnership structure means that the people who would need to enforce those standards are the same people whose income depends on the revenue those corners generate.

The Question Nobody Wants to Answer

PwC’s response to the Australian scandal was to sell the government consulting business. The logic was that removing the government advisory practice would remove the conflict of interest that enabled the breach. But the MAAL leak did not happen because PwC had a government consulting arm. It happened because a partner with access to confidential information had colleagues with a commercial incentive to use it, and no control in the firm was strong enough to prevent the transfer.

Selling the government practice addressed the symptom. The information flowed because the incentive to share exceeded the consequence of sharing. That incentive calculus has not changed. It cannot change within the current model, because the current model depends on partners cross-selling across practice lines, sharing client intelligence to win adjacent engagements, and treating the firm’s collective knowledge base as a competitive asset. Confidentiality walls ask partners to do the opposite of what the business model rewards them for doing. Every time.

KPMG Australia demonstrated this in 2026, when confidential audit-client data crossed the firm’s “ethical divider” into teams pitching for rival clients’ audit work. Different firm, different data, different regulatory obligations, but an analogous mechanism: confidential information crossed an internal boundary at the point where doing so could support a commercially important pitch. The whistleblower who reported it was treated the way PwC’s regulatory interlocutors had been treated: as the problem, not the solution.

Two analogous failures at two different firms in three years, in the same country. If a control fails twice in comparable circumstances, the issue is probably not that both firms implemented the control badly. The more plausible conclusion is that policy-only walls cannot hold where both sides share a P&L and the commercial information can move without technical friction.

A$2.5 million in advisory fees. An A$820 million revenue decline in a single year. A government consulting business transferred for a dollar. At least 53 recipients of confidential intelligence and, in the disclosed record, not one flag raised. Those numbers tell you everything about where the incentives actually point, and how far the compliance architecture is from being able to redirect them.

Is there a design for internal confidentiality walls that actually works at scale? Not in theory. Not as described in a transparency report. In practice, inside a Big 4 partnership, where the people behind the wall and the people in front of it share a compensation pool and a single P&L. Has anyone seen it work?

“Amazing race, amazing teamwork, Amazing platform to network across service lines.”

“Amazing job doing amazing things with amazing people!”

“Amazing news with even more amazing partnership.”

Every one of these was posted within the last twelve months by a serving partner at a Big 4 firm. (The capital A in the middle of the first one is original. We checked.) Nobody had to hunt for them, either. Ten minutes of scrolling partner LinkedIn and the feed delivers them to you, the way the ocean delivers plastic.

Stay a little longer and you get the full range of the genre. The team is amazing. The client is amazing. The deal is amazing. The offsite is amazing, as is the city hosting it, the lounge at its airport and, in one documented case, the air. The panel was amazing (the poster was on it). The insights were amazing (the poster delivered them). When “amazing” needs a rest, “brilliant” and “fabulous” come off the bench, and an “incredible journey” is always warming up. One post in our collection reports “collective strength of EY MENA unleashed” at an internal leadership summit, which sounds less like professional services than the third act of a superhero film. A partner elsewhere in the folder came away from a session with “great learnings!!”, two exclamation marks, presumably because one could not have carried all that learning alone.

One of us keeps a running file of these. (It is not a small file.) The current record is a thirty-word post that uses “amazing” six times, which works out to one every five words, a cadence otherwise achieved only by toddlers at aquariums. We are aware of how petty it is to count. We counted anyway, because the number is the observation: the most senior people in the world’s most expensive advisory firms have converged, in public, on a single adjective, and the adjective is the one that says the least.

Same partner, different room

The generous reading is that the enthusiasm is sincere, and we would extend it if we had not spent thirty years in the rooms where the same partners discuss the same teams and clients with the door shut. The private register exists. It is fluent and specific, and it does not contain the word amazing.

Take a composite week, one all of us have lived through more often than we would like to admit. Monday, the partner posts about the amazing team and the amazing energy at the client workshop. Tuesday is talent calibration, where two members of that team are described as underperforming the grade and a third as a flight risk who should be encouraged toward the exit before promotion season complicates the math. Wednesday is the account review, where the amazing client has not paid the last two invoices, cannot make a decision without four pre-meetings, and has pushed the scope to a place where somebody finally says write-off. Thursday the partner spends forty minutes on the phone trying to move the team’s weakest senior onto someone else’s engagement, praising them in terms Monday’s post would recognize. Friday, another post: amazing week, amazing people, grateful.

None of the private talk is scandalous. It is the ordinary candor of people paid to assess things, and it proves the diagnostic muscle still works fine. What the communications machinery actually controls is location. Enthusiasm got licensed for outdoor use; candor moved indoors, like smoking.

And the audience knows. The team praised in Monday’s post was the subject of Tuesday’s calibration and heard about it within the week, because juniors always hear. The client has read enough posts about other amazing clients to decode the ones about themselves. The performance continues anyway, which tells you the applause it seeks is not theirs.

Partners used to sound like owners

For most of the profession’s history, partners made almost no public sound at all. Client rooms, yes; the record, rarely. When the partner voice did appear in public, it showed up in technical journals, in comment letters informing the standard-setters that they were wrong, in conference speeches that contained actual positions. That writing was dry, precise, sometimes combative, and it had the one credential the current feed cannot counterfeit: the confidence of a professional who answered to clients and to the partnership and to nobody else. Some of us wrote those pieces. We fought over single sentences for days, because the sentences were ours and the name underneath them meant something. Nobody ever asked us to be amazing.

The shift to the current register happened inside fifteen years, which is fast for a profession that took the better part of a century to agree on how to depreciate a building. Four things drove it, and they compound.

Four things changed

First, the title kept its name and lost its meaning. A partner in 1995 was an owner: capital genuinely at risk, a vote that counted, a personal franchise, and unlimited liability to concentrate the mind. The past two decades added salaried partners, income partners, directors doing partner work without the equity, and, above the whole pyramid, a global executive layer that behaves like the management of any multinational, because that is what it is. A present-day partner has a capital account and a boss. People with bosses communicate like people with bosses. Middle managers do not publish opinions; they publish enthusiasm, aimed upward. We watched colleagues make that adjustment in real time. The quickest studies did best.

Second, the scandals taught management which control is cheapest. Improving audit quality costs money, takes years, and shows up in nobody’s quarter. Controlling what partners say about it costs a policy memo. So after every failure of the past twenty years, the reliable institutional output was a communications upgrade: the social media guideline that grew from one page into a manual, the media training that teaches everyone to bridge to the positive, the review chain that added legal, then brand, then risk. After #MeToo the ratchet turned again, and the list of things a partner must never touch in public reached its current form. No clients (confidentiality). No competitors (defamation). No regulators (relationship). No politics (policy). No commentary on the firm’s own troubles (career). Take all of that out of a working professional’s life and ask what remains to post about. Adjectives.

Third, marketing discovered the distribution channel. Somewhere in the mid-2010s, a firm with three thousand partners noticed it was sitting on three thousand broadcast towers, and the machinery followed: social selling programs, employee advocacy platforms, the Tuesday email from brand containing pre-drafted posts and a cheerful invitation to personalize before sharing. Several firms track the output on internal dashboards, and visibility has become a line in year-end conversations. The posts are also recruiting collateral, since a 24-year-old choosing between offers reads the feed, and the feed must therefore show grinning teams at offsites rather than anyone at a desk at 1 a.m. A meaningful share of what appears under a partner’s name was drafted two floors away from the partner. We have seen the content calendars.

Fourth, the internal tournament moved onto the feed. Partners once differentiated on a client book or a technical reputation, and that competition happened in rooms. As the firms institutionalized their big relationships — global accounts, rotating lead partners, succession by committee — the personal rolodex depreciated, and visible devotion appreciated in its place. A compensation committee cannot easily measure judgment. It can count posts. Tagging the regional managing partner in a tribute to an amazing leadership summit is a compensation argument, made in public, with timestamps. And the likes travel the direction everything travels in a partnership: upward. The team likes the partner’s post inside an hour, the partner likes the practice leader’s, the practice leader likes the CEO’s. Forty reactions, thirty-eight of them from people whose year-end ratings the poster controls, and every participant knows it and none of them says it.

The null post

Combine the four and only one register remains available. A population of formerly independent professionals, employees now in everything but tax treatment, barred from specifics, supplied with pre-drafted enthusiasm, and rewarded for public displays of loyalty, will converge on the word that commits to nothing. No claim, no number, no comparison, no opinion for a general counsel to flag. “Amazing” is the null post: maximum visibility, zero risk. The LinkedIn edition of the eighty-slide deck with no recommendation in it.

The vocabulary is a symptom. The old dry voice, whatever its charms, was evidence of the product clients believed they were buying: a senior professional with the standing to say an unwelcome thing. A partner who cannot risk a public opinion on the industry, the standard, or the weather is telling you, well in advance, how the contested judgment call with your CFO will go in the fourth quarter. Clients read the feed, whether or not they think of it as reading. So does the senior talent the firms keep wondering how they lost; a 32-year-old with options looks at what a partner is now permitted to sound like and prices the title accordingly.

None of which makes the people in our folder fools. Posting this way is the rational move inside the current incentive set, most partners privately read their own output exactly the way you do, and the eye-rolling in the partner lounge is close to universal. The interesting question is who built the incentive set, and the pattern is the clue: adjective density tracks corporatization. Big Law developed the same condition on roughly the same schedule, as its partnerships centralized. The founders of ten-person boutiques, meanwhile, still post opinions with verbs in them, because they still own what they built and nobody reviews their drafts.

So, a challenge for this comment section, offered as a genuine research question and only partly as a trap: can you find one public statement from the past twelve months, by a serving Big 4 or MBB partner, that a client, a regulator, or the firm’s own brand team could plausibly object to? An article, a post, a conference remark, anything on the record. Post the link. We went looking ourselves and came back with a folder full of amazing.

Last week the FRC fined PwC £3.24 million for failing the audit of Babcock International’s 2019 and 2020 financial statements. The firm admitted to breaches across cash pooling, financing arrangements, aircraft-related costs, goodwill impairments, intangible assets, and long-term contracts. Audit engagement partner John Waters received a separate £59,000 penalty. Both were “severely reprimanded.”

This was the second time the FRC sanctioned PwC over Babcock.

In March 2023, PwC and two different former partners were fined a combined £8 million (discounted to £5.65 million) for failures in the 2017 and 2018 audits of the same company. The FRC’s investigation that time found something extraordinary: PwC’s audit team had never obtained or read a 30-year Public Private Partnership contract worth £3 billion in lifetime revenue, accounting for approximately £77 million in Babcock’s 2018 revenues alone. A separate contract valued at roughly €640 million was written in French. The audit team did not read French. They did not commission a translation.

Four consecutive years of audit failure. Two investigations. Two sets of partners sanctioned. One client. The same firm.

Babcock is not a corner shop. It is a FTSE-listed defence contractor that manages nuclear submarine refits, military flight training, and critical UK government infrastructure. The public-interest dimension of getting this audit right could not be more obvious, and PwC missed it four years running.

The arithmetic of deterrence

PwC UK’s consolidated group reported £6.35 billion in revenue for the year ending June 2025, with profits of £1.37 billion. Average distributable profit per UK partner was £865,000.

The 2026 Babcock fine of £3.24 million represents 0.05% of that revenue. Add the 2023 fine and the combined total reaches approximately £8.89 million across both investigations, or less than 0.14% of a single year’s UK revenue. The combined penalty for four years of admitted audit failure at a major defence contractor is roughly what ten PwC UK partners took home last year.

For comparison, when Hong Kong’s AFRC sanctioned PwC over its Evergrande audit, the combined penalties reached HK$1.3 billion (roughly US$166 million), and the firm was banned from taking on new publicly listed audit clients for six months. PwC lost more than 50 major clients, including Alibaba and Bank of China. Partners were fired. The China business was gutted.

Hong Kong treated an audit failure as what it is: a threat to market integrity requiring consequences that the firm would actually feel. The FRC treated an audit failure as an administrative matter to be resolved through cooperative settlement, with a 40% discount for admitting what was already proven.

The cooperation discount and what it rewards

Both Babcock sanctions were reduced substantially because PwC demonstrated “exceptional cooperation” and made “early admissions.” In 2026, the starting fine of £5.5 million was cut to £3.24 million. In 2023, £7.5 million became £5.65 million.

This mechanism deserves scrutiny. The FRC has created a system where a firm’s fastest route to minimising penalties is immediate admission. That sounds reasonable in isolation: cooperation is generally preferable to obstruction, and early resolution saves public money. The problem is what it produces in aggregate. When the expected cost of admitting to years of audit failure at a defence contractor is £3 million, no rational partner in a firm earning £6.35 billion will lose sleep over the risk. The cooperation discount does not incentivise better auditing. It incentivises efficient confession.

Compare this to PwC’s own internal economics. If a senior manager’s utilisation drops five percentage points below target for two consecutive quarters, their career trajectory changes. If an engagement runs 15% over budget, the partner faces a difficult conversation with the practice leader. The firm’s internal penalties for underperformance are felt immediately and personally. The FRC’s penalties for failing a public-interest audit of a nuclear-submarine contractor are absorbed into overhead without anyone’s compensation being affected.

PwC’s statement

PwC responded with a sentence that could have been, and probably was, pulled from a template: “We’re sorry that some aspects of these audits were not of the standard expected. Audit quality is a constant focus for the firm and the impact of our commitment to continuous improvement has been underlined by recent inspection results.”

Read that again without the institutional framing and translate it into plain English: “We failed the audit. We are sorry. We are improving.” It is the same sentence the firm has issued after every sanction for as long as any of us can remember. The Babcock 2017–18 failures. The Wyelands Bank failures. The London Capital & Finance failures (which earned a first-ever FCA fine of £15 million, notably from a different regulator with more appetite for consequential penalties). The Evergrande catastrophe. The Colonial BancGroup settlement. After each one: sorry, improving, committed to quality. At what point does an apology become a press release format rather than an institutional response?

The FRC itself noted that Waters “faced difficult circumstances, taking over the 2019 audit at short notice without a handover.” The pandemic then disrupted the 2020 engagement. Both points are fair. But they raise a question the FRC did not ask publicly: why was PwC’s internal system for partner transitions so weak that a new audit partner could be dropped into a complex defence-sector engagement without adequate handover, on a client whose previous four audits would later be found to contain serious failures? The handover gap is not an excuse. It is another symptom.

What this tells us

We have argued previously that the Big 4 should be broken up, that the audit-consulting combination produces conflicts the firms cannot internally manage, and that fines calibrated to single-digit millions against firms earning tens of billions are the regulatory equivalent of a parking ticket.

The Babcock case adds a data point that is useful precisely because it is boring. Nobody stole anything. Nobody committed fraud. Nobody leaked confidential tax intelligence to 53 partners (that was PwC Australia). This is just an audit team that did not do its job adequately, four years in a row, on a client whose work directly involves national security. And the system’s response, from first failure to final settlement, took the better part of a decade and produced a total penalty that PwC UK earns back in roughly five hours of operation.

We have been adding to our running timeline of Big 4 and MBB scandals, and the Babcock case is now included. It sits in a crowded section. PwC alone has accumulated FRC fines, PCAOB penalties, a record Hong Kong sanction, a six-month China business suspension, a Saudi Arabia advisory ban, a £335 million Colonial BancGroup settlement, and the Australia tax-leak scandal that cost the firm its government consulting business, its CEO, and approximately A$820 million in losses.

