The Restorer’s Trap:
Why Capable Leaders Choose Managed Decline

IA most capable man, doing entirely the wrong thing

Somewhere at cruising altitude, where the air is paid for and the wine list is handed to you on heavy paper, a man named Arun Varma is writing code.

This is not, in the ordinary course of things, what chief executive officers do. Chief executive officers set direction. They align stakeholders. They say things about transformation that someone else wrote. They ought not sit in business class with a laptop open to a terminal window, building a pipeline that connects three data sources to a prioritised portfolio view — because they happened to be on a distribution list where seventy account managers across all of EMEA spent four days failing to do what Arun is now doing in the time it takes to reach cruising altitude.

He knows he shouldn’t. A CEO who builds the sales intelligence his own organisation cannot produce is telling somebody three levels below him two things simultaneously, neither of them kind: I see that you cannot do this, and — rather worse — I have nothing more important to do than prove it.

He sends the email anyway. It contains the analysis, a brief note, and no explicit accusation, which is of course the most damning of all. The account manager will read it, understand approximately a third of it, and respond with the particular enthusiasm of a man who has just been handed a gift he cannot use in a language he does not speak. The deal will not close, the marketing strategy will not be successful. Not because the analysis is wrong — the analysis is, in fact, beautiful — but because the system that receives it was not designed to act on beauty. It was designed to execute commands. He does not follow up. Following up would mean asking whether the thing he sent had been understood, and asking whether one has been understood is an admission that one might not have been, and that conversation would remind him of things he prefers not to be reminded of.

Arun lands. He takes a car to the regional headquarters — regional being the word the firm uses for the office where the decisions are made, as opposed to the office where the CEO sits, which is always elsewhere. The distance is not merely geographical.

The boardroom is arranged in the way boardrooms are when the people who arranged it want everyone to know that certain chairs were occupied long before the current occupant arrived, and will be occupied long after he leaves. Arun takes his seat — or rather a seat. The deck is already loaded. It contains the word transformation eleven times, the word AI fourteen times, and the word revenue in a context that, on close inspection, means the opposite of what it appears to mean. In the back row, the two-and-a-half strategy consultants the firm has never quite managed to absorb — the ones everyone calls the slide builders — are sitting with the particular composure of people who have decided that observation is more entertaining than participation.

He does not shout. He used to. He doesn’t anymore. He was raised in a house where voices were never raised, not because anger was absent but because raising one’s voice was an admission that one had lost control of the room, and losing control of the room was — in the vocabulary of the women who raised him — simply not done. His grandmother’s favourite adage was that she considered working for money vulgar — a view that formed part of a consistent worldview that made sense, given her position. But Arun’s position is not hers, and with it he lacks the aloofness that made her composure effortless. He needed counselling to stop shouting at people. Not all the rules he learned still apply.

IIThe distance is compressing

The business model of the global IT services industry rests on a single structural assumption: that the distance between what a client wants and what a client has is long, expensive, and requires large numbers of human beings to cross. Requirements are gathered. Architectures are drawn. Code is written, tested, deployed, maintained, and — when the requirements were wrong, which they usually were — rewritten. Every stage is a billable event. The five largest Indian IT firms alone employ 1.56 million people to perform this crossing.

The distance is compressing. And the evidence is no longer anecdotal.

In the two years through mid-2025, TCS, Infosys, Wipro, HCLTech, and Tech Mahindra collectively reduced headcount by more than 42,000 employees. TCS alone shed roughly 23,500 in FY26, then posted its first year-on-year decline in dollar revenue. Bench strength declined by approximately 25% across the industry. Entry-level hiring fell 30 to 35 percent. Demand for mid-level delivery roles dropped 20 to 30 percent. And yet revenue per employee rose across every major firm. For the first time in the industry’s history, revenue growth has decoupled from headcount growth. The firms are producing more output with fewer people, which is, if one pauses to consider the implications, the beginning of a confession that the people were never the product. The crossing was.

