The 54-Centimetre Gap
McKinsey's April 2026 diagnosis is unusually honest. Operations functions are less prepared than they were in 2022. The prescription is the part worth talking about.
Nedeljko Madzarac, the Swiss supervising site manager, walked onto the half-built Hochrheinbrücke on the afternoon of Christmas Eve 2003. Madzarac was not alone. The team had just poured the concrete for the second stage of the bridge and they were out on the structure for the last scheduled inspection before the holidays. Madzarac looked across at the German half of the bridge, coming toward them from the opposite bank of the Rhine. The two halves were supposed to meet in the middle.
They did not. The Swiss side was fifty-four centimetres lower than the German side.
Madzarac’s team discussed the situation among themselves. There was not much to discuss. Two stages had already been concreted. The gap between the two halves was physical, visible, and not going anywhere. Madzarac picked up the phone and called the Swiss engineering office.
“We had already concreted two stages when it became apparent that something wasn’t right with the height,” Madzarac later told the Berner Rundschau3.
Two zeros
The Hochrheinbrücke was being built jointly by the two Laufenburgs, the Swiss one on the south bank of the High Rhine and the German one on the north. Each country’s engineers would build their own half. The two would meet in the middle. The project was a twelve-million-franc prestressed concrete road bridge designed to take heavy traffic out of the medieval town centres on both banks. It was the kind of joint infrastructure that was supposed to be routine.
The error began in a place nobody usually looks: the definition of sea level.
Germany and Switzerland do not use the same zero. Germany uses the Amsterdam Ordnance Datum, copied from the Dutch, referenced to the North Sea. Switzerland uses the Marseille tidal gauge, referenced to the Mediterranean, with its primary calibration marker fixed to a rock in Lake Geneva called the Repère Pierre du Niton. The Swiss chose this system in the early 1800s, when Geneva was a French-speaking scientific hub and aligning with France was the obvious thing to do. The Mediterranean is an evaporative basin with weak tides. It sits on average twenty-seven centimetres lower than the North Sea. The Swiss zero and the German zero are twenty-seven centimetres apart, and that gap is geography, not error4.
The engineers on the Laufenburg project knew this. The offset was drawn up correctly on paper. Beat von Arx, the project manager for the Swiss canton of Aargau’s Civil Works department, was very clear about this afterwards. “The difference of 27 cm was certainly known, and everything had been drawn up correctly on paper.”3
During implementation, something happened that nobody has been able to explain. Instead of adding twenty-seven centimetres to the Swiss side of the design to compensate, the operation was inverted. Twenty-seven centimetres was subtracted. The correction did not cancel the geography. It doubled it. The gap the engineers thought they had closed by a full twenty-seven centimetres had instead been opened by another twenty-seven. Fifty-four centimetres of empty air where the bridge was supposed to meet itself.
The sign error passed through an endless chain of engineers, checkers, and approvals. Nobody caught it. It survived until the concrete was in the ground and a man standing on the bridge looked across and saw what the drawings had not.
Awareness is not protection
The most important thing about the Laufenburg bridge is not that engineers made a mistake. Engineers make mistakes every day. The important thing is the specific shape of the mistake.
The people who built the Hochrheinbrücke were not ignorant. They had full awareness of the problem. They had drawn up the correct offset. They knew which datum each country used and why. They had accounted for the difference in every document that mattered. The awareness was complete.
What they lacked was the step between awareness and action. The step that checks the sign. The step that says: the correction was supposed to make the gap smaller. Did it? Nobody ran that check. They applied the offset, assumed it had been applied in the right direction, and poured the concrete.
Awareness made them confident. Confidence made them build. Building made them pour concrete on a bridge that was not going to meet itself. The awareness got used to justify the action, and the action went unchecked, because the verification step was the one thing that nobody had designed into the process.
This is not a story about sea levels. It is a story about what happens when awareness is allowed to substitute for verification.
What McKinsey almost wrote
On the third of April 2026, McKinsey published one of the most honest diagnostic pieces I have read from them in years. The article, by Julian Fischer and Vera Trautwein, is called “From disruption to advantage: A leadership moment.”1 Its central claim is unusual for a consultancy to make to its own client base: operations functions in 2026 are less prepared for disruption than they were in 2022.
The data backs the claim. Tier 2 supplier risk awareness peaked four years ago and has fallen every year since. The weakest score in McKinsey’s own supply chain survey is for supply visibility over the next five years. Digitisation budgets are getting raided to fund ERP replacement projects at exactly the moment that resilience depends on digital capability. The three contextual myths the piece dispels, that transparency has improved since the pandemic, that digitisation is accelerating, that geopolitical disruption is driving global retreat, are all false. Operations functions are less ready than they were. The environment has become more volatile. The gap between the two has widened every year.
The diagnosis is the most uncomfortable thing McKinsey has said to its own client base in some time. It is also correct.
And it is describing, in the vocabulary of operations, exactly the thing that happened on the bridge.
Operations leaders are not ignorant. Awareness of Tier 1 supplier risk is over ninety percent. Visibility tools have been bought. Control towers have been stood up. The dashboards are lit. Executives have been briefed. In exactly the way the engineers at Laufenburg knew about the twenty-seven-centimetre offset and had drawn it up correctly on paper, operations leaders know about the risks in their supply chains. They can describe them at length. They have the slide decks.
What they do not have is the verification step. The thing that checks whether awareness became action, whether the action went in the right direction, whether the direction was sustained under pressure. McKinsey is measuring the gap between what operations leaders know and what operations leaders can do about what they know. That gap has been widening every year since 2022. McKinsey describes it. McKinsey does not name it.