At some point, the phrase “isolated incident” stops being an available defence. We passed that point several scandals ago.

The regulator that never arrives

The FRC was supposed to be replaced by a new statutory regulator, the Audit, Reporting and Governance Authority (ARGA), with expanded powers including the ability to impose operational penalties and potentially ban firms from specific client categories. The Conservative government shelved the plan. Labour revived it, but a legislative timetable has not materialised. The House of Commons Library noted as recently as July 2026 that ARGA’s creation remains outstanding.

Until ARGA exists, the FRC’s enforcement ceiling is capped at a level that the Big Four treat as an operational cost. The record FRC fine remains £21 million, levied against KPMG for the Carillion audit. KPMG UK’s revenue that year was approximately £2.6 billion. Even the record fine was less than 1% of revenue.

The firms know this. Every compliance partner at every Big Four firm has done the same calculation we just did. The expected cost of an audit failure, discounted by the probability of detection and the cooperation reduction, is negligible relative to the revenue generated by keeping the engagement staffed at minimum cost. The FRC cannot fix this without powers it does not have and a government that is in no hurry to grant them.

Babcock, for its part, has moved on. The company restated its 2021 financials to correct the errors the audits missed, and the share price recovered. The market has priced in the expectation that Big Four audits will occasionally fail and that the consequences will be absorbed. That expectation is, itself, the indictment.

We updated our scandal timeline to include the July 2026 Babcock fine. Combined with the March 2023 action, PwC has now been sanctioned for every Babcock audit from FY2017 through FY2020.

For anyone who has worked on a defence-sector audit at a Big Four firm: what does the partner transition process actually look like when a new engagement partner takes over mid-cycle? Is the handover gap that the FRC described as “challenging circumstances” the exception, or is it how these transitions typically work?

A friend on the buy side invited me to sit in on an advisory selection last month. Three firms. One mandate. Mid-market industrial company looking at a portfolio review and potential divestiture. The kind of engagement that used to be genuinely interesting work.

Each firm got ninety minutes. I sat through all three.

Every deck opened with a “proprietary framework.” Firm A called theirs the Strategic Value Matrix. Firm B called theirs the Portfolio Optimisation Lens. Firm C used something like the Growth-Complexity Navigator. All three were 2x2 matrices. High-low on one axis, high-low on the other. The quadrant labels differed; the logic was identical. Plot your business units, see which ones fall in the “divest” box, recommend further analysis on the ones near the boundary.

I have seen this matrix hundreds of times. I have presented versions of it myself, under at least four different firm logos over the course of my career. The template has not changed in twenty years. What has changed is the confidence with which partners now present it as proprietary thinking. It is not proprietary. It is a recombination of ideas from BCG’s growth-share matrix (1968), McKinsey’s GE-McKinsey nine-box (1970s), and the portfolio logic taught in every MBA strategy course since the 1990s. Everyone in the industry knows this. Nobody says it in the room.

Two of the three firms proposed a “phased approach.” Phase 1: diagnostic and data gathering, six to eight weeks. Phase 2: strategic assessment and scenario modelling, six to eight weeks. Phase 3: recommendations and implementation roadmap, four to six weeks. The timelines were so similar that if you shuffled the decks and removed the logos, you could not tell which firm proposed which. The third firm proposed the same phases under different names and added a fourth (”stakeholder alignment”), presumably to justify charging for a few extra weeks.

Then came the detail that stayed with me. The partner leading Firm C’s pitch had been at Firm A two years earlier. I know because I worked with him at Firm A. He was pitching the same client, using the same framework, with the same phased methodology he had used at his previous firm. The only differences: his business card, the colour palette on his slides, and the proprietary name stamped across the top of the 2x2.

I do not begrudge him the career move. Partners move between firms constantly, and they take their playbooks with them because the playbooks are the only consistently portable asset a partner has. But his presence at that pitch table made visible something the industry prefers to keep invisible: the firms are selling interchangeable products under competing brand names.

This convergence did not happen overnight, and we did not need this particular pitch to notice it. We started seeing it decades ago. The decks were already getting similar by mid-2000s. Talent rotation was homogenising the analytical traditions. When partners and senior managers cycle between firms every few years (sometimes every 18 months at the VP and principal level), they carry their methods with them. The receiving firm absorbs those methods. McKinsey partners who move to Deloitte bring McKinsey’s frameworks. BCG managers who move to EY bring BCG’s hypothesis structure. Over a decade of constant cross-pollination, the approaches merged. Consolidation accelerated it: when PwC acquired Booz & Company in 2014 to create Strategy&, and Deloitte absorbed Monitor Group to build Monitor Deloitte, the acquired boutiques lost their distinctive analytical cultures within a few years. The industrialisation of methodology continued it, with every firm building internal “knowledge libraries” and “accelerators” that packaged engagement types into repeatable templates.

But there was still, ten years ago, a residual difference in quality. A team that had done its homework, spent time with the client’s operations, interrogated the data with genuine curiosity, could produce something the client had not already thought of. The framework might be familiar, but the thinking applied within it could surprise you. Occasionally, a pitch would contain a genuine insight, a connection the client had missed, an angle that reframed the problem in a way that justified the fees. Those moments were already rare by 2016. They have now almost vanished. AI finished the job.

Every firm now runs its pitch preparation through large language models. The analysts use them to draft market analyses, synthesise industry reports, generate scenario frameworks, and write the narrative sections of the deck. This is rational behaviour and we are not objecting to it on principle. What we are observing is what it produces in aggregate: when six competing teams feed the same public data into the same family of foundation models and ask for a strategic assessment, the outputs converge on the same median analysis, expressed in the same register, structured with the same logical scaffolding. The idiosyncratic brilliance of a team that had genuinely wrestled with a problem, argued about it internally, and produced something unexpected gets replaced by fluent, competent, and utterly interchangeable prose.

We have, among the contributors to this publication, become embarrassingly good at identifying which model produced which section of a pitch deck. Claude has a particular cadence: measured, structured, slightly formal, with a tendency toward precise qualifications. ChatGPT produces a breezier, more expansive register, often with a characteristic pattern of listing implications in groups of three. Copilot-assisted sections carry a recognisable Microsoft house style, and Gemini has its own tells (a certain flatness when synthesising, an inclination toward neutral balancing of perspectives). Read enough AI-assisted pitch decks across enough firms and you start pattern-matching the model before you have finished the second paragraph.

We mention this not to embarrass anyone. Every firm is doing it. We would have done it too, had these tools existed during our tenure. The point is that “proprietary framework” has become an even more absurd label than it was before. The frameworks were already interchangeable when they came from the same business-school case studies and the same partner playbooks rotating between firms. Now they come from the same four or five language models available to everyone. The label “proprietary” applied to AI-generated analysis is not just a marketing claim. It is a fiction.

What used to differentiate a great team from a competent one was the willingness to go beyond the template: to spend a weekend reading the client’s SEC filings from five years back, to call twelve industry contacts and synthesise what they heard, to argue with each other until the analysis said something the client could not have produced alone. AI has made the template-level work almost free, which should in theory liberate teams to spend more time on the genuinely original thinking. In practice, the economics run the other direction. If AI can produce an 80% draft in two hours, the engagement manager ships the 80% draft. The remaining 20%, the part that required judgment, experience, and actual intellectual commitment, gets trimmed because the utilisation model does not reward the extra time it takes to produce it.

That brings us to what made the whole exercise feel slightly absurd. Before the three pitches, I had coffee with two members of the client’s internal strategy team. They are a group of four: two former consultants (one ex-McKinsey, one ex-Deloitte), an industry veteran with fifteen years in the sector, and a data analyst who had built a set of automated models on their company’s operational and financial data. They had already produced a 40-page draft analysis of the portfolio. It included granular unit-level economics that none of the three pitching firms could access without six weeks of data requests. It incorporated scenario modelling that covered more permutations than any of the firms’ proposed Phase 2 workstreams. And it had something none of the external firms could offer: the internal team knew which plant managers would resist a divestiture, which union contracts contained change-of-control provisions, and which business units had undocumented customer relationships that would not survive a transfer.

The client was running the advisory selection because the board wanted external validation. The CEO told my friend as much: “We know what the answer is. We need a name on the cover.” That is the insurance policy theory of hiring consultants (a subject for another post in this publication), and it is rational behaviour on the CEO’s part. But it means the three firms were competing to provide a branded stamp of approval on analysis that had already been done, more thoroughly, by four people who actually understood the business.

The fees proposed ranged from £800,000 to £1.4 million. The internal team’s cost for producing the superior analysis was their existing salaries and roughly £15,000 in software subscriptions.

I left the final pitch early. On the way out, I passed the engagement manager from Firm B in the hallway, rehearsing talking points on her phone. She looked tense. I recognised the expression from fifteen years of watching my own teams prepare for competitive pitches: the mixture of performance anxiety and the quiet awareness that what you are about to present is, in every way that matters to the client, indistinguishable from what the competitors presented an hour ago.

She will probably make partner in a few years. She is good at her job. The job has just stopped requiring anything that the other firms’ people cannot do equally well.

When was the last time you saw a genuinely differentiated pitch from a Big 4 or MBB firm? Not a better deck or a smoother partner. A fundamentally different way of thinking about the client’s problem. If you can name one, we would like to hear about it. If you cannot, that is the point.

KPMG Australia announced today that John Sams will be the firm’s new CEO, effective immediately. The board described the process as “rigorous,” with candidates considered “from within KPMG Australia, across the international network and externally.”

They picked their own CFO.

Sams joined KPMG UK as a graduate in 2003, transferred to KPMG Australia in 2006, made partner, ran the infrastructure advisory practice, became CFO in October 2025, and was given the additional title of COO on 3 June 2026. That last date matters. June 3 was five days after CEO Andrew Yates and head of audit Julian McPherson resigned. Whoever added “COO” to Sams’ title in the middle of a crisis was not responding to an organisational gap. They were staging a succession.

The selection panel that oversaw this “rigorous” search was chaired by Mike Ebeid, who joined KPMG’s board as an Independent Adviser in July 2025. Ebeid was appointed by Andrew Yates. Jennifer Westacott, also on the panel, was brought in as a Special Adviser to the firm on the same day, also by Yates. The people charged with finding a leader to fix the culture were themselves installed by the person who presided over the culture that failed.

None of this is unusual. It is, in fact, exactly how partnerships handle these transitions. And that is the problem.

When a firm promotes an insider after a governance crisis, it sends two messages to two audiences. To the outside world, it says: we searched widely and found the best candidate. To the partnership, it says: nothing about the power structure has changed. The second message is the one that matters, because the partnership is the institution. If the partners believed an outsider might arrive with a mandate to restructure compensation, dissolve practice groups, or fire senior partners who were complicit, the anxiety inside the firm would be genuine. The appointment of a 23-year KPMG veteran eliminates that anxiety.

Sams’ public statement follows the crisis-management script so closely it could have been generated from a template. He will be “courageous.” He will take “tough decisions.” He will address “culture, leadership and governance” with “resolve and endurance.” He will “deliver the Action Plan in full.” Every one of those phrases has appeared in at least five previous Big 4 post-scandal CEO statements over the past decade. I know, because I helped draft two of them.

The Action Plan itself runs to several pages of commitments: governance reform, culture review, enhanced controls, external oversight. It reads well. Action Plans always do. The question, as always, is whether the person tasked with executing the plan has the independence and the incentive to follow through when the reforms become inconvenient for the partners who elected them.

Sams was not elected by an independent board answerable to shareholders. He was selected by a board whose independent members were appointed by his predecessor, and he serves at the pleasure of a partnership that will judge him on revenue, retention, and the speed at which this scandal stops being a topic of conversation. Those incentives do not point toward deep reform. They point toward managed recovery: enough visible change to satisfy regulators and the parliamentary committee, not so much that the partnership’s economics are disrupted.

I hope I am wrong about Sams. A 23-year insider who actually forces the hard changes would be more effective than any outside appointment, because he knows where the problems are buried. But the track record of internal crisis appointments at the Big 4 is not encouraging. The firm needs to ask itself an honest question: if it had genuinely been willing to hire an outsider, would the process have looked any different from the one that produced this result?

Has your firm ever conducted an “external search” that ended with an internal candidate? What did the partnership read into it? We already know the answer at ours.

Every professional services firm runs on the same equation, and every partner knows it by heart even if they never write it down. The equation is simple: hire ten people, bill them at rates that exceed their cost by a factor of three or four, keep them busy at least 75% of the time, and extract the margin as partner profit. The people at the bottom do the work. The people at the top collect the spread. The entire organisational structure exists to maximise that spread.

This is the pyramid model. It made the Big Four and MBB what they are. And it is now eating their quality from the inside.

We are not the first to observe that professional services firms are built on labour arbitrage. But most commentary on the pyramid treats it as a business model critique, something to be debated in the abstract. We want to do something different: show you the specific numbers, the specific pressure points, and the specific mechanism by which the pursuit of higher utilisation and wider billing-rate spreads produces the declining quality that regulators are now measuring in black and white.

How the pyramid actually works

The economics are stark once you strip away the professional mystique.

A first-year analyst at a Big Four firm in the United States earns somewhere between $65,000 and $85,000 in base salary, depending on city and service line. The firm’s fully loaded cost for that analyst, once you add benefits, office space, technology, training, and administrative overhead, is typically estimated at $120,000 to $150,000 per year. That same analyst is billed to the client at rates between $200 and $400 per hour, depending on whether they sit in audit or advisory. At 1,500 billable hours per year (roughly 75% utilisation of available working time), that analyst generates $300,000 to $600,000 in fee revenue.

The gross contribution from a single first-year analyst, before write-offs and non-billable time, can reach $200,000 to $450,000 per year. Multiply that across the base of the pyramid and you have the economics of a Big Four partnership.

At the top of the billing schedule, GSA federal supply schedules published in 2024 list not-to-exceed ceiling rates for government contracts: a McKinsey senior partner at $1,194 per hour. A BCG senior partner: $1,116 per hour. McKinsey engagement managers sit at $834. Associates and analysts range from $327 to $498 per hour. Commercial rates are often higher. The gap between what a first-year associate costs the firm and what the client pays for that associate’s time is where the entire business model lives.

Partners understand this intuitively. They talk about it in terms of “realisation” (the percentage of standard rates actually collected) and “leverage” (the ratio of junior staff to partners on an engagement). An engagement with one partner, one manager, and eight associates or analysts is highly leveraged. An engagement with two partners and three senior managers is poorly leveraged. The economics of the first are spectacular. The economics of the second are mediocre.

The incentive, therefore, is always to increase leverage. Put more junior people on the engagement. Reduce the partner’s time. Bill the junior hours at the highest rate the client will tolerate. The partner who masters this arithmetic makes more money for the firm, gets promoted, and sets the template for how engagements should be staffed.

The utilisation ratchet

If leverage is the width of the pyramid, utilisation is the intensity at which it operates. Utilisation measures the percentage of a consultant’s available working hours that are billed to a client. The industry targets a range of 75% to 85%. Firms treat the lower bound as a performance floor and the upper bound as an aspiration, though they rarely say so in writing.