The stock market has noticed, even if not all board decks have. Accenture’s share price fell more than 20 percent in calendar 2025 and continued declining into 2026, underperforming the S&P 500 by roughly 45 percentage points over twelve months. DXC Technology is in its eighth consecutive year of revenue decline. The firms selling AI transformation are being punished by capital markets while the firms building AI are being rewarded — a divergence that tells you everything you need to know about where investors believe the value is migrating.

The strategy firms are not immune, though they might have hoped the altitude would protect them. McKinsey’s workforce has fallen from over 45,000 to roughly 40,000 — with 5,000 positions cut in 2025 alone. Revenue growth has stalled. The firm that spent a century advising other companies on transformation now requires, by its own chairman’s admission, a rather significant one of its own. BCG’s revenue rose 10 percent in 2024, but largely by absorbing clients unnerved by McKinsey’s legal difficulties — not by solving the structural problem. All three firms have deployed internal AI assistants — McKinsey’s is called Lilli, Bain’s is called Sage, and BCG uses one called Deckster to build slide presentations, which is either very efficient or very sad depending on one’s view of what strategy consulting was supposed to be. Harvard Business Review described the resulting organisational shape as a shift from pyramid to obelisk: fewer people at every level. This is a polite way of saying that the business model was a labour arbitrage, and the arbitrage is closing.

The analyst calls are instructive, if one enjoys dramatic irony. TCS booked $165 million in restructuring charges across FY26 while cutting roughly 23,500 positions — even as it claimed $1.8 billion in annualised AI revenue. Infosys’s chairman, Nandan Nilekani, described the transition as requiring “root-and-branch surgery.” The company guided revenue growth at zero to three percent. AI projects constitute 5.5 percent of Infosys’s revenue. The other 94.5 percent is the old model, undisturbed. Accenture reported $5.9 billion in generative AI bookings for fiscal 2025, then announced in December that it would stop breaking out AI metrics entirely, because AI was now “embedded in some way across nearly everything we do.” The translation is less elegant than the announcement: the company can no longer distinguish AI revenue from regular consulting revenue, which is less a sign of integration than of relabelling. Jefferies put it most plainly: AI adoption could produce two to three percent annual deflation in traditional IT services revenues. The firms’ own tools are cannibalising their own revenue base.

Underneath every one of these projects, if one is willing to look, is the same structure that has been underneath every project for thirty years: people, billing hours, sitting in chairs. One has merely given the chairs a more contemporary name. And the deal pipelines confirm it. Wins are concentrated on cost reduction and vendor consolidation, not new digital projects. Clients are buying cheaper, not buying different.

The pattern is older than software. Assembly machine language programmers discovered it first: when the distance between intent and execution compresses by an order of magnitude, the guild that profited from the crossing becomes irrelevant. Assembly did not disappear; it became a specialty — lucrative, arcane, and approximately as relevant to the mainstream technology economy as Latin is to modern literature. Not something upon which one builds a global services enterprise. What is happening now is not the same arithmetic compression the industry has survived before — moving from assembler to C, from C to Java, from Java to frameworks. Those shifts moved the programmer from one language to another. This one moves the client from one side of the translation to the other. The rate card is inverting: clients refuse juniors not because they have found an alternative but because the work juniors did is no longer work. What remains requires seniority and judgment, which clients are delighted to purchase — at junior rates, because the supply of experienced consultants who need employment has become rather generous.

IIIThe Restorer’s Trap

When a business model begins to die, it is rarely because the people inside it did something wrong. It is because something outside it changed. Technology and innovation are not internal decisions — they are imposed externally, by markets, by physics, by the pace at which someone else’s invention makes your process redundant. The only thing within an organisation’s control is its response: ignore the shift, adopt it superficially, or use existing assets to build something new. Most firms discover three options, each of which they describe as a strategy, a word that here means a direction chosen under the impression that choosing is the same as moving.

The first door is defense. One protects the existing model through cost reduction. Fewer people, lower rates, same service. This is operational hospice care administered with great professionalism. Although, on reflection, and having met Arun’s European Head of HR, perhaps without.

