The Afternoon and the Year After.
A board can approve a new strategy in an afternoon. A discussion, then a vote. Done by four o'clock. The organisation expected to deliver it must now reinterpret thousands of decisions that were sensible last quarter and are wrong today.
The gap begins with a difference in speed. A strategy can change in one meeting. An organisation changes through priorities, relationships and routines that people repeat every day.
Most senior leaders have lived through the year that follows. The plan was sound. The people were capable. And still, somewhere between the board paper and the work, the plan lost. Nobody could say exactly where. By the time anyone could, the window for fixing it cheaply had closed.
This paper is about that gap, and about why the first explanation for it is so often wrong.
A decision travels at the speed of a meeting. A pattern travels at the speed of habit.
Where the Friction Comes From.
The old strategy does not leave the building when the new one arrives. It lives on in the priorities people defend, who holds each decision and which meetings still shape the work. It also remains in what gets measured, what gets rewarded and the relationships people rely on when things get difficult.
And none of it is dysfunction: these are the exact habits that made the previous strategy work. An organisation is an attention architecture, accumulated over years, and until someone deliberately redirects it, it keeps pointing where it has always pointed.
So the new demand lands on an arrangement built for the old one, and people absorb the difference. A commercial director holds two priorities that cannot both be first. A decision that used to take a week now crosses three functions and takes a month, so people escalate. Coordination nobody designed gets done invisibly, in the evenings, by whoever cares most. Performance gets misread, because success is still being judged against a strategy that no longer applies. People become the shock absorbers of strategic change, and shock absorbers wear out quietly.
The business consequence is real. Upper-echelons research connects senior-team composition and working dynamics with organisational choices and outcomes. The same research warns against treating executive demographics as a direct measure of underlying behaviour (Hambrick & Mason, 1984; Carpenter, Geletkanycz & Sanders, 2004). How the senior team carries a strategic shift matters. It is also easy to oversimplify.
The Diagnostic Error.
Friction rarely announces its cause. It arrives wearing the face of a person.
A decision is slow, so the leader lacks pace. Perhaps. Or the authority for that decision is split between two roles, and the evidence needed to make it arrives late.
Two functions are locked in commercial conflict, so the team lacks alignment. Perhaps. Or the new strategy rewards outcomes those functions cannot both achieve.
An executive is struggling, so the person is not capable. Perhaps. Or the role changed while the support around it did not, and success is still defined by the previous strategy.
The organisation seems tired of change, so people are resisting it. Perhaps. Or several strategic shifts are competing for the same finite attention and capacity.
Every one of these first explanations might be true. That is what makes them dangerous. Each is flattering to the plan and cheap to act on: replace the leader, push harder. The alternatives are hypotheses that deserve investigation, and in most organisations they never get it, because investigating a system is harder than blaming a person.
Friction rarely announces its cause. It arrives wearing the face of a person.
The cost of getting this wrong is not symmetrical. Remove a capable executive whose role was undeliverable, and you keep the problem while losing a year to discover that. Everyone watching learns what happens to whoever inherits the seat.
What Evidence Is For.
If the first explanation is unreliable, the obvious remedy is judgement: put experienced people in a room and let them read the situation. The research on unaided expert judgement should give any leader pause.
Across 136 comparisons, Grove and colleagues found that mechanical prediction was generally as accurate as, or more accurate than, unaided clinical judgement (Grove et al., 2000). Kahneman and colleagues document how professionals can reach materially different judgements from the same case (Kahneman, Sibony & Sunstein, 2021). Tetlock's work shows why explicit forecasts, regular updates and scoring against outcomes matter (Tetlock, 2005; Tetlock & Gardner, 2015).
Experience itself is not the problem. Unstructured experience is noisy, and noise is invisible from the inside. Two seasoned leaders can read the same organisation and reach confident, opposite conclusions. The organisation will follow whichever of them holds the pen.
Two seasoned leaders can read the same organisation and reach confident, opposite conclusions.
The same discipline applies to reading people. Personality measured in isolation is a weak predictor of performance; the signal appears when a person's stable patterns are read against a specific context and a specific demand (Sackett et al., 2022). And pressure tightens the grip of those patterns. Under load, attention narrows to fewer cues (Easterbrook, 1959). Behaviour shifts towards the habitual, a mechanism with converging though not yet settled evidence behind it (Schwabe & Wolf, 2009). Traits express themselves most strongly in ambiguous, high-stakes situations, which is precisely where plans are won or lost (Tett & Guterman, 2000; Meyer, Dalal & Hermida, 2010).
So the descriptive work belongs to evidence. What changed. Where the new demand now exceeds the present support. Which handovers the plan depends on, and who owns them end to end. How leaders and teams actually decide and escalate under pressure, as opposed to how the organisation chart says they do. Evidence earns this role for a modest reason: it can be checked.
Where Judgement Stays.
It would be convenient if the answer were to hand the whole reading to a model. A meta-analysis of 106 experiments found that human and AI combinations often underperformed the better of the two working alone, even though the combination improved on human performance on average (Vaccaro, Almaatouq & Malone, 2024). The result is not an argument against combining them. It is an argument for testing the combination against both alternatives, for the task in front of it.
In organisational life, important context often sits outside the available data. People close to the work know where trust is weak, which dependency broke three months ago and what was left unsaid in the last meeting. No instrument sees all of this.
Judgement also affects whether people will use the evidence. People abandon an algorithm after one visible error, even when it outperforms them; give them a bounded ability to adjust its output and both adoption and accuracy rise (Dietvorst, Simmons & Massey, 2018). The study suggests that a structured, limited override can improve use and, in the context studied, accuracy.
Honesty sets another limit. Any judgement about whether an organisation can carry a plan is time-bound. It concerns this strategy, this arrangement of roles and dependencies, over a defined period. It is never a prediction about a person's future. Where the evidence is insufficient, the method should say so rather than guess. That makes every stated confidence easier to interpret.
The person is never the data point. A leader is not a score, and a hard quarter is not a verdict. The discipline is to describe the arrangement around the person, with evidence, and to leave the decision where it belongs: with the people accountable for it.
The Question Worth Asking.
Most post-mortems of a stalled strategy ask one of two questions. Was the strategy good? Were the people good? Both usually come back yes, and the post-mortem goes nowhere. The failure may sit in the relationship between the plan and the organisation. It may sit between two roles, between a role and its mandate, or between a team and dependencies it could not control. It may also sit between a leader's operating pattern and the pressure created by the new plan.
The better question is harder and more useful. What must become different for this organisation to carry the strategy?
Asking it early changes what is possible. The moments to ask are recognisable: a new strategy, a change of ownership, a new operating model, a new external pressure, a new dependency, or new evidence that the current view is wrong. Each is a moment when the demand on the organisation and the arrangement of the organisation move apart. Each is far cheaper to examine before friction becomes failure than after.
And the answers come on a ladder, from surface to structural. A priority may need clarifying, or a trade-off naming. The work may need resequencing, a decision may need a different owner, a role may need redesigning. And when the demand and the evidence warrant it, leadership needs to change. A responsible reading keeps several of these open and states why one should be tested first. Changing the system around a person is never an excuse to avoid a difficult people decision. It is the discipline of locating the cause before making one.
A strategy is not translated until people can tell what they must decide differently on Monday.
If the answer remains vague, the translation is unfinished.
Atlas exists because we believe this reading deserves the same rigour as the financial model that sits beside it. Our own method is young, and we hold our claims to the standard this paper recommends. The question itself is available to any leader now, and the best time to ask it is before the afternoon the board says yes.