The SPI Professional Services Benchmark surveyed 403 firms in 2024 and found that the average billable utilisation across the industry had fallen to 68.9%, the lowest figure in five years. The benchmark identifies 75% as the optimal threshold; above 80%, firms risk burnout and quality erosion. The difference between 70% and 80% utilisation for a firm with 1,000 consultants, at an average billing rate of $300 per hour, is roughly $60 million in annual revenue. Same people, same offices, same overhead.

That sensitivity explains why utilisation is tracked weekly, discussed in every partner meeting, and treated as the single most important operational metric in every Big Four and MBB firm we worked at. When your utilisation report comes back at 68%, your practice leader calls you. When it drops below 65%, you are on a list. When it stays there for two consecutive quarters, your trajectory at the firm changes permanently.

The pressure flows downhill. Partners who face utilisation shortfalls staff engagements more aggressively, which means assigning fewer people to each project and expecting each person to carry more billable hours. We have seen, in our own firms, the implicit bargain that forms under this pressure: if you bill 80%, nobody asks whether the hours were productive. If you bill 65%, everyone asks everything.

We sat in those meetings. We watched partners present their utilisation dashboards with the same gravity that a hospital administrator would present mortality rates. And in a sense, the analogy is apt. Below a certain utilisation threshold, the firm’s financial health deteriorates rapidly. Above a certain threshold, the quality of the work deteriorates instead. The question the model never answers, and the one the partners never ask aloud, is: where is the crossover point? At what utilisation rate does the pressure to bill start producing work that should not carry the firm’s name?

The billing-rate gap

The pyramid creates a second, related problem: the widening distance between what the client pays and what the client gets.

A Big Four advisory engagement manager billing at $500 to $700 per hour has, in most cases, four to seven years of post-university experience. (Audit rates are lower, typically $300 to $500 for equivalent seniority, but the leverage dynamics are identical.) Some of that experience is excellent. Some of it consists of having run the same methodology on six similar clients and knowing which slides to reuse. The client pays a rate that implies seasoned expertise. The client receives someone who is competent, hardworking, and following a playbook written by someone more senior who is no longer on the engagement.

Across the industry, firms typically charge 2.5 to 4 times the loaded cost of the consultant actually doing the work. That multiplier is the pyramid in a single number. When a client pays $600 per hour for a senior consultant, perhaps $150 to $200 of that represents the consultant’s compensation and direct costs. The remaining $400 to $450 funds the partner’s draw, the firm’s overhead, and the margin that makes the partnership economics work.

None of this is secret. Sophisticated procurement teams at large corporations have understood these economics for years. What has changed is the willingness to accept them. Fifteen years ago, the brand premium justified the spread: you paid McKinsey rates because the McKinsey name on the report carried weight in the boardroom. Today, procurement teams are asking to see the resumes of the actual team members, demanding fixed-fee structures, and questioning why the partner who pitched the work billed four hours in a twelve-week engagement.

The pyramid depends on the client not asking that question, or not caring about the answer. Increasingly, clients are asking. And they care.

What the PCAOB data actually shows

If the pyramid model’s effect on quality were theoretical, this would be a thought exercise. It is not theoretical. The Public Company Accounting Oversight Board has been measuring audit quality at the largest firms for over fifteen years, and the data tells a specific, quantifiable story.

The PCAOB inspects audits and identifies “Part I.A deficiencies,” cases where the auditor did not obtain sufficient appropriate audit evidence to support the audit opinion. A deficiency does not necessarily mean the financial statements are wrong. It means the auditor did not do enough work to determine whether they are right — a distinction that matters legally but should concern investors either way.

The trajectory from 2020 to 2024 is consistent with the pattern you would expect if the pyramid’s incentive structure were a contributing factor.

In 2020, the Big Four’s aggregate deficiency rate was 12%. By 2021 it had risen to 16%. By 2022 it reached 26%, where it stayed through 2023. In 2024, following intense PCAOB pressure and public criticism, the rate improved to 20%. At the all-firm level, the picture is worse: deficiency rates climbed from 29% in 2020 to 46% in 2023 before improving to 39% in 2024.

Consider those numbers for a moment. In 2023, nearly half of all audits inspected by the PCAOB lacked sufficient evidence. Among the Big Four, one in four audits was deficient. Among the firms outside the Big Four, the numbers were catastrophic: BDO USA posted an 86% deficiency rate in 2023, meaning that in 25 of its 29 inspected audits, the PCAOB found the firm had not gathered enough evidence to support its opinion. Grant Thornton reached 54%.

The 2024 improvement is real and worth acknowledging. Deloitte dropped to 14%, the lowest among the Big Four. PwC reached 16%. KPMG hit 20%. EY, which had the worst Big Four record at 46% as recently as 2022, improved to 28%. BDO fell from 86% to 60%, a significant move in the right direction. The gap between the best and worst performers narrowed from 68 percentage points in 2023 to 46 in 2024.

But the improvement itself reveals the mechanism. By mid-2024, EY had already shed 84 audit clients since January 2023, a net loss of 63 public-company engagements at a cost of approximately $215 million in audit fees. By June 2025, the total had grown to 132 departures and a net loss of 101 clients. Deloitte, KPMG, and PwC all gained clients in the same period. EY’s vice chair for assurance acknowledged the strategy was intentional: reduce the portfolio to allow auditors to dedicate more resources to each remaining engagement.

In other words, EY improved quality by reducing the pyramid’s workload. Fewer engagements per auditor. More time per audit. Less utilisation pressure. The fix for the problem created by the model was to partially reverse the model’s own logic.

The up-or-out accelerant

The pyramid requires constant fuel. Junior professionals enter at the base, work for two to four years, and either advance or leave. The industry calls this “up or out,” and it is presented as a meritocratic tournament. In practice, it is a staffing model. The firms need a constant supply of inexpensive junior labour to maintain the billing-rate spread, and they need most of that labour to leave before it becomes expensive. (The ideal associate, from the model’s perspective, is one who bills intensely for three years and then leaves to make room for the next cohort — taking nothing with them but a line on their CV.)

According to SPI’s benchmark data, attrition rates across professional services ran at approximately 11.7% in 2024, down slightly from the five-year average of 12.8%. At the Big Four, voluntary turnover among junior staff is typically higher, often approaching 15% to 20% in competitive labour markets. The firms budget for this. They plan for it. When attrition drops below expected levels, as it did during parts of 2023 and 2024 when the external job market tightened, the firms face an overstaffing problem that they solve with layoffs, reduced intake, or both.

Deloitte UK promoted 60 employees to partner in 2025, down from 81 the year before; the 2026 round dropped further to 48. Partner promotions across the Big Four have contracted. KPMG is consolidating its global network from over 100 national entities to roughly 30. PwC reduced its global headcount by 5,600 in its 2025 financial year.

The up-or-out model creates a specific quality problem that every partner reading this will recognise. When a second-year associate leaves, they take with them whatever institutional knowledge they accumulated. Their replacement is a first-year associate who must learn the client, the industry, the methodology, and the firm’s internal systems from scratch. The client pays the same rate. The engagement manager absorbs the transition cost in unpaid hours of supervision. The deliverable quality dips during the transition, recovers partially as the new associate gets up to speed, and dips again the following year when that associate, now experienced and productive, also leaves.

Multiply this cycle across every engagement team at every Big Four firm, and you have a permanent quality drag built into the model’s DNA. The firms know this. They have known it for decades. The reason they tolerate it is that the economic benefit of cheap junior labour exceeds the economic cost of the churn. Quality is the variable that absorbs the difference.

The constraint that loosened

For most of the history of these firms, quality was maintained by two forces: partner involvement and client scrutiny.

Partner involvement meant that a senior professional with fifteen or twenty years of experience personally reviewed significant workpapers, challenged analytical conclusions, and took direct accountability for the engagement’s output. The partner’s reputation was on the line. Their name (in audit) or their client relationships (in advisory) provided a check on the pyramid’s tendency to push work downward.

Client scrutiny meant that the buyer of the work was sophisticated enough to notice when quality slipped. A CFO who reviewed the audit workpapers. A board member who read the strategy report and asked hard questions. A procurement director who compared this year’s deliverable to last year’s and noticed that the analysis was thinner.

Both constraints have weakened.

Partner involvement declined as the economics pushed partners to manage more engagements simultaneously. A partner overseeing eight audit clients cannot give each one the attention they would have given when overseeing four. The partner-to-engagement ratio has shifted in every firm we worked at, and the direction is always the same: more clients per partner, less time per client. The partner shows up for the kickoff meeting and the final presentation. Everything between — and that is where the actual audit work lives — is reviewed by managers and senior associates whose incentive is to keep the engagement moving, not to slow it down with questions.

Client scrutiny declined as corporate buyers increasingly purchased the brand rather than the work. “We hired Deloitte” is a statement that protects the executive who made the decision. If the work is mediocre, that is Deloitte’s problem; the decision to hire them was sound. This dynamic removes the buyer’s incentive to evaluate quality closely, which removes the external pressure that once forced the firms to maintain standards despite the pyramid’s internal pressure to cut costs.

When both constraints loosen simultaneously, the model does what it was always designed to do: maximise margin. Utilisation goes up. Leverage increases. Junior-to-senior ratios widen. The billing-rate spread grows. And quality declines, because quality was never the objective of the model. Quality was the constraint on the model. When the constraint weakens, the model optimises for what it was built to optimise for.

The numbers behind the improvement

The 2024 PCAOB data deserves closer reading, because the improvement, while genuine, does not mean the model has corrected itself.

A 20% deficiency rate for the Big Four means that one in five inspected audits lacked sufficient evidence. In 2020, that number was roughly one in eight. The “improved” 2024 rate is still materially worse than where these firms stood four years earlier.

The individual firm data tells a more granular story. Deloitte’s 14% rate in 2024 is the strongest performance among the Big Four, and its inspectors reviewed 63 audits to reach that figure. EY’s 28% rate, despite the client shedding and billion-dollar investment, still means that inspectors found significant deficiencies in 18 of 64 reviewed audits. In two of those 18 engagements, the firm’s opinion on internal controls was determined to be incorrect and had to be revised. In one, the client restated its financial statements entirely.

Outside the Big Four, the deficiency rates remain alarming. BDO’s 60% in 2024, while a genuine improvement from 86%, still means that 18 of 30 inspected audits were deficient. The PCAOB found 2.4 deficiencies per inspected BDO audit, eight times the rate at Deloitte. Grant Thornton posted a 48% deficiency rate, down from 54%.

The firms that improved did so through identifiable actions: investing in technology, standardising procedures, shedding high-risk clients, and responding to regulatory pressure. Those are real steps. But they are steps taken against the direction of the model’s natural incentives. The moment the regulatory pressure eases, the model’s incentives reassert themselves. Hire cheaper. Staff leaner. Bill higher. Utilise more.

And there are reasons to think the pressure is already easing. In February 2026, the SEC swore in Demetrios Logothetis as the new PCAOB Chair, replacing Erica Williams, whom SEC Chair Paul Atkins had pushed out in July 2025. Logothetis is a retired EY partner with forty years at the firm. The SEC simultaneously cut the PCAOB’s budget and reduced the Chair’s salary by 52%. You can draw your own conclusions about the direction of regulatory intensity.

We have watched this cycle before. The post-Enron reforms of the early 2000s produced a temporary improvement in audit quality. The improvement lasted until the memory of Arthur Andersen faded and the pressure to grow reassumed its primacy. The post-2008 scrutiny produced another temporary correction. The pattern is consistent: scandal produces regulation, regulation produces improvement, improvement reduces pressure, reduced pressure allows the model to reassert itself, and quality declines again.

What the model cannot fix

The pyramid is the firm itself, not a policy that management can simply revise. The partnership economics, the compensation structure, the staffing ratios, the career tournament, the billing model, the client relationship architecture are all load-bearing elements of the same structure. You cannot remove the utilisation pressure without reducing partner income. You cannot reduce leverage without increasing the cost per engagement, which means raising fees or cutting margins. You cannot slow the up-or-out churn without creating a workforce that ages in place and becomes too expensive for the model to support.

Some firms are experimenting at the edges. EY’s client shedding is one example. KPMG’s consolidation of its global network from over 100 entities to roughly 30 is another; it is an acknowledgement that most of the federation’s country operations could not justify their cost structures independently. Deloitte restructured its core business lines in 2024. PwC reduced headcount.

But none of these moves address the fundamental architecture. They are efficiency measures within the existing model. The pyramid remains. The billing-rate spread remains. The utilisation targets remain. The up-or-out churn remains. And as long as those elements remain, the incentive to trade quality for margin remains with them.

The Big Four generated over $220 billion in combined revenue in their 2025 financial years, employing roughly 1.5 million people. Those are not organisations that can pivot to a fundamentally different model without risking the economics that fund partner compensation. The partners vote on changes. The partners benefit from the status quo. The math is straightforward.

Where the floor might be

The industry’s own benchmarks suggest that 75% billable utilisation is the threshold below which profitability deteriorates rapidly. EBITDA across surveyed professional services firms fell from 15.4% in 2023 to 9.8% in 2024 as average utilisation dropped to 68.9%. Revenue per consultant declined to $199,000.

But the quality threshold runs in the opposite direction. Above 80% to 85% utilisation, the SPI benchmark notes that firms risk burnout, attrition, and delivery quality erosion. The PCAOB data is consistent with this: the deficiency rate escalation from 2020 to 2023 coincided with a period when firms were pushing utilisation to recover from pandemic-era disruption and meet post-pandemic demand with leaner teams. Other factors contributed too, including audit complexity, remote working, and staffing shortages. But the pyramid’s incentive structure sat underneath all of them, amplifying every pressure.

The window between “profitable enough to sustain the partnership” and “not so pressured that the work product degrades” is narrow, perhaps ten percentage points wide. And it is narrowing further as AI compresses the value of the analytical hours that junior staff bill, as clients demand fixed fees that limit the upside of high utilisation, and as the talent pipeline produces graduates with less institutional loyalty and shorter expected tenure.

We spent our careers inside this model. We benefited from it. We managed our own utilisation dashboards and our own leverage ratios, and we contributed to the pressure that cascaded down to the people doing the work. We are describing a system we participated in, not one we observed from outside.

The question we never answered, and that no partner meeting we ever attended attempted to answer, is the one we are putting to you now.

What utilisation rate is the floor? At what point does the math make quality impossible? Is there a number below which the pyramid cannot compress without the work becoming something no client should rely on? And if you are a current partner reading this: do you know what your firm’s number is? Has anyone ever tried to calculate it?

We think the PCAOB data is the beginning of that calculation. One in five Big Four audits failing to gather sufficient evidence is the model’s output, measured in black and white.

Between 2023 and 2025, the Big 4 and MBB collectively poured billions into artificial intelligence. By our count, the public commitments alone exceed $10 billion. PwC committed $1 billion across its US operations and became OpenAI’s largest enterprise customer, handing ChatGPT Enterprise licenses to over 100,000 employees. KPMG entered a multi-year alliance with Microsoft widely reported at $2 billion. Deloitte launched a $2 billion technology and AI program it branded “Industry Advantage.” EY invested $1.4 billion in its EY.ai platform, building an internal large language model called EYQ on top of Microsoft Azure and Azure OpenAI. McKinsey deployed an internal AI platform called Lilli, which the firm says is used by over 75% of its employees and has saved more than 1.5 million hours of search and synthesis work.