The second door is the hybrid — or, less charitably, cargo culting. Copilots for developers. Automation layers on top of existing processes like fresh paint on a structure with subsidence. It is the response of a person who, upon being told the building is on fire, produces a thoughtful plan to renovate the kitchen. Arun’s firm has its own version: a strategic partnership with Anthropic, no less — which is rather like owning a Steinway and using it to hold the post. Twenty percent of his workforce are now Certified Claude Architects. They have the certificates. They have the badges for their email signatures. What they do not have is the habit of opening a terminal and doing the thing the certificate says they can do, which is why Arun finds himself doing it at thirty-five thousand feet.

The third door is the one that would work. Radical reconstitution. Rebuild from the substrate up. Accept that the firm which emerges will be smaller, faster, and unrecognisable to anyone who loved the old one.

Almost nobody chooses the third door. One does not wish to call this cowardice, because cowardice implies a choice, and what is happening in most boardrooms is something more interesting than a choice. It is a disposition. A way of being in the room that makes the third door invisible not because it cannot be seen but because seeing it would require becoming a different kind of person, and one is rather attached to the kind of person one is. One has spent thirty years becoming this person. And the wardrobe depends on it.

Arun’s grandmother would have understood the disposition perfectly, having perfected it herself. When the family estate was lost — in the kind of economic and societal upheaval that rearranges fortunes once or twice a century — she did not fight or lament. She opened a bottle of champagne and observed that the light from the burning house was really quite a show. It was the most dignified response imaginable. But it taught the generations that followed a dangerous lesson: to accept what only appears inevitable. Not to fight for a different outcome while there is still time.

Arun’s cost programme is the champagne.

Three questions determine whether a leadership team is caught in what we might call the Restorer’s Trap: trying to get everything back to where it was, despite crushing evidence of futility.

Is the CEO’s identity separable from the operating model?

Can the leader describe who they are without describing how the firm currently works? Arun cannot. His career, his restoration, his sense of what he has built — all of it is the firm as it currently operates. An attack on the model is experienced as a personal attack, which is why every proposal to change the model is received not as strategy but as insult.

Was the leadership team built by the current leader, or inherited?

Arun governs downward and negotiates upward. Below him: managers he chose for loyalty rather than for excellence, who produce results that are always, in the most dangerous sense of the word, acceptable. Above and around him: a board which considers him a parvenu and an executive team built by the man before him, who gave him the estate but not permission to renovate it. The distance between tolerance and authority is the distance that defines his career.

Or — the crueller possibility — the permission is there for the asking, and he cannot ask. Asking would require admitting that the decline began on his watch, which would require admitting that the decline is, in some sense, his. The model did not die under him; it began dying under him, and the distinction matters to exactly one person. He cannot propose a renovation because a renovation is a confession. He must instead propose a restoration, because a restoration is a promise: that the estate will be put right, that the career will arc upward rather than down when read end to end. Every day he delays asking is a day the narrative holds.

Can the organisation survive honesty?

Not individual honesty — institutional honesty. When someone says the difficult thing in a meeting, does the response follow a specific and predictable sequence: the thing is heard, the thing is acknowledged, the speaker is thanked for their candour, and nothing changes? This is more demoralising than being ignored. Being ignored at least preserves the fantasy that one might, if heard, make a difference. Being heard and producing no effect is a proof of something one would rather not have proved. And the agreeable senior managers — the ones Arun promoted — don’t even understand what is happening, let alone possess the ability to do anything about it. They lack both the skillset and the disposition, the absence of which is precisely why they were promoted to their positions in the first place.

If the answer to all three questions points inward, the organisation is in the Restorer’s Trap. These questions can be asked by a CEO of themselves or by a board about their CEO, with the minor complication that the person inside the trap is generally the last to recognise it, because recognising it would require being the kind of person who is not in it. If you find the questions easy to answer, you may wish to consider the possibility that you are answering them incorrectly. A trapped organisation will choose the first door and call it the second, or the second and call it the third, and the champagne will be excellent, and the light will be, one must admit, rather a show.