The prescription has destinations, not a path
McKinsey’s prescription, to their credit, is specific. They argue operations leaders need to refocus on three transitions.
- From awareness to foresight.
- From fixed capacity to flex capacity.
- From low-cost sourcing to low-exposure sourcing.
Each of these is a correct destination. A forward-looking operations function should be getting to all three.
The piece does not say what bridge gets you there.
The closest the article comes to naming a mechanism is in one of its client examples. A global industrial equipment manufacturer facing volatile regional demand implemented a networkwide control tower integrating plant capacity, labour availability, and supplier constraints. The result was real: service levels improved by ten to fifteen percent, idle capacity fell by more than twenty percent, and all of it was achieved without additional capital expenditure. McKinsey uses this as evidence that the “from fixed to flex capacity” transition is possible.
The result was not produced by the integration. It was produced by what the company built on top of the integration. The control tower gave them visibility. The visibility was not the outcome. What turned the visibility into a ten-point service level improvement was a decision architecture that took a signal and converted it, inside the window where the conversion still mattered, into a structured response. The integration was the datum. The thing on top was the sign check.
McKinsey describes the integration. McKinsey does not describe the thing on top. Without the thing on top, integration produces what Laufenburg produced: confident action in the wrong direction.
What they almost named
I have written about this layer twice already in the past two weeks.
On the first of April I wrote “The Missing Layer”, arguing that the AI market is pricing in a six-trillion-dollar expansion into knowledge work while ignoring the one layer of the enterprise that does not exist yet: the layer where decisions are made and the reasoning behind them is recorded. On the seventh of April I wrote “The Decision Layer Nobody Built”, showing that the largest study of chatbot usage ever conducted found that the dominant enterprise use of AI is decision support, and that none of it is being captured by any system of record.
Those essays argued the case from two specific angles: venture-capital market structure and chatbot usage data. The McKinsey piece gives me a third angle, operational resilience, and the same layer is missing in the same way.
The decision layer is what sits between a signal and a structured response. It is what checks the sign on the offset. It is what verifies that the awareness on the dashboard translated into action on the ground, that the direction of the action matched the intent of the diagnosis, and that the translation survived the chain of approvals without being inverted somewhere along the way.
Without it, foresight is a smarter dashboard. Flex capacity is a more expensive control tower. Low-exposure sourcing is a more anxious procurement team. Every one of McKinsey’s three transitions runs through this layer. None of them works without it.
Almost no enterprise has built it. Not because they do not need it. Because they have not had a name for it, and the vocabulary of enterprise strategy has, until very recently, been describing the gap in different languages without recognising that it was describing the same thing.
The category forming in public
The category is starting to form in public. On the thirteenth of April, Jaya Gupta of Foundation Capital published a piece called “The Moat Is State”2. Gupta’s argument is that the durable value in AI is not going to sit in the model. It is going to sit in the state that accumulates around the model: memory, organisational context, behavioural calibration, the record of how a system has learned a specific enterprise. Every major player in enterprise AI, Gupta argues, is making a bet on which layer of that state they will own. Most enterprises do not realise that is what they are deciding when they pick a vendor. They think they are choosing a model. In practice they are choosing who will own the reasoning that accumulates inside their operations.
Foundation Capital is looking at this from the venture-capital altitude. McKinsey is looking at it from the chief-operating-officer altitude. I have been looking at it from the architect’s altitude, because I am trying to build it. Three vantages, three vocabularies, one architectural moment. The category is real. It is forming right now, in public, in three different languages, and the vocabularies will converge before the end of the year.
It is also forming faster than most enterprises can choose. The decisions being made in 2026 about how to architect AI into operations, about who owns the state that builds up around those decisions, about where the layer between awareness and action will actually live, will be hard to reverse by 2028. They are the sign check on a much larger bridge.
The line the prescription needs
McKinsey’s diagnosis is right. Operations functions in 2026 are less prepared than they were in 2022. The prescription, three destinations with no mechanism to reach them, needs one more line.
Awareness without verification doubles the problem. At Laufenburg the doubling showed up as fifty-four centimetres of empty air over the Rhine. In 2026 operations functions it shows up as a four-year drop in preparedness that McKinsey has measured, correctly attributed, and cannot explain. The explanation is the one they did not write. It is the layer between knowing and doing.
The bridge opened in the summer of 2004. Locals in Laufenburg still call it the difference bridge. The engineers went back, adjusted the German abutment, equalised the superstructure, redid parts of the Swiss concrete, and made the two halves meet. The cost landed on the Swiss engineering consortium’s liability insurance. Beat von Arx promised, and was able to deliver, that it would cost the taxpayer nothing.
Most enterprises will not be so lucky. Their bridges are not made of concrete. The mistakes are not visible from the middle of the span. The insurance does not cover the gap. And the correction, when it eventually comes, costs not twelve million Swiss francs but the difference between being prepared and being left behind.
McKinsey has given operations leaders an unusually honest diagnosis of where they stand. The diagnosis does not tell them what to build. The thing to build is the layer between what they already know and what they are actually able to do about it. Without it, every dashboard they buy will make the problem worse by making them more confident about actions that have not been verified.
The sign on the offset matters. Someone has to check it.
Sources
2. Jaya Gupta, “The Moat Is State,” Foundation Capital, April 13, 2026
3. Nedeljko Madzarac quote as reported in the Berner Rundschau, January 2004. Beat von Arx quotes as reported in the Stuttgarter Zeitung, January 20, 2004 and Weser Kurier, January 15, 2004.
4. Vertical datum background: Amsterdam Ordnance Datum (NAP), Marseille tidal gauge, Repère Pierre du Niton, Lake Geneva.