These are serious numbers. Billions of dollars, hundreds of thousands of licensed seats, executive pronouncements about reinvention.

And after all of it, the consulting business model has barely changed.

The pyramid is narrowing at the base, but nobody redesigned the structure. The billing model has started to crack in places, but three-quarters of the industry still charges by the hour. Partners still show up for kickoffs and final presentations. Juniors still build the slides. The only difference is that the juniors now use a chatbot to draft them faster, while being told this makes the firm “AI-native.”

We want to be specific about what those billions bought and what they did not, because the gap between the press releases and the operating reality is where the diagnosis lives.

What the money bought

Start with the internal tools. Every major firm now has one. McKinsey has Lilli, which indexes more than 100,000 internal documents and 40-plus knowledge sources. BCG has Deckster for slide editing. Bain has Sage, powered by OpenAI. PwC built ChatPwC on GPT-4 before upgrading to the full ChatGPT Enterprise rollout. EY built EYQ on Microsoft’s Azure stack.

These tools work. We know people at these firms who use them daily. McKinsey’s own case studies report roughly 30% time savings on information gathering and synthesis. PwC has reported substantial productivity gains from its generative AI tools. Senior consultants told us they can now produce a first-pass competitive analysis in two hours that would have taken three junior staff two weeks. The technology compresses what used to be the most labour-intensive part of engagement delivery.

None of that is in dispute. The technology is real, and in certain applications, it is good.

The question is what the firms did with those productivity gains. And the answer, when you look at the numbers, is: they kept the gains, cut the juniors, and left most of the billing model untouched.

The pyramid with fewer people at the bottom

In the UK, graduate-level accountancy job adverts fell about 44% year-on-year, with Big 4 firms cutting their intakes sharply. EY delayed graduate start dates for the third consecutive year, with some 2025 hires not starting until 2026.

At firm level, the cuts went deeper. According to executive search firm Patrick Morgan, KPMG reduced its UK workforce by 7% in 2024, PwC by 5%, Deloitte by 5%, and EY by 3%. The reductions continued into 2025, with KPMG and PwC each shrinking UK headcounts by another 4%.

McKinsey is the most candid about what is happening. CEO Bob Sternfels said in early 2026 that the firm now operates with roughly 25,000 AI agents alongside about 40,000 human employees, and expects the ratio to approach parity. McKinsey saved 1.5 million hours of search-and-synthesis work through AI in the past year. The firm’s headcount has dropped from roughly 45,000 to around 40,000 over 18 months, with the reductions concentrated in junior research, back-office, and non-client-facing roles.

Kate Smaje, McKinsey’s global leader of technology and AI, told reporters that the firm no longer needs “armies of business analysts creating PowerPoints” because the technology can handle much of that work.

So the base of the pyramid is being compressed. Fewer graduates, fewer analysts, fewer of the people who used to do the 60% of engagement work that could be described as “structured information processing.” The firms are producing comparable output with smaller teams.

Now ask the follow-up that none of the press releases address: if a team of eight can now deliver what a team of twelve used to, is the client paying for eight or twelve?

The billing model almost nobody wants to change

The answer varies by engagement, but the direction is clear. Engagement fees have not dropped to match the productivity gains. In many cases, they have increased, because the firms are repositioning themselves as “AI-enabled” and charging a premium for the label.

The logic goes like this: AI makes our teams more productive, which means the output per hour is higher, which means the value per hour is higher, which justifies the existing fee structure. This is a defensible argument on paper. It is also the exact argument that a taxi company would have made about GPS navigation in 2010: the technology makes our service better, so we should charge more per ride. Uber made that argument irrelevant.

Credit where it is due: McKinsey has moved further than most. The firm disclosed in late 2025 that about a quarter of its global fees now come from outcome-based pricing rather than hourly billing. That is a real shift, and it is worth watching. But McKinsey is the exception, not the pattern. Three-quarters of its own revenue is still traditional, and the Big 4 have barely begun to experiment with alternative pricing. The industry’s centre of gravity remains the billable hour.

The risk for the firms is that clients are starting to do the same math. A senior procurement officer at a FTSE 100 company told us last month that her team now asks a specific question during every consulting RFP: “What percentage of this engagement will be delivered using AI tools, and how is that reflected in the fee?” Two years ago, nobody asked that. Today, she said, it is a standard line item in the evaluation criteria.

KPMG may have accelerated this shift in the worst possible way. The Financial Times reported in February that KPMG International pushed Grant Thornton, the firm that audits KPMG’s own books, to cut its fees on the basis that AI should be making the audit faster and cheaper. KPMG threatened to switch auditors if Grant Thornton did not agree. Grant Thornton agreed. The audit fee dropped 14%, from $416,000 to $357,000.

The irony writes itself. KPMG acted as the aggressive procurement client, demanding an “AI discount” from its own service provider. Every one of KPMG’s consulting clients now has the playbook. If KPMG believes AI justifies lower fees when it is the buyer, why should the logic be any different when KPMG is the seller?

BCG’s numbers tell the story

Among the major firms, BCG is the most transparent about where revenue growth is actually coming from, and the picture should worry anyone who believes traditional strategy consulting has a stable future.

BCG reported $14.4 billion in revenue for 2025, up 7% from $13.5 billion the prior year. That sounds healthy. Then you look at the composition.

AI- and tech-focused services now represent over 40% of BCG’s total revenue. AI services specifically grew 25% year-over-year. The firm hired AI engineers, data scientists, IT architects, and industry specialists. It launched the BCG X AI Science Institute.

BCG does not break out the growth rate of its non-AI business, but the arithmetic is suggestive. When a firm grows 7% overall and its fastest-growing segment (already 40% of revenue) is compounding at 25%, the remaining business is not growing much. It may be shrinking in real terms. The traditional strategy and organizational work that BCG was built on, the practice that charges $500,000 for a four-week engagement to tell a CEO what three of their direct reports already know, is no longer the growth engine.

BCG’s own press release calls this “22 consecutive years of growth.” Technically accurate. But the firm is becoming a technology implementation business with a strategy brand bolted on, and the revenue composition makes that trajectory visible.

McKinsey’s pattern is similar, though the firm discloses less. Bain reports that AI- and tech-enabled work is already roughly 30% of its consulting business, projected to reach 50%. We are not saying this is wrong. Firms should follow revenue growth. (We spent decades telling clients exactly that.) What we are saying is that the press coverage treats these firms as if they are successfully integrating AI into a traditional consulting model when the reality is closer to the opposite: the technology business is cannibalizing the consulting business, and the firms are relabeling the result as intentional.

The credibility gap

These firms sell AI transformation to clients for fees that often start at seven figures. They produce reports on AI strategy. They advise boards on AI governance. They staff entire practices around helping organizations deploy AI responsibly. That is a significant revenue stream, and the firms market it aggressively.

During this same period, three of the four Big 4 firms have been caught publishing reports with AI-fabricated citations. (It’s only a matter of days before it’s four out of four - you read it here first).

Deloitte Australia delivered a 237-page government welfare report that contained fabricated references, including a quote from a Federal Court judgment that cited a judge whose name was misspelled. The report had been produced using Azure OpenAI GPT-4o. Deloitte refunded a portion of the A$440,000 contract.

Deloitte Canada submitted a 526-page health workforce plan to the Government of Newfoundland and Labrador, costing C$1.6 million, that contained at least four citations to academic papers that do not exist. Deloitte stood by its recommendations.

EY Canada published a cybersecurity marketing report in which GPTZero, the AI detection firm that has become the de facto auditor of the auditors, found that 16 of 27 citations were fabricated, misattributed, or pointed to dead links. Sixty percent. The report cited a “McKinsey & Company: Loyalty Economics Report (2022)” that does not exist. GPTZero traced this to what it calls a “secondhand hallucination”: a fabricated reference that originated in a blog post and was then laundered into the report as if it were a real source.

KPMG International published a report on AI itself, titled “Total Experience: Redefining Excellence in the Age of Agentic AI.” GPTZero found that 40 of 45 citations were fabricated, mangled, or misattributed. UBS, the NHS, Swiss Federal Railways, and Transport for London told the Financial Times that the report’s claims about their AI usage were untrue or misleading.

An AI report about AI, with AI-generated citations to AI projects that do not exist, published by a firm charging clients millions to implement AI responsibly. Read that sequence again and try to maintain the position that these firms have their own house in order.

The consistent response from each firm has been some variation of: “The substance and recommendations are unaffected.” The defence concedes more than it intends. The fabricated evidence does not undermine the conclusions, which is another way of saying the conclusions were never based on evidence in the first place. If the recommendations survive the removal of the sources that supposedly supported them, those sources were decorative. GPTZero coined a term for this: “vibe citing.” Generating references that feel plausible without checking whether they exist. As a description of the consulting industry’s relationship with evidence, it is uncomfortably precise.

As of mid-2026, PwC is the only Big 4 firm not publicly implicated in an AI-fabrication scandal of this kind.

Partner politics and the gravity well

Inside the firms, the people who understand the technology best are the most frustrated by what is happening with it. We have spoken to partners and directors at four different firms who describe the same dynamic, often using almost the same words.

One, a technology partner at a Big 4 firm, described it this way: the firm invested heavily in AI tools, trained thousands of people to use them, and then told the practice leaders to “integrate AI into delivery.” The practice leaders heard: make the existing model more efficient. They did not hear: rethink the model. Why would they? The existing model generates their compensation. A practice leader whose team of twelve is reduced to eight has just lost 33% of the billable headcount that feeds the partner’s revenue attribution. Nobody volunteers for that.

Another, a former BCG director, put it more bluntly: the partners who control the P&L of traditional strategy engagements have every incentive to use AI to make the old model faster and no incentive to replace the old model with something that might be better for clients but worse for the partner’s economics. The technology team proposes new delivery models. The practice leadership nods, agrees in principle, and then staffs the next engagement exactly the way they staffed the last one.

This is the pattern we observed with every attempted reform inside these firms. EY spent over $600 million on Project Everest trying to separate audit and consulting, and the US partners killed it in part because splitting the tax practice would have altered their economics. The same gravity applies to AI. Any change that threatens the current partnership structure gets slowed, diluted, or quietly abandoned once the press coverage fades.

What would genuine transformation look like?

If these firms were serious about AI changing their business model, you would see at least some of the following.

Engagement pricing would shift from hourly billing to fixed-fee or outcome-based models. McKinsey is experimenting with this; roughly 25% of its fees are now outcome-based. That is worth noting and worth watching. But for the remaining three-quarters of McKinsey’s business, and for the vast majority of Big 4 consulting, the billable hour persists. If AI compresses the hours, and the firms claim the value per hour has increased, then they should be willing to price on value rather than time. Most are not.

The leverage ratio would change visibly. If you need fewer juniors per partner, the pyramid should be narrowing at the base and broadening in the middle, with delivery redesigned around smaller, more senior teams augmented by AI. Some of this is happening through hiring cuts, but it is happening as a cost measure, not as a delivery redesign. The teams are smaller. The engagement structure is identical.

Clients would see the AI. Today, most clients on a Big 4 or MBB engagement have no visibility into which parts of their deliverable were produced by AI and which were produced by a person. There is no disclosure standard, no transparency norm. When we asked a partner at one firm why clients are not told, the answer was candid: “If the client knew how much of this was AI-generated, they would ask why they are paying partner rates.”

The training pipeline would be redesigned. The traditional model trained graduates through repetition: you learned to be a consultant by building slide decks, running analyses, and sitting in client meetings for two to four years until the pattern-matching became instinctive. If AI is now doing the analytical repetition, what replaces it as a training mechanism? Nobody has a good answer. UK accountancy graduate job adverts fell 44% in a single year, and the firms have not announced what replaces the on-the-job apprenticeship those roles used to provide.

If AI were genuinely transforming these firms, the operating model would look different: the fees, the staffing, the disclosure practices, the partnership economics. Instead, what has mainly changed is the cost base. AI made it cheaper to deliver the same product with fewer people. The savings went to partner compensation, not to clients and not to reinvention.

Clients are doing the math

The firms have bought themselves time, perhaps two to five years, by being early AI adopters in their own operations. But the clock is running. Procurement teams are getting better at distinguishing between “we use AI” as a capability claim and “we use AI” as a marketing claim. Boutique firms that were never built on the pyramid model can adopt AI without the partnership politics that choke change at the incumbents. Companies are building internal strategy teams and discovering that a senior hire with AI tools can replicate a large portion of what they used to buy from McKinsey.

BCG’s revenue composition is the leading indicator. The technology business is growing. The traditional business is stalling. The firms that figure out how to price, staff, and deliver for an AI-enabled world will survive in something resembling their current form. The firms that use AI to make the old model slightly cheaper, while cutting the junior staff who were supposed to be their future, will discover that they optimized themselves into irrelevance.

We spent our careers inside these firms. We know the partner politics, the compensation incentives, the gravitational pull of this-quarter’s-numbers on every strategic decision. And what we see, looking at billions in AI investment, is an industry that bought the technology and has been slow to change the business.

The pyramid is narrower at the base. The slides are faster. The business model, for the most part, is the same.

Has anyone here been on an engagement where AI genuinely changed the delivery model, not just sped up slide production? We are interested in examples where the staffing, the pricing, or the deliverable format was fundamentally different because of AI. If those engagements exist, we want to hear about them. If they don’t, that absence is the point.

Last updated 22 July 2026. We will update this post as new cases emerge. Bookmark it.

In August 2005, KPMG signed what was then the largest criminal tax settlement in American history: $456 million and an admission of criminal wrongdoing, under a deferred prosecution agreement, for designing and selling tax shelters that manufactured at least $11 billion in phony tax losses. The firm promised it had changed. In June 2019 the SEC fined the same firm $50 million for altering completed audit files using confidential inspection plans stolen from inside its own regulator, and for exam cheating among its audit professionals. Same firm, fourteen years apart. In April 2024 the PCAOB handed down the largest fine in its history, $25 million, to KPMG Netherlands, where more than 500 people had shared answers on internal exams and the firm misled investigators about it. And last month, KPMG Australia’s chairman told a parliamentary committee that confidential client data had crossed the firm’s “ethical divider” into a team pitching against that client’s interests.

We open with KPMG only because someone has to go first. Deloitte, PwC and EY each have an equivalent sheet, and so do McKinsey, BCG and Bain. Every incident below was described, at the time, as isolated: a rogue partner, a one-off lapse in a single country. Read one entry and the explanation is plausible. Read forty in a row and “isolated” turns into a description of the pattern itself.

This is the record, firm by firm and in date order: what happened, what it cost, what the regulators did, and one italic sentence under each entry on how it looked from inside a partnership. The facts come from regulator announcements, court records and named investigative reporting, with links; the italics are ours. Some ground rules: we name firms; we mostly do not name individuals unless they led a firm, testified publicly, or were convicted. Where a matter rests on allegations or journalism rather than an adjudicated finding, we say so. Figures and currency conversions are as reported by the primary source, and the list is long without being exhaustive.