And if the answer to all three questions points outward — does the structure provide space for new business models to form? Suppose someone in the firm can see that the compression of the delivery model could benefit both the firm and its clients — a joint venture, perhaps, or an outcome-based engagement where the efficiency gain is shared rather than hidden. They could propose it. They do not. Not because the idea is bad, and not because they would be silenced, but because they are sold by the hour, and a person sold by the hour is a billing unit, not a strategic actor. The only institutional unit institutionally empowered to discuss new commercial structures is the M&A department, and the M&A department is not thinking about this problem, because it is not a problem that arrives in the shape of an acquisition deal, so the M&A department is not trained to recognise it. Honesty without opportunity is a diagnosis with no operating theatre. The organisation must take a view one step removed from what it is currently doing, challenge itself, and provide the substrate in which alternatives can take root — or the first three answers, however encouraging, are academic.

Here is the confusion at the heart of Arun’s paralysis: he is managing a professional estate while restoring a private one, and he has mistaken the two for the same activity. They are not. Managing an estate — even a burning one — means speaking with the insurrectionists, coming to terms, developing the land once the fields have been cleared. It is unsentimental, forward-looking, occasionally ugly work. Restoring an estate means preserving what was lost, which requires believing that what was lost still has value or is achievable, which requires not looking too closely at the fields. Arun applies the psychology of restoration to a problem that requires the psychology of management, because restoration is the project that gives his life its shape, and management is merely what he does for a living.

He could make the case for the third door. He has the data. The CFO has the data. He does not make this case, because making it would require telling the board they were wrong, which would require a kind of fight that would remind him of exactly where he stands in this firm, and he prefers not to be reminded. One opens the champagne instead. One remarks on the light.

The frameworks the industry uses to evaluate these doors — the transformation playbooks, the operating model canvases, the maturity matrices — were built for a world in which the distance between intent and execution was long and expensive. One does not wish to be unkind about other people’s life’s work, but it is worth observing that a resistance army is of limited utility after the war is over. The frameworks are not wrong. They are obsolete, which is more painful, because one cannot argue with obsolescence. One can only outlive it or be outlived by it.

IVWhat the survivors look like

The canonical example of a firm that survived structural compression is not an information technology company. It is the Minnesota Mining and Manufacturing Company — 3M — which began its life in 1902 digging corundum out of the ground and nearly died when the corundum turned out to be worthless. What happened next is instructive: the company did not attempt to become a better mining operation. It asked what it actually knew how to do — work with abrasives, adhesives, materials — and rebuilt itself around the capability rather than the product. Sandpaper became adhesive tape. Adhesive tape became Post-it Notes. Post-it Notes became medical devices. The identity was never we are a mining company. It was we solve problems with materials science. When any single product line died, the company did not die with it, because the company was never the product line. It is too early to say what might become of the IT services companies that choose the third door. What will become of the others, however, is abundantly clear.

This is the builder’s psychology applied at institutional scale. And it illuminates, by contrast, why the Restorer’s Trap is so lethal. A firm whose identity is we cross the distance between intent and execution using large numbers of people cannot survive that distance closing. It was never the crosser. It was the distance. A firm whose identity is we help clients create what they need can survive anything, because the method is disposable and the mission is not.

The firms that navigate structural compression share three observable characteristics.

First, they are led by builders or by effective managers — not by restorers. Builders create something new when the model breaks. Effective managers navigate the transition honestly, even when it is ugly — they speak with the insurrectionists, develop the land, make unsentimental decisions. Restorers try to return to a state that no longer exists, which is not management but its corruption. The CEO’s identity determines which of the three they become. But even a builder or an effective manager cannot walk through the third door alone, because the act of dismantling what the board promoted them to protect feels like betrayal — and nobody is promoted for their willingness to betray. The firms that survive are the ones whose boards understood this and acted as the mirror: telling the CEO, explicitly, that what they were chosen to protect is now what they are being asked to dismantle, and that this is not ingratitude but mandate. Only the board can release a CEO from the obligation that traps them. Without that release, even the right leader will optimise rather than reconstitute, because optimisation feels like leadership and dismantlement feels like treachery. This is easier for boards that chose their CEO for what comes next. It is nearly impossible for boards that chose their CEO for what came before.

Second, they changed what they sell. The proof of transformation is not in the prototype or the demo. It is in the contract. A firm that has genuinely adapted to the compression will have shifted from billing hours to billing outcomes — because if you are still selling a quantity of human effort by the hour, you are still selling the old model, regardless of what those humans are doing during the hours. Outcome-based pricing means the firm absorbs the efficiency gain rather than billing it, and no board of a public company has cheerfully voted to compress its own margins for the benefit of a client who hasn’t asked. The firms that survive are the ones whose leadership forced that vote anyway, because they understood that a margin built on a structural assumption that is no longer true is not a margin. It is a countdown.

Third, they hired for clarity, not loyalty. The slide builders — the strategy consultants Arun’s firm mocked for charging a fortune to think rather than to do — turn out to have been making a prediction, not an excuse. They always said that code was a delivery mechanism, not a destination. Now code writes itself, and the people who spent their careers on the thinking side find themselves building working systems while their engineering colleagues attend certification programmes. The asset was never the code. It was always the judgment that preceded the code. Firms that hired for that judgment — even at the cost of employing people who are occasionally, unbearably, right — are the ones with a future. Firms that hired for compliance have a cost programme.

VThe light is really quite a show

It is November. Arun is home — the real home. His wife has arranged the house for Diwali in the way she has arranged it every year for thirty years, which is to say beautifully and with the quiet competence of a person who has long since stopped expecting her husband to notice the details and has made peace with this, or something adjacent to it.

His daughter is visiting. She is twenty-six. She works in venture capital in San Francisco, where she funds companies that do, in various ways, the thing that is making her father’s firm unnecessary. She thinks of her father as a man who runs a large and successful company, because that is what he is, in the way that a man standing on a cliff is, technically, at altitude.

They are on the terrace. It is late. She has been telling him about a portfolio company — a team of nine that has built a platform that takes a client’s business requirements and produces a working, tested, deployed enterprise application. The entire development lifecycle. The thing that Arun’s firm does with forty people over six months, this company does with three people and a system in four weeks. She is excited about it. She does not notice that her father has become very still.

She asks — directly and without particular concern for the feelings of the person being asked — why his firm isn’t doing that. Not the platform. Not the technology. The pivot. Why isn’t the firm rebuilding from the ground up, the way — and here she uses a phrase that Arun finds physically painful — the way any rational actor would.

He considers his answer. He considers the board, who gave him a building he is not permitted to redesign. He considers his managers — his loyal, capable, fatally agreeable people. He considers the prototype that nobody showed him but that he knows about, because he is not a fool and has never been a fool, which is the part of the story that hurts the most. He considers the script he wrote on the plane. The one that worked. The one he sent. The one that changed nothing.

He does not tell her the truth. Instead he tells her about the cost programme. He explains that transformation is a journey, that the firm is investing in capability, that the rate card pressure is cyclical. He uses the word selectively twice. He uses the word foundation three times. He does not use the word fear at all.

There is a pause. The candles flicker. Somewhere in the house his wife is closing the kitchen, performing the small domestic rituals that will outlast every strategy deck and every substrate shift and every stock price, because they are built on a foundation that does not require quarterly validation.

His daughter asks, more softly now, not as a venture capitalist but as a daughter: But you can see all of this. Why aren’t you changing it?

Arun looks at her for a long time.

Some things are not improved by fighting them. Your great-grandmother taught me that.

She smiles, because it sounds like wisdom, like tradition, like the kind of thing fathers say to daughters on festival evenings when the candles are burning and the world outside can wait until morning.

He smiles back. His smile being, in its way, a masterpiece of the form.