The title says “since Arthur Andersen,” so start there. Andersen was convicted in June 2002 of obstruction of justice for shredding Enron audit documents. The Supreme Court overturned that conviction three years later on jury-instruction grounds, and by then it made no difference: clients had fled within weeks and the firm had surrendered its licences. Eighty-nine years of history ended inside twelve months. The verdict arrived too late to matter either way; the indictment alone had been fatal. The surviving firms absorbed that lesson thoroughly, which is one reason nearly every entry below ends in a settlement.

Deloitte

2018. Taylor, Bean & Whitaker. Deloitte audited mortgage lender Taylor, Bean & Whitaker from 2002 to 2008 while TBW ran a fraud with Colonial Bank built on fictitious and double-pledged mortgage loans; both companies failed in August 2009. Deloitte paid $149.5 million in 2018 to resolve potential False Claims Act liability for missing it. (HousingWire)

Nine years from collapse to settlement is long enough for every partner involved to retire on full points.

2019. Serco Geografix. The FRC fined Deloitte £6.5 million, discounted to £4.2 million for settling, over the 2011 and 2012 audits of the Serco subsidiary at the centre of the electronic-tagging overbilling affair, and added £300,000 in costs. The engagement partner was fined personally. (Consultancy.uk)

The settlement discount alone exceeded a partner’s annual profit share. Deterrence was not the design goal.

2020. Autonomy. A then-record £15 million FRC fine and a severe reprimand for “serious and serial failures” in the 2009 and 2010 audits of Autonomy, the software company HP bought for $11.1 billion in 2011 and wrote down by $8.8 billion a year later. Two former audit partners were fined, one of them excluded from the profession for five years. (FRC)

HP’s deal team still gets mocked for this acquisition. They at least had an excuse: they relied on the audits.

2021. 1MDB. Deloitte audited Malaysia’s 1MDB and its SRC International unit from 2011 to 2014, resigned in 2016, was fined 2.2 million ringgit in 2019 for failing to report irregularities in a linked bond issuance, and paid the Malaysian government $80 million in 2021 to settle all claims, the largest 1MDB settlement by any audit firm in Southeast Asia. Billions moved out of the fund through shell structures of exactly the kind audit methodology exists to catch. (Bloomberg)

Deloitte’s defence, that the fraud was deliberately concealed, describes every fraud in history.

2022. Mitie and SIG. Two smaller FRC actions in one year: £1.45 million after reduction over goodwill-impairment testing in Mitie’s FY2016 audit, and £906,250 after reduction over the 2015 and 2016 audits of SIG plc, where the FRC found breaches of requirements it called fundamental to independent audit. (FRC)

The amounts were minor; the findings were the tell. Impairment testing and revenue basics are first-year material, not judgment calls.

2021 to 2025. The exam-cheating and quality-control run. A string of PCAOB actions across the Deloitte network: $350,000 for Deloitte Canada in 2021 after staff altered the clocks on their computers to backdate work-paper sign-offs; $900,000 for Deloitte Colombia over quality-control failures tied to a 2016 Bancolombia audit; $1 million each for Deloitte Indonesia and Deloitte Philippines for widespread answer-sharing on internal training exams, with a Philippines leader barred; and $3 million for Deloitte Netherlands within a combined $8.5 million Dutch action that also caught EY and PwC. (PCAOB)

The geography matters more than the sums. Answer-sharing turned up wherever regulators looked closely, on four continents and at every one of the four firms.

2025. The Canberra report. Deloitte Australia refunded part of a A$440,000 contract after a 237-page assurance review for the Department of Employment and Workplace Relations was found to contain fabricated references, including a quote that appears nowhere in the Federal Court judgment it was attributed to, with the judge’s name misspelled. A University of Sydney researcher caught it. Deloitte then disclosed the report had been produced with help from Azure OpenAI’s GPT-4o and issued a corrected version. (Fortune)

It was, of all things, an assurance review, the one product whose entire value is that somebody checked.

2025. The Newfoundland report. Weeks later in Canada: a 526-page health-workforce plan for Newfoundland and Labrador, priced near C$1.6 million, contained at least four citations to papers that do not exist. Deloitte Canada said AI had not written the report, only “selectively” supported some citations, and stood by its recommendations. (Fortune)

If the recommendations stand regardless of whether the evidence exists, ask what the 526 pages were for.

PwC

2019. Colonial BancGroup. PwC audited Colonial BancGroup while the Taylor Bean fraud ran through it. The FDIC sued; a federal judge found PwC negligent in December 2017 and ordered $625.3 million in damages; the parties settled for $335 million in 2019. (HousingWire)

Two firms audited two ends of the same fraud for six years and neither saw it. It comes back in the pattern section below.

2023. Americanas. Brazil's largest accounting fraud, and a two-firm story. Americanas S.A. disclosed roughly R$20 billion of accounting inconsistencies in January 2023 and entered bankruptcy protection with about R$43 billion of debt. Its independent investigation found ~R$25 billion of inflated results through fictitious cooperative advertising entries and ~R$21 billion of understated financial debt through supplier finance booked as trade payables. KPMG audited from 2016 until an early termination in the third quarter of 2019, six days after its engagement partner escalated a control letter on the accounts the fraud ran through; PwC returned, its three-year cooling-off having just elapsed, and signed unmodified opinions for 2019 to 2021. The CVM opened an administrative accusation against KPMG in November 2024. PwC remains under review. No final sanction against either firm. (Reuters)

Mandatory rotation is the reform the profession points to as proof the independence problem is solved. Here it gave the client a lawful exit from the auditor asking questions and a lawful predecessor to replace it with.

2020. Luanda Leaks. The ICIJ’s investigation into Isabel dos Santos, daughter of Angola’s former president, showed PwC charging rich fees to advise on Angolan tax avoidance and on deals around state oil company Sonangol, including $21.4 million routed to the firm through a Dubai shell company for a 2017 Sonangol “modernization” project. Portuguese police raided PwC’s Lisbon offices in January 2023 at Angola’s request. These are journalistic findings, not adjudicated ones. (ICIJ)

Every firm runs a client-acceptance committee. Files like this one show how often it functions as a filing requirement rather than a brake.

2023. The Australian tax leaks. This is the one that rewired a national government’s relationship with the profession. From 2013, PwC partner Peter Collins advised Treasury on the confidential design of new multinational anti-avoidance laws, signed three confidentiality agreements, and shared the intelligence internally anyway. PwC then marketed structures to sidestep the very laws Collins was helping write, earning about $2.5 million advising 14 US companies. Internal emails released by a Senate inquiry showed dozens of partners in the loop. CEO Tom Seymour resigned in May 2023. PwC sold its entire government-consulting business, rebadged as Scyne Advisory, for one Australian dollar, accepted a $642,000 fine over 170 false claims of legal professional privilege, lost hundreds of partners and staff, and saw FY2023–24 revenue fall by A$820 million, roughly a quarter, to A$2.35 billion. (Overview)

We have both sold and staffed confidential government working groups. The wall between that work and the tax practice was a sentence in an engagement letter, and everyone senior knew a sentence was all it was.

2024. Evergrande, Beijing’s verdict. China’s Ministry of Finance and the CSRC suspended PwC Zhong Tian, the firm’s mainland audit arm, for six months and took 441 million yuan (about $62 million) in fines and confiscations, the harshest penalty a Big 4 firm has ever received in China. Regulators said PwC issued “false audit reports” on Evergrande, found 88 percent of sampled project records inconsistent with reality, and concluded the developer had inflated mainland revenue by roughly $80 billion across 2019 and 2020. More than 50 major clients left, Bank of China, Alibaba and Tencent among them, and PwC fired six partners. (Consultancy.asia)

The ban hurt more than any fine could. Fines get absorbed; fifty lost audit clients are a franchise event, and regulators everywhere took notes.

2025. Riyadh. Saudi Arabia’s Public Investment Fund barred PwC from new advisory work across the fund and its subsidiaries for a year, cutting the firm out of Vision 2030 work. No official reason was ever given; Bloomberg’s reporting tied the fallout to PwC’s attempt to hire away NEOM’s chief internal audit officer. PwC cut about 1,500 staff and 60 partners across the Middle East before the ban was lifted in late January 2026. (Bloomberg)

When a sovereign client can vaporise a regional practice with one unexplained letter, the trusted-advisor language in the pitch deck reads differently.

2024 to 2026. The fine ledger. The PCAOB fined the US firm $2.75 million and PwC Australia $600,000 in March 2024 over quality-control and reporting failures. The Dutch firm was caught in the 2025 exam-cheating action. PwC Singapore was fined $1.5 million in 2025. In the UK, fines came over Wyelands Bank and, this month, £3.2 million over the 2019 and 2020 Babcock audits, PwC’s second penalty over that client after a combined action of almost £8 million in 2023, alongside London Capital & Finance actions that included the FCA’s first-ever fine of an auditor, £15 million, for failing to report suspected fraud. (City A.M.)

Five jurisdictions in twenty-four months is the point at which “local issue” requires a new map.

2026. Evergrande, Hong Kong’s turn. Hong Kong’s AFRC fined PwC HK$300 million and imposed a six-month ban on new public-interest-entity clients, the first restriction of its kind on a PIE auditor; combined with a HK$1 billion SFC compensation fund for shareholders, the total reached HK$1.3 billion, about US$166 million, over audits that had blessed the same overstated revenue Beijing acted on, roughly 564 billion yuan. (Yahoo Finance)

The same audit failure, priced twice by two regulators, and the combined bill still rounds to two weeks of PwC’s global revenue.

EY

2020. Three clients, one year. EY affiliates were the auditors of NMC Health, which failed after revealing more than $4 billion in hidden debt; Luckin Coffee, which admitted about $314 million in fabricated sales; and Wirecard. All three imploded within months of each other. (Forbes)

Auditing is a portfolio business, and 2020 was the year EY’s tail risk correlated.

2022. The ethics exam. The SEC fined EY $100 million, at the time the largest penalty ever imposed on an audit firm, after finding that audit professionals had cheated on the ethics component of the CPA exam and on continuing-education courses, and that the firm misled the SEC’s enforcement staff during the investigation, denying any current cheating problem one day after learning of one. EY admitted the facts. (SEC)

The exam in question was the ethics exam. We would not dare invent that detail.

2023. Wirecard. EY signed unqualified opinions on Wirecard for close to a decade before the German payments company went into insolvency in June 2020, once the company itself conceded that €1.9 billion in cash, about a quarter of its balance sheet, probably never existed. Germany’s audit watchdog APAS banned EY from taking new listed-company audit clients in Germany for two years, fined the firm €500,000 plus smaller sums against five individual auditors, and investigators described audit work that was at minimum negligent and in places grossly so. EY dropped its appeal in 2024. (CNN)

Confirming cash is week-one work for a first-year. Whatever else went wrong across a decade of clean opinions, it began with the simplest test on the audit plan.

2023. Project Everest. EY spent more than a year and roughly $600 million designing a split of its audit and consulting arms, then abandoned the plan in April 2023 when the US executive committee refused to proceed, with the fight centring on who would keep the lucrative tax practice. The firm walked away carrying over $700 million in debt and no split. (AccountancyAge)

The only Big 4 firm to seriously attempt the reform regulators keep hinting at could not get it past its own partners. File that away for every future debate about voluntary separation.

2025 to 2026. NMC Health. Administrators of the failed hospital group sued EY for about £2 billion over 2012–2018 audits that missed more than $4 billion of hidden debt. The High Court trial opened in May 2025 and ran through the summer; in February 2026, with judgment still pending, EY settled on confidential terms with no admission of liability, and the administrators’ report filed in May put the payment at £105.5 million. An FRC probe that has reportedly identified serious failings continues, and EY’s UK firm set aside a record £188 million for fines and legal claims. (Bloomberg; Irish Times)

A nine-figure provision is a forecast. EY’s own finance function is telling you what it expects the next few years to look like.

2025. The Dutch exams. EY Netherlands was one of the three firms in the PCAOB’s combined $8.5 million Dutch exam-cheating action. (CFO Dive)

See Deloitte, above. Same offence, same years, different logo.

2025 to 2026. The UK fine ledger. The FRC fined EY £4.9 million over Thomas Cook and £325,000 over Stirling Water Seafield, and has open probes into unauthorised audit reports and an independence question involving Shell. (AccountancyAge)

None of these made the front page, which is a data point in itself. The market has repriced what counts as news from a Big 4 audit practice.

2026. The loyalty-fraud report. EY Canada published a cybersecurity marketing report in which the AI-detection firm GPTZero found 16 of 27 citations fabricated, misattributed or dead, including a confident reference to a McKinsey study that has never existed, itself traced back to someone else’s blog post. EY pulled the report the same day. (GPTZero)

A fake McKinsey citation inside an EY report is the industry’s supply chain in miniature.

2026. The Prime Minister’s bank account. EY dismissed two graduate employees it had seconded to Commonwealth Bank; one of them, along with a second man, was then criminally charged over allegedly accessing Prime Minister Anthony Albanese’s personal banking data. The case is before a Sydney court and no findings have been made. (The Register)

Secondment bills like consulting and behaves like employment: the firm takes the fee, the client holds the access controls, and neither side monitors conduct. This entry is what that gap looks like.

KPMG

2005. The shelters. The opening story, in full, ran to $456 million, an admission of criminal wrongdoing, and a deferred prosecution agreement over shelters with names like FLIP, OPIS, BLIPS and SOS, which produced at least $11 billion in fictitious losses and cost the US Treasury at least $2.5 billion. Nine individuals were charged, among them a former deputy chairman. (DOJ)

Remember the deferred prosecution structure. It recurs on this page like a chorus.

2017. Gupta state capture. KPMG South Africa audited Gupta family companies for about 15 years, work that included Linkway Trading, through which R30 million of public money meant for poor dairy farmers in the Free State moved via Dubai to pay for a lavish 2013 family wedding. The firm also produced the discredited rogue-unit report for the revenue service, used to force out finance minister Pravin Gordhan, and later withdrew it. In September 2017 the local CEO, chairman, COO and five senior partners resigned, and KPMG conceded its work “fell considerably short” of its standards. The audit regulator struck the lead partner off the register in 2019. (Daily Maverick)

The resignations followed the newspapers rather than any internal review, which tells you which alarm system a partnership actually wires up.

2019. The stolen inspection list. The SEC fined KPMG $50 million after the firm obtained confidential information from a PCAOB insider identifying which of its audits would be inspected, then altered completed work papers to improve its results, and after finding widespread answer-sharing on internal training exams. KPMG admitted wrongdoing. (SEC)

The remedy for a high deficiency rate was, in practice, advance knowledge of the sample. Cheaper than quality, until the SEC priced it.

2023. Carillion. KPMG signed unqualified opinions on the UK construction group for 2014 through 2016; Carillion went into liquidation in January 2018 after announcing £1.045 billion in contract provisions. The FRC fined KPMG a record £21 million, reduced from £30 million for cooperation, with its chief executive calling the work “a textbook case study in failure,” and sanctioned two former partners, one excluded from the profession for ten years. A separate 2022 tribunal had already fined the firm £14.4 million after staff forged documents to mislead FRC inspectors. (FRC)

The second fine deserves its own line: staff forged documents aimed at the inspectors who inspect the auditors. At that point the org chart of trust has run out of levels.

2024. The record fine. The PCAOB’s largest penalty ever, $25 million, went to KPMG Netherlands, where more than 500 professionals shared exam answers over five years, the practice reached the firm’s former head of assurance, who was barred for life, and the firm repeatedly misrepresented what it knew to investigators. (PCAOB)

Five hundred participants over five years leaves no room for the word “isolated.” At that scale the behaviour has a training programme.

2025. Form AP. The PCAOB censured and fined nine KPMG network firms a combined $3.375 million for failing to disclose, on the Form AP filing that names every firm participating in an audit, which other firms worked on theirs; smaller UK fines over Carr’s Group and N Brown, and an earlier $500,000 PCAOB fine for KPMG Japan, sit alongside it. (PCAOB)

Who actually did the work is the entire premise of an audit signature. The sums are small; the subject is not.

2026. The AI-ethics exam. KPMG Australia confirmed in February that 28 staff had used AI tools to cheat on the firm’s mandatory AI-ethics training since mid-2025, among them a partner-level registered auditor fined more than A$10,000. This is the same national firm the PCAOB sanctioned $450,000 in 2021 over cheating that implicated as many as 1,100 partners and staff, and the same global network that announced itself, in 2024, as the first organisation in the world to earn ISO 42001 certification for AI governance, a service it also sells. (Report)

We read this entry aloud to three former colleagues. All three laughed before they winced.

2026. The audit-data leak. The fire is still burning on this one. KPMG staff moved confidential information from audit clients Lendlease and Optus to colleagues pitching for the audits of Westpac, Dexus and Telstra. Chairman Martin Sheppard told Parliament on 19 June that unredacted Optus data crossed what the firm called its ethical divider and reached the team pursuing rival Telstra. A whistleblower had first raised concerns on 30 May 2024; the firm’s documented response was to search his laptop, deny him a pay rise, pull his client work and threaten dismissal. Since March: CEO Andrew Yates resigned (an A$1.7 million notice payment plus A$2.4 million in retirement entitlements), the head of audit left, the COO stepped aside, the firm appointed its first independent chair, Lendlease ended an audit relationship of roughly 68 years, ASIC opened a formal investigation, the Department of Finance froze KPMG out of new federal contracts until the end of September, the Greens referred the firm to the National Anti-Corruption Commission, and the AFR reports preparations to cut as many as a thousand jobs and up to a fifth of partner pay. A second parliamentary hearing is set for 14 August. (Bloomberg; Canberra Times)

Every firm’s first crisis instinct is containment; we have sat in the rooms where that instinct wins. What distinguishes this case is that the containment file became evidence.

2026. The Total Experience report. KPMG International published an AI thought-leadership report in October 2025; by the time GPTZero finished with it in June, 40 of 45 citations proved fabricated, mangled or misattributed, and UBS, the NHS, Swiss Federal Railways and Transport for London told the Financial Times that its claims about their AI usage were untrue or misleading. KPMG removed the report. (GPTZero; The Register)

Four institutions had to publicly correct a Big 4 firm’s account of their own operations. The report was marketing, which is the problem: the product being marketed is accuracy.

McKinsey

2018. The Saudi influencer report. A 2016 McKinsey document analysed Twitter sentiment after Saudi austerity measures and identified three people driving negative online conversation about the government. All three were subsequently targeted: writer Khalid al-Alkami was arrested, a second account-holder disappeared, and Omar Abdulaziz, a close associate of Jamal Khashoggi, had his phone hacked and relatives detained, with torture reported. McKinsey said the report was for internal use. During the crown prince’s rise the firm was embedded enough in Riyadh that the Planning Ministry earned the nickname “Ministry of McKinsey.” (Courthouse News)

An internal-use defence concedes the report existed and named the names. Everything after that concession is a dispute about distribution.

2018. Eskom and Trillian. McKinsey took roughly R1 billion from South Africa’s state power utility on a turnaround contract, run alongside a Gupta-linked partner firm, that was later found unlawfully procured. The firm repaid the R1 billion (about $74 million) with interest in 2018 and a further $39 million in 2020 over other state-owned-enterprise work. (SA Government News)

Repayment with interest is the professional-services equivalent of a guilty plea entered in a whisper.

2019. ICE. ProPublica and the New York Times revealed that McKinsey’s engagement with ICE, begun under one administration as organisational work, was redirected under the next toward executing the immigration crackdown, with the firm proposing cuts to spending on detainee food, medical care and supervision. Total billings passed $20 million, and the team ghostwrote the contracting document that justified its own $2.2 million extension. (ProPublica)

Every consultant has drafted the client memo that requests more consulting. Few have done it for a detention system’s food budget.

2020. Luanda Leaks. McKinsey’s slice of the dos Santos-era Sonangol work was $15.4 million, routed through the same Dubai shell company as PwC’s and BCG’s fees. As with PwC’s entry above, these findings come from journalism, and no court has ruled on them. (ICIJ)

Three rival firms shared one Dubai invoicing address, which says more about client selection than any values page.

2021 and 2024. Purdue. McKinsey advised Purdue Pharma for over a decade on how to “turbocharge” OxyContin sales, work that included targeting the highest-volume prescribers and a 2017 proposal to pay distributors rebates tied to overdoses linked to pills they moved. When states began suing Purdue, senior McKinsey people emailed about deleting documents. The firm paid $573 million to 47 states, DC and five territories in 2021 without admitting wrongdoing, then $650 million in December 2024 to resolve a Department of Justice criminal investigation, entering a five-year deferred prosecution agreement while a former senior partner pleaded guilty to obstruction of justice for destroying records. In April 2026 the firm agreed to contribute a further $125 million to Purdue’s bankruptcy estate, again without admitting wrongdoing, taking its opioid-related settlements past $1.7 billion in total. (FiercePharma; Reuters)

Strip the industry framing away and describe the engagement plainly: paid experts optimised sales of a product at the centre of a mass-casualty epidemic, and the firm’s recorded reflex when accountability approached was the delete key. No partnership-governance reform addresses this. It is a question of what work you will take.

2022. Paris. A French Senate report alleged that a McKinsey executive gave false sworn testimony about the firm’s tax payments in France, prompting a fiscal investigation that remains in the allegation column. (The Hill)

Tax advice is the product; the firm’s own tax posture is the free sample everyone inspects.

Ongoing. Bankruptcy disclosures. McKinsey’s restructuring arm has faced repeated allegations that it failed to disclose connections to interested parties in US corporate bankruptcy cases, where disclosure by advisers is a legal requirement; in 2019 the firm paid $15 million to the US Trustee to resolve disclosure questions across three cases.

Disclosure rules exist because the court cannot see the conflicts on its own. A firm that treats them as negotiable has told you its theory of the referee.

2024. The China file. The House Select Committee on the Chinese Communist Party alleged that McKinsey concealed extensive work for Chinese government bodies, including on five-year plans and Made in China 2025, while holding more than $480 million in US Department of Defense contracts, and that the firm’s global managing partner misrepresented that work under oath in February 2024. These are the committee’s allegations. (House Select Committee)

A client-acceptance policy meets exactly one test that matters at this scale, and simultaneous service to both sides of a strategic rivalry is that test.

2024. The South Africa DPA. Closing the Eskom-era loop, and separate from the 2018 repayments: McKinsey Africa agreed to a $122.85 million criminal penalty under a three-year deferred prosecution agreement over bribery of Transnet and Eskom officials between 2012 and 2016, conduct that had earned the firm about $85 million. A former partner had pleaded guilty in 2022, and a parallel South African resolution added roughly $50 million. (DOJ)

Count the deferred prosecution agreements on this page. For firms whose product is judgment, the DPA has become a recurring line item, and line items get budgeted.

BCG

2020. Luanda Leaks. BCG’s slice of the dos Santos-era work was the largest of the three consulting firms: $31.2 million through the Dubai shell company for the 2017 Sonangol project, plus a stint helping run the jewellery house De Grisogono, acquired with Angolan state loans. Portuguese police raided the Lisbon office in January 2023 as part of the dos Santos investigation. As with the PwC and McKinsey entries, these are journalistic findings. (Consultancy.com.au)

De Grisogono deserves its own business-school case: a strategy firm helping operate a jewellery house bought with public money.

2024. The DOJ declination. A separate Angola matter with a cleaner paper trail: between 2011 and 2017, BCG’s Lisbon office paid about $4.3 million in commissions to an agent connected to Angolan officials, who helped the firm win eleven contracts with the economy ministry and one with the central bank, worth about $22.5 million in revenue. BCG self-disclosed after finding a 2014 email, fired those involved, and in August 2024 the US Department of Justice declined prosecution in exchange for $14.4 million in disgorged profits. (DOJ letter)

A declination is the best available outcome in a foreign-bribery case, and BCG earned it by the book: disclose, cooperate, fire, disgorge. Eleven ministry contracts through one connected agent still describes a business-development channel, whatever the engagement letters called it.

2024 to 2025. Gaza. Between October 2024 and May 2025, BCG helped establish and run the US and Israeli-backed Gaza Humanitarian Foundation and, under a workstream reported as Project Aurora, built financial models that included costing the “voluntary relocation” of Palestinians out of Gaza, a concept a senior UN official condemned as ethnic cleansing. BCG described the work as pro bono; the Financial Times, which led the reporting, put the contracted value above US$4 million and billings above $1 million a month. The CEO apologised, two partners were fired for what the firm called unauthorised work, the chief risk officer and the head of social impact stepped down, and Save the Children and the World Food Programme reviewed or paused their ties to the firm. (Middle East Eye)

Two fired partners is the rogue-actor defence at a firm whose entire method is that nothing reaches a client unreviewed. Pick which claim to believe about BCG; both cannot be true.

Bain

2022. SARS. Bain’s 2015–2017 restructuring of the South African Revenue Service under commissioner Tom Moyane gutted the agency’s enforcement capacity. Two judicial commissions found that Bain’s local managing partner had met President Jacob Zuma repeatedly and that the firm’s work facilitated state capture. South Africa’s National Treasury banned Bain from public contracts for ten years, until 2032, citing “corrupt and fraudulent practices”; the UK Cabinet Office had imposed a three-year ban of its own in 2022, then lifted it for the global firm in March 2023 while keeping the South African arm excluded. Bain repaid its fees with interest, apologised, wound down its client-facing South African business, and continues to deny wilfully enabling state capture. Whistleblower Athol Williams left the country after testifying. (TimesLIVE)

A ten-year sovereign ban is the closest thing consulting has to a licence revocation, and the industry’s response was to keep pitching as if it had happened to someone else’s profession.

The AI report problem

Three of the four Big 4 firms have now been caught publishing reports with fabricated, AI-hallucinated citations: Deloitte twice, in Australia and Canada, then EY Canada, then KPMG International, with fabrication running from a handful of fake references to 40 out of 45. PwC is, so far, the only Big 4 firm with a clean sheet on this, and no MBB firm has yet been caught. Most of the forensic work has come from GPTZero, which coined the term “vibe citing” for hallucinated references and traced fake citations laundering from one document into another. Each firm caught reached for the same defence, that the substance and recommendations were unaffected. That defence deserves the scrutiny the citations never got: if the conclusions hold regardless of whether their evidence exists, the evidence was decoration, and the client paid for decoration.

The regulators kept score

Behind the case-by-case entries sits a quieter dataset. The PCAOB’s Part I.A deficiency rate measures the share of inspected audits where the file did not contain sufficient evidence to support the opinion issued. For the US Big 4, that rate ran 12 percent in 2020, 16 in 2021, 26 across 2022 and 2023, and back to 20 in 2024. Across all inspected firms it hit 40 percent in 2022, peaked at 46 percent in 2023, and eased to 39 in 2024; BDO USA touched 86 percent. Enforcement moved in parallel: PCAOB civil penalties went from about $1.1 million in 2021 to a record $11 million in 2022 and $11.9 million by November 2023, and since 2021 the board has sanctioned at least nine firms for exam cheating. One number puts the escalation in scale: the single fine against KPMG Netherlands in April 2024 was more than double the board’s entire record 2022 total. (Accounting Today; Floyd Advisory)

Australia has gone furthest. After two identical wall failures at two firms in three years, new Big 4 federal contracts fell from A$637 million to A$348 million between 2024 and 2025, and this month Canberra moved to expand ASIC’s powers, publishing a Treasury options paper that describes conduct from the large firms “that is not fair and honest.” ASIC has widened its audit-conduct scrutiny to all four.

The 2024 improvement in the PCAOB numbers is real, and the firms will cite it. Weigh it honestly: the improvement arrived under the most intense regulatory attention the profession has faced since 2002, and it still leaves a fifth of inspected Big 4 audits without sufficient evidence for the opinions they carried. In what other product category would a 20 percent verified defect rate be presented as recovery?

Three patterns

Sort the forty-odd entries above and they resolve into three piles.

The first pile holds audit failures on companies that then failed: Wirecard and EY, Carillion and KPMG, Evergrande and PwC, NMC and EY, Americanas and PwC and KPMG, 1MDB and Deloitte, Autonomy and Deloitte, and Colonial BancGroup, where PwC and Deloitte audited two ends of the same fraud. The economics of this pile are stable and comfortable. The fine lands years later, sized in the single or low double-digit millions, against client losses in the billions; the partners involved have usually retired. What changes firm behaviour is a different instrument, and the record shows it plainly: Germany’s two-year ban moved EY more than any fine, China’s six-month suspension cost PwC over fifty clients, and Hong Kong followed with a PIE restriction. Exclusion has changed behaviour where money never has.

Pile two collects the conflicts between advisory work and a public-interest role: Canberra’s tax intelligence inside PwC, audit-client data inside KPMG Australia’s pursuit teams, the three-firm Dubai invoicing address in Luanda, McKinsey advising the FDA while advising Purdue, the Saudi and SARS engagements. The walls that were supposed to prevent all of this were real in exactly one sense: they existed as sentences in policy documents, acknowledged during onboarding. Between us, the contributors to this publication sat through decades of those acknowledgements, and we watched information cross the walls whenever it made a pitch stronger. Two identical wall failures in one country within three years should end the argument about whether this is a people problem or an incentive problem.

The third pile deserves the most attention, because it involves no client, no fraudulent counterparty and no tempting pile of money: exam cheating at every Big 4 firm across the US, the Netherlands, Australia, Indonesia and the Philippines; stolen regulator inspection plans; documents forged for FRC inspectors; investigators misled at the SEC and the PCAOB; AI-fabricated evidence in published reports; and an AI-ethics exam cheated on with AI. The first two piles have external villains available. The third is the firms alone with their own integrity infrastructure, and the infrastructure lost. We have sat through the all-hands calls that follow each of these announcements, and a couple of us have helped draft the talking points that client-facing teams receive the same afternoon. The training module that follows is always about forty minutes. The quiz at the end can be passed on a skim. That is the soil the third pile grows in.

The strongest defence, taken seriously

The defence of the firms goes like this, and it deserves its best version. These are organisations with revenue between $53 billion and $70 billion each, operating in well over a hundred countries, across a window of more than twenty years; at that scale some misconduct is a statistical certainty, and a list like this one is survivorship bias in reverse. Most of the work, most of the time, is competent; clients keep buying it, and the US Big 4 still audit about 80 percent of the market capitalisation of US-listed companies because no credible alternative exists at that scale. The PCAOB’s own 2024 numbers show correction. Some entries above are journalism or committee allegations rather than verdicts. And Andersen’s fate proves the ultimate sanction is available when conduct warrants it.

Much of that is true, and we say so as people who spent careers inside these buildings delivering work we remain proud of. The defence explains the frequency of failure. It cannot explain the kind. A base-rate argument works for a retailer or a manufacturer, businesses that sell products and sometimes ship bad ones. These firms sell trust: the audit franchise exists because the state grants a handful of private partnerships the role of certifying everyone else’s honesty. For that business, three features of the record above are disqualifying. Repeat offences follow settlements as if the settlements were subscriptions; KPMG’s arc from 2005 to 2026 is the clearest run, and network-wide exam cheating continued for years after EY’s $100 million made the stakes unmistakable. Penalties price in as a cost of doing business; against Deloitte’s $70.5 billion in annual revenue, an eight-figure fine is a rounding decision. And a striking share of the misconduct targets the oversight system itself: stolen inspection selections, forged documents for inspectors, misled enforcement staff, allegedly false testimony before two legislatures. Ordinary corporate scandal cheats customers or markets. Several entries on this page cheat the referee, and a profession that cheats its referee forfeits the case for self-regulation, which is the case the entire model rests on.

Now tell us what’s missing

This page will grow. We will add entries as regulators, courts and journalists produce them, and we will correct anything a reader shows us we got wrong; the caveats above about allegations and currency conversions stand.

What’s missing from this list? If you’ve seen something from the inside that hasn’t been reported, this is the place. Two requests before you type: stay within whatever confidentiality obligations bind you, and bring specifics that can be checked. Patterns and public documents beat client names. The comments are open.

In 2023, PwC partner Peter Collins was found to have shared confidential Australian government tax intelligence with dozens of partners and staff across the firm. In 2026, KPMG Australia’s chairman admitted that confidential Optus information had crossed the firm’s “ethical divider” into a team pitching for rival Telstra’s audit. In both cases, confidential information crossed internal walls that were supposed to be impenetrable, because the financial incentive to share it was stronger than the compliance incentive to contain it. At KPMG, the firm’s first instinct was to investigate the whistleblower rather than the conduct he reported: his laptop was searched, he was denied a pay rise, had client work removed, and was threatened with dismissal. And at both firms, the “one bad apple” defence collapsed under the weight of evidence showing that the information sharing was widespread.

Two firms, same country, same failure mechanism. Three years apart.

We are supposed to believe this is a governance failure. That better policies, stronger internal walls, more robust (there’s a word the firms love) compliance training will prevent it from happening again. We spent a collective 170 years inside these institutions. We sat in the meetings where information that should have stayed behind the wall was shared casually, because the partner sharing it was trying to win a pitch and the information made the pitch stronger. We signed the compliance attestations and watched them being ignored. The wall is not weak. The wall is fictional. And the fiction persists because it is profitable.

The Big 4 should be broken up. Audit and consulting should be separated, fully and permanently, at every firm, in every jurisdiction. The argument for maintaining the combined model has been exposed as intellectually bankrupt by the firms’ own conduct. We are not going to hedge that position, because the evidence does not support hedging.

The incentive problem that no policy can fix

The audit-consulting conflict is simple to describe and impossible to resolve within a combined firm. An auditor’s job is to provide an independent opinion on whether a company’s financial statements are accurate. Independence requires that the auditor has no financial interest in the client’s commercial success beyond the audit fee. When the same firm that audits a company also advises that company on strategy, technology, tax structuring, and operational improvement, the auditor has a direct financial interest in keeping the client happy. The advisory fees dwarf the audit fees. A partner who irritates the client with tough audit findings jeopardises the consulting relationship that pays for the partner’s bonus, the practice group’s headcount, and the firm’s growth targets.

This is not a hypothetical risk. It is the operating reality of every combined Big 4 firm, and it has been for two decades.

In 2023, the Big 4 generated $66.5 billion from audit and assurance services. They generated $95.4 billion from advisory and consulting. Advisory revenue exceeded audit revenue by 43%. At Deloitte, consulting alone generated more revenue than the entire audit practice of any other Big 4 firm. The economics are lopsided enough that audit has become, in effect, a loss leader: the mandatory, regulated service that gets the firm through the door, where the fees multiply through advisory work sold to the same client.

We have watched this dynamic play out from the inside of partner compensation meetings. The audit partners who bring in the highest-value clients are the ones whose clients also buy consulting services. The partner whose audit client generates $2 million in audit fees and $8 million in advisory fees is a more valuable partner than the one whose client generates $3 million in audit fees and nothing else. Everyone in the room understands this. Nobody says it aloud, because saying it would make the conflict of interest explicit. So the compensation formula rewards “client relationship management” and “cross-service integration” and other euphemisms that mean the same thing: the audit partner who feeds the consulting pipeline gets paid more.

No compliance policy can survive that incentive structure. The firms have tried. They have created ethical walls, Chinese walls, information barriers, ethical dividers (KPMG Australia’s preferred term for the one that failed). They have appointed chief ethics officers and built compliance training modules and required annual attestations. All of it amounts to a policy document that asks individual partners to act against their own financial interest, against the firm’s cultural expectations, and against the implicit rewards that determine their compensation. When the policy conflicts with the incentive, the incentive wins. Every time. At every firm. The only variable is whether someone reports it.

The evidence the firms produced against themselves

Arthur Andersen earned more from consulting fees to Enron than from auditing Enron’s books. When the fraud was exposed in 2001, the firm’s Houston office shredded documents. Andersen was convicted of obstruction of justice in 2002; the conviction was later overturned by the Supreme Court, but by then the firm had already surrendered its licences and ceased to exist. The surviving firms began shedding their consulting practices even before Congress passed Sarbanes-Oxley in mid-2002. Ernst & Young’s consulting practice went to Capgemini in 2000. KPMG Consulting was spun off in 2001 and eventually became BearingPoint. PwC’s consulting arm was sold to IBM in 2002. Deloitte was the only firm that did not complete a full divestiture of its consulting business. Sarbanes-Oxley then reinforced the separation by restricting the non-audit services that audit firms could sell to audit clients.

Within a decade, every firm that had divested its consulting practice had rebuilt one. By 2015, advisory revenue had surpassed pre-Andersen levels. The Sarbanes-Oxley restrictions applied only to services sold to audit clients, and the firms worked around them by expanding advisory services to non-audit clients, building the consulting revenue base until it once again dominated the firm’s economics. The cycle completed itself in less than fifteen years.

Since then, the evidence has accumulated to the point where denial requires active effort.

PwC Australia: a partner advising the government on confidential tax legislation simultaneously shared that intelligence with PwC colleagues who used it to help multinational clients circumvent the laws PwC had helped write. Internal documents show the firm generated roughly $2.5 million advising 14 US companies on how to avoid the very rules its partner had shaped. When the Senate inquiry released internal emails, they showed that dozens of partners across the firm were aware of the information sharing. PwC eventually sold its government consulting division for A$1 and lost hundreds of partners and staff. Its FY2023-24 revenue dropped by A$820 million, with profits falling 24%.

KPMG Australia: staff used confidential Lendlease board papers and Optus data to pitch for audit contracts at Westpac, Dexus, and Telstra. The whistleblower who reported it had his laptop searched, was denied a pay rise, had client work removed, and was threatened with dismissal. KPMG has since been sidelined from new federal government contracts, its CEO resigned (reportedly collecting A$1.7 million in notice pay plus A$2.4 million in retirement entitlements), Lendlease ended a 68-year audit relationship, and the firm is preparing to cut over 1,000 jobs. Partner pay may drop by 20%.

EY and Wirecard: EY audited Wirecard for close to a decade, issuing unqualified opinions while €1.9 billion in cash balances probably never existed. The audit team failed to independently confirm the bank balances over multiple engagement cycles. Germany’s audit regulator APAS banned EY from taking on new public-interest-entity audits for two years.

PwC and Evergrande: Chinese regulators found that 88% of PwC’s project records were inconsistent with reality. PwC Zhong Tian was suspended for six months and fined the equivalent of $62 million. PwC lost more than 50 major clients, including Bank of China, Alibaba, and Tencent. Hong Kong regulators added a further $166 million in combined fines and compensation.

Every one of these failures occurred within a combined audit-consulting firm. Every one involved auditors whose independence was compromised, implicitly or explicitly, by the firm’s commercial interests. The pattern is consistent across firms, across countries, and across decades.

EY tried to solve it. EY proved it cannot be solved voluntarily.

In 2022, EY launched Project Everest, the most ambitious attempt to separate audit and consulting from within. The plan would have split EY into two independent entities: an audit-focused firm and a consulting-focused firm. It was, on paper, exactly the organisational reform that regulators and critics had been demanding.

It failed. The US executive committee refused to proceed, primarily because the partners could not agree on how to divide the tax practice (which serves both audit and consulting clients) and because the personal economics of the split would have reduced the income of the most senior US partners. The failure consumed roughly $600 million in costs and left EY carrying over $700 million in debt. The firm began cutting costs and laying off staff to recover. In the UK alone, EY has since set aside a record £188 million for fines and legal claims.

Project Everest is the single most important data point in this debate. The firm that tried hardest to resolve the conflict voluntarily proved that partnership economics make voluntary separation impossible. The partners who benefited most from the combined model had veto power over the split, and they used it. The conflict of interest that the split was designed to eliminate was also the force that killed it. If the firms cannot do this to themselves, someone else has to do it to them.

The counterargument, honestly engaged

The strongest case for maintaining the combined model is that separation would weaken audit quality rather than improve it. Auditors need access to specialists in technology, valuation, tax, and sector-specific expertise. A standalone audit firm, stripped of its consulting arm, would lose that bench of specialists. Smaller audit firms already have higher PCAOB deficiency rates than the Big 4 (the all-firm aggregate hit 46% in 2023, while the Big 4 US rate was 26%), in part because they lack the specialist resources that large, diversified firms can deploy.

This is a serious argument, and it deserves a serious answer.

The answer is that the audit arm does not need to be profitable on its own. It needs to be independent. Right now, it is neither. Sarbanes-Oxley permits audit firms to engage non-audit specialists for audit-related work. A separated audit firm can hire actuaries, IT specialists, and valuation experts as contractors or employees. The UK’s FRC operational separation framework, which all four firms completed implementing by June 2024, already allows audit practices to borrow expertise from other service lines within the same firm. The question is whether the audit partner’s compensation, career trajectory, and daily incentives are tied to the consulting revenue pipeline. In a separated firm, they are not. In a combined firm, they always will be.

PCAOB inspection data reinforces the case for separation. The Big 4 US deficiency rate jumped from 12% in 2020 to 26% by 2022 and 2023. It has improved slightly to 20% in 2024, but the direction over the past five years is clear: being large and combined did not prevent a deterioration in audit quality. It may have contributed to it, because the same institutional pressure to prioritise client relationships and revenue growth that drives the conflict-of-interest scandals also drives the corner-cutting that PCAOB inspectors flag.

The direction of travel

The UK moved first. The FRC’s 22 principles of operational separation, completed by all four firms in 2024, require audit practices to produce separate profit-and-loss accounts, operate with independent audit boards, and demonstrate that audit is not cross-subsidised by consulting. The FRC stopped short of a complete breakup, but the trajectory is unmistakable. The Competition and Markets Authority recommended operational separation, with full breakup as the backstop if it proved insufficient.

Australia is moving faster, propelled by two consecutive scandals. ASIC has opened a formal investigation into KPMG and widened scrutiny to audit-conduct complaints across all Big 4 firms. The Australian Treasury published a 56-page paper in July 2026 criticising behaviour from accounting, auditing, and consulting firms that, in the Treasury’s words, is “not fair and honest.” The government has announced it will strengthen ASIC’s powers over the Big 4, which are currently structured as partnerships outside ASIC’s direct supervision. New federal government contracts with the Big 4 dropped from A$637 million to A$348 million in a single year.

In the US, the Senate has held hearings on audit quality and Big 4 conflicts of interest. The PCAOB under Chair Erica Williams escalated enforcement sharply, with civil penalties reaching $11.9 million by November 2023 alone (up from $1.1 million in 2021). Williams has called the profession’s deficiency rates “unacceptable.” Whether the US will follow the UK and Australia toward mandatory separation depends on political will, but the regulatory consensus has shifted. A decade ago, operational separation was considered radical. Today, it is the moderate position. Full separation is the question being debated.

What we learned from the inside

We managed practices within these firms. We sat on the committees that reviewed audit quality. We participated in the partner compensation discussions where the cross-selling incentives were baked into the reward structure. We watched colleagues share information they should not have shared, win pitches they should not have won, and retain clients they should not have retained, because the consulting revenue attached to those clients made them untouchable.

We do not say this with satisfaction. Several of us built our careers in practices that benefited from the combined model. The cross-selling worked in both directions: audit relationships opened doors for consulting engagements, and consulting relationships generated audit referrals. The model was good for partner economics. It was bad for audit independence, and audit independence is the thing that justifies giving these firms the franchise to sign off on the financial statements that public markets rely on.

The firms will argue that separation is unnecessary, that internal reforms are working, that the compliance architecture has been strengthened, that the recent scandals are isolated incidents being addressed. They have been making this argument for 24 years, since Arthur Andersen. The compliance architecture has been strengthened repeatedly, and the same failures keep occurring, because the compliance architecture is addressing a symptom while the incentive structure is producing the disease.

We have 170 combined years inside these institutions. We have watched the cycle repeat three times: scandal, reform promise, policy adjustment, quiet reversion to the status quo. The only intervention that broke the cycle was the one that was never tried: separation.

It should be tried now. Voluntarily if the firms can summon the courage that EY’s partners lacked. By regulators if they cannot.

If you were designing the system from scratch, would you ever allow the same firm to audit a company and advise it? If not, why do we tolerate it now?

Julian McPherson, until recently KPMG Australia’s head of audit, appeared before a parliamentary committee on 19 June and admitted that when a whistleblower alleged misuse of confidential client information across practice boundaries, the firm treated it as an employment matter. Not an ethics issue. Not a compliance breach. An HR problem. Senator Deborah O’Neill’s response from the chair cut through the procedural language: “You looked at this and you went, oh, this guy’s a problem.”

That single admission tells you more about how these firms operate than every press release, governance reform, and leadership transition combined. It is also the one detail that most of the coverage has missed.

The headlines have covered the facts well enough. KPMG staff shared confidential Lendlease board papers internally while pursuing audit mandates at Westpac, and unredacted Optus data crossed what the firm calls an “ethical divider” into a team pitching for Telstra’s audit. Chairman Martin Sheppard admitted to the committee that the divider had failed. CEO Andrew Yates resigned in May with A$1.7 million in notice pay plus A$2.4 million in retirement benefits. The AFR reports the firm is preparing job cuts that could exceed a thousand; partner pay reductions of up to 20 per cent have been discussed. ASIC has opened a formal investigation. KPMG agreed with the Department of Finance to pause bidding for new federal contracts for three months, though existing contracts (about 40 per cent of which carry extension options) were excluded. Senator Barbara Pocock called the arrangement “a holiday, not a punishment.”

All of this has been reported. What has been underreported is the pattern, because recognising the pattern means asking a question that nobody in a position of authority over these firms wants to answer.

Three years apart, same mechanism

This is the second Big 4 meltdown in Australia in three years. In 2023, PwC’s tax-leak scandal revealed that partner Peter Collins had shared confidential government tax-policy intelligence with dozens of advisory colleagues (internal correspondence released by the Tax Practitioners Board referenced more than 50 PwC partners and staff) while serving as a government advisor on anti-avoidance legislation. PwC used that intelligence to advise multinational clients on structuring around the laws Collins helped design. CEO Tom Seymour resigned. Internal emails showed the information had been distributed far beyond Collins. PwC sold its government-consulting arm for A$1. Profits fell roughly 25 per cent.

Different firm. Different data. Identical mechanism: confidential information crossed an internal wall because the financial incentive to share it was stronger than the compliance incentive to contain it. At KPMG, the firm’s initial response was to investigate the whistleblower rather than the conduct. At PwC, the firm maintained its “one bad apple” defence until Senate-released emails showed the scale of internal distribution. In both cases, the containment instinct preceded the accountability instinct.

If two different firms, in the same country, three years apart, produce the same failure through the same mechanism, you are not looking at two incidents. You are looking at a design flaw.

What the firm probably looked like from the inside

Anyone who has been in a partnership during a crisis will recognise the choreography. It follows a sequence so consistent that you could set a timer by it.

The whistleblower emailed McPherson in May 2024, alleging that partners pursued “revenue growth at all costs.” Instead of investigating the allegations, KPMG authorised a search of the whistleblower’s laptop. According to testimony before the parliamentary committee, he was denied a pay rise, had client work withdrawn, and was threatened with dismissal. Subsequent covert searches of his laptop later that year reportedly uncovered material that supported the original allegations. By that point, the firm’s containment strategy had been running for months.

I have watched this sequence play out from inside a partnership. The first instinct is never investigation. It is triage: who knows, how bad is this, and can it be contained within the practice group? Partners closest to the issue begin calculating personal exposure. The conversations happen in side channels, over coffee, in the margins of other meetings. The compliance team is brought in eventually, but by the time they arrive, the senior partners have already mapped the blast radius and begun positioning.

The “ethical divider” that Sheppard admitted had failed was never a wall in any meaningful sense. In every firm I worked at, it was a policy document acknowledged during annual compliance training and then set aside, because the economic logic ran in the opposite direction. Audit partners sat on information that would help their consulting colleagues win work. Consulting partners needed competitive intelligence to pitch against rival firms’ clients. The divider existed to satisfy regulators. The incentive structure existed to generate revenue. When those two forces collided, I never saw compliance win.

The partners who shared the information did not think of themselves as breaching confidentiality. They thought of themselves as helping the firm. The incentive system told them that was their job. The compliance system told them to sign a form once a year confirming they understood the rules. The gap between those two signals is where every one of these scandals originates.

The fallout is accelerating

The business consequences are no longer hypothetical. New federal contracts awarded to all four Big 4 firms in Australia dropped from A$637 million in 2024 to A$348 million in 2025, according to Reuters analysis of government tenders. Lendlease dropped KPMG as its auditor after 68 years. KPMG’s public-sector division has paused bidding for Commonwealth and NSW work until the end of September.

In early July, the Treasury released a 56-page options paper canvassing reforms that include separating audit from consulting, capping the number of partners (from over 1,000 to 400, mirroring limits on law firms), mandatory firm rotation, and licensing audit firms under ASIC the way financial services businesses are licensed. On 16 July, the government formally directed ASIC to increase its oversight of all four firms. Assistant Treasurer Daniel Mulino said the government had observed “behaviour from large accounting, auditing, and consulting firms in Australia that is not fair and honest.”

That language is worth pausing on. A government minister, on the record, describing the conduct of the Big 4 as “not fair and honest.” Three years ago, after PwC, the language was about “disappointing lapses.” The register has changed.

Why Australia, and why twice

Australia has shown an unusual willingness to drag these firms into public view. Parliamentary committees held open hearings. Senators named names. Regulators moved quickly enough that the firms could not run the clock. The US and the UK have strong enforcement agencies, but neither has matched Australia’s combination of speed, public scrutiny, and political pressure over the past three years.

The implication is uncomfortable: if the wall between audit and consulting fails twice in three years in the jurisdiction where partners are most likely to be publicly held to account, the question is not what went wrong at these two firms. The question is whether the wall holds anywhere, or whether it only appears to hold where nobody is looking.

I spent years in firms where the divider was treated as a compliance formality. Information moved through informal channels: a conversation before a practice-group meeting, a forwarded email with the subject line slightly changed, a mention over lunch framed as a hypothetical. The partners sharing it were not acting out of malice. They were acting within a system that rewarded revenue growth and penalised nothing else with equivalent force. The difference between KPMG Australia and the firms I worked at is not that the wall failed. In my experience, the wall was never as solid as the compliance manual described. The difference is that a whistleblower put it in writing, and a regulator followed through.

The question the Treasury paper cannot answer

Australia’s options paper goes further than any current proposal outside the UK (which already requires operational separation). Mandatory structural separation. Licensing. Partner caps. Rotation. If implemented in full, they would remake the industry in that country.

But the proposals rest on an assumption that better rules and stronger enforcement can fix a problem created by the economic architecture of these firms. The wall between audit and consulting fails because the two sides of the firm share a profit pool. Partners on both sides are paid from the same pot. Strengthening the wall while leaving the shared economics intact is like installing a better lock on a door that nobody has any incentive to close. The PwC scandal produced a parliamentary inquiry and dozens of recommendations. Most of those recommendations remain unimplemented three years later. KPMG’s scandal has now produced another parliamentary inquiry and another set of proposals. The firms will engage constructively with the consultation process. They always do.

The open question is whether “engaging constructively” and “changing anything” are the same activity. The track record suggests they are not.

How many firms have you worked at where the internal walls between audit and advisory were genuinely enforced? Not the policy. Not the annual training module. The day-to-day reality of how information moved between practice groups. We are asking because we already know the answer for the firms we worked at, and we suspect yours was the same.

Every partner writes an exit memo. Some firms call it a transition document, some a handover note. At one firm we know, they call it the “legacy brief,” with no apparent irony. Whatever the name, the ritual is the same: before you leave, you write down what the incoming partner needs to know. Client relationships. Engagement history. Where the risks sit. What the numbers actually look like behind the dashboard the executive committee presents at town halls.

We have written a lot of exit memos. Between us, we have left firms 14 times at partner level. Some of us made partner at one firm, left, and made partner again at another. One contributor held senior partner roles at two Big 4 firms and an MBB. We have written exit memos that ran to forty pages and exit memos that fit on a single sheet. We have written memos where the honest version bore no resemblance to the version we submitted.

This publication is the honest version.

The exit memos we filed were constrained by diplomacy, by ongoing commercial relationships, by the knowledge that the people reading them would be deciding our deferred compensation. So we described client risks without naming the partners who created them. We flagged engagement economics without pointing out that the margins depended on staffing models we considered reckless. We noted “areas for development” in the practice without saying plainly that the quality had declined because nobody with authority was willing to pay the cost of maintaining it.

Those memos were useful. They were also incomplete. This publication exists because the incomplete version, multiplied across thousands of departing partners over two decades, has left the outside world with a dangerously inaccurate picture of what these firms have become.

Who we are

We are former partners and senior partners of Big 4 and MBB firms. Ninety years at partner level between us. A hundred and seventy years inside the machine in total, counting from our first days as associates and analysts. Some of us were direct-admit partners. We are not going to tell you our names.

That is not false modesty or cowardice. Most of us still work in the professional services orbit as advisors, board members, investors, and clients of the firms we left. Attaching names would make this publication about us rather than about the argument. It would also end several professional relationships that, selfishly, we prefer to keep.

So instead of credentials by name, here are credentials by experience. Between us, we have sat on partner compensation committees and watched the sausage get made. (It is worse than you think, and the opacity is deliberate.) We have run engagement quality reviews that flagged serious deficiencies, then watched those reviews get filed without action because acting would have meant writing down revenue. We have hired thousands of graduates and watched the training pipeline that used to produce excellent professionals get hollowed out by cost reduction. We have signed off on deliverables we were proud of and signed off on deliverables that made us wonder who we had become.

We know how partner election cycles create incentives that have nothing to do with client service. We know what happens in the first executive committee meeting after a scandal breaks. (The first concern is always the insurance position.) We know the economics of the leverage model in granular, uncomfortable detail: what utilization rate a partner needs to hit bonus, how many hours of a $1,200-per-day engagement are performed by someone billing for the first time, what happens to quality when the ratio of experienced professionals to juniors passes the point where mentoring becomes impossible.

We know all of this because we lived it. Some of us contributed to the problems we are about to describe. We are not writing from moral high ground. We are writing from a position of specific, hard-earned knowledge.

What we remember

There was a time when the work was worth something. We need to say this up front, because the criticism that follows will be more useful if you understand it comes from people who once believed in these institutions.

The training at its best was extraordinary. In our first years, we learned to think about problems with a rigour that no MBA programme could match, because the problems were real, the clients were paying, and the consequences of getting it wrong were immediate. A good partner in the 1990s was a genuine expert. Not just a salesperson with a Rolodex for cold calls, but someone who had built relationships over years with people who took their call because they had earned it by being consistently useful. The partner knew their clients’ industries, sometimes better than the clients did. The client was paying for that person’s judgment, their willingness to deliver a difficult message and stand behind it.

That kind of partner still exists, scattered across the firms. But the model that produced them has been systematically dismantled, and what replaced it produces something else entirely.

What changed

Around 2005, the buyers started purchasing the brand rather than the partner. Procurement teams centralised buying. RFPs replaced relationships. Boards discovered that “we hired Deloitte” or “McKinsey recommended it” provided career insurance for decision-makers regardless of whether the advice was any good. Once the buyer is purchasing the logo, the logo no longer needs to employ the kinds of people who built its reputation in the first place. It needs people who can sell, manage the pyramid, keep utilization high, and navigate the internal politics well enough to survive the next partner election. Deep expertise became optional. The client was paying for the name on the cover page.

The quality erosion followed. A typical Big 4 engagement today bills around 60% of its hours through professionals with fewer than two years of experience. The client pays rates that presume senior judgment. The partner shows up for the kickoff meeting and the final presentation. Everything between is a production line. We have all received the client feedback surveys that say “we expected more senior involvement.” We have all sat in the post-engagement reviews where that feedback was acknowledged and nothing changed, because changing it would have meant the partner being present, and the partner was already spread across five engagements simultaneously, because that is what the utilization model requires.

The ethical failures are a symptom of the same economics. KPMG’s $456 million criminal settlement for fraudulent tax shelters. McKinsey’s $650 million settlement over the opioid crisis. EY missing €1.9 billion that never existed at Wirecard. PwC leaking confidential government tax policy to corporate clients in Australia. Three of four Big 4 firms caught publishing reports with AI-fabricated citations in the past eighteen months alone. This is not a series of isolated incidents. The incentive system rewards growth and punishes caution. When the person deciding whether to accept an engagement is the same person whose compensation depends on revenue, the answer is almost always yes. When the compliance function reports to the partners it is supposed to constrain, the constraints loosen over time. Inevitably.

We watched this happen from the inside. We participated in some of it. When we raised concerns, the response was usually a version of “trust the process” or “this is how the profession works.” For a long time, the profession did work, in the narrow sense that the firms kept growing and the partners kept getting paid. What stopped working was the quality. And then the scandals started arriving faster than the firms could absorb them.

What we believe

The collapse of the current model is overdue. We do not mean that the firms will disappear tomorrow, or that nobody inside them is doing good work. We mean that the model itself has been running on inherited credibility for over a decade. The combination of extreme leverage, audit-consulting conflicts, partnership governance designed for consensus rather than decision, and a market reputation that diverges further from performance every year is not sustainable. The credibility account is drawing down.

We would rather see these firms reborn than buried. We spent our careers inside them. The institutional capabilities they possess, the global networks, the accumulated knowledge across thousands of engagements, the ability to mobilise hundreds of professionals across borders on short notice, those capabilities are worth preserving. What is not worth preserving is the way they are currently deployed: the cost-cutting that guts quality, the partner politics that block adaptation, the relentless optimisation of the leverage ratio at the expense of the people doing the work and the clients paying for it.

The choice is between reinvention and decline. Reinvention would mean confronting questions that no current managing partner has the tenure security to ask. Whether audit and consulting should share a brand. Whether partner compensation should be transparent. Whether the leverage model can survive AI-driven productivity gains without prices collapsing. Whether the firms can retain their best people when boutiques and PE-backed platforms offer better economics and fewer committee meetings. Decline means continuing to do what these firms have done for the past fifteen years: cut costs, issue press releases about innovation, win on brand recognition, and hope that nobody notices the gap between the reputation and the work.

We are betting on decline, because we have sat in the rooms where reinvention proposals get discussed, and we have watched them die. They die because the partnership governance model gives veto power to the people who benefit most from the status quo. EY spent $600 million trying to split its audit and consulting businesses. The partners voted it down because the partners who would lose from separation outnumbered the partners who would gain from it, and the governance rules require supermajorities. That dynamic will play out again and again on every consequential question these firms face.

What we are not

We are not bitter ex-employees with a grievance. When something works, we will say so. We have a post coming on what the Big 4 actually do well, because the lazy version of this argument, the one that says everything inside is broken and everyone inside is complicit, is wrong and counterproductive.

We are not a competitor. We are not selling an alternative service, running a boutique firm, or using this as a recruitment funnel. We are not anti-consulting. We believe in the value of expert advisory when it is delivered with genuine expertise, honest pricing, and accountability for outcomes. We are critical of how that advisory is delivered at scale by firms that have confused brand recognition with capability.

We are not going to be a one-note publication, either. We will take positions and defend them, but we will also publish pieces that argue the other side, because the best thing this publication can become is a place where people who know these firms from the inside can have the conversation that LinkedIn will never host. LinkedIn is where partners perform enthusiasm. Every city is amazing. Every initiative is transformative. Every team is incredible. The partners’ silence about what is actually happening inside these firms is not agreement. It is compliance with social media policies that require positive messaging and punish candour.

We are the people who have left. We can say what they cannot.

Why now

Because AI is about to accelerate every dynamic we have described. The leverage model sells human hours of analysis. AI compresses the value of those hours. The firms know this. They have collectively invested over $10 billion in AI initiatives since 2023. They are also selling AI transformation engagements that, in our experience, amount to a strategic assessment recommending the client hire specialists for implementation. That is a seven-figure slide deck with a chatbot demo attached.

Because the scandals are arriving faster than the firms can process them. KPMG Australia’s audit-consulting wall breached for the second time in a country where PwC’s confidential tax leak was still fresh in regulatory memory. McKinsey facing a congressional inquiry over whether its global managing partner misrepresented the firm’s work for the Chinese military while holding $480 million in US defence contracts. BCG’s CEO apologising after the firm modelled “voluntary relocation” plans for Palestinians. Four separate Big 4 firms caught publishing reports with citations that did not exist, generated by the AI tools these same firms are selling to clients as the future of professional services.

Because the people who know the most are the people who can say the least. Current partners cannot write what we are writing. We are under no such constraint.

One last thing. We are not jaded. Jaded people stop caring. We are writing because we care enough to say what current partners cannot say on the record.

For those who have been inside these firms: when did you first notice the gap between what was promised and what was delivered? Was there a specific engagement, a specific meeting, a specific number on a utilization report? We want to hear it. The comments are open.