Why this matters
Very few leadership teams experience a failure of strategy execution as an event. There is rarely a moment at which a plan visibly collapses. What happens instead is quieter and considerably harder to name: a set of priorities is agreed, work begins across the organisation, activity is genuinely high, reporting is broadly positive — and eighteen months later the position the strategy was written to reach has not been reached, without anyone being able to point to the decision that caused it.
That absence of a single identifiable failure is the reason the problem persists. Problems with a visible cause get corrected. Problems that present as a slight, distributed shortfall across many areas at once tend to be absorbed into the ordinary noise of running an organisation. Each individual variance is small enough to be explained. Collectively they are the whole story.
The consequence is not simply that a plan underdelivers. It is that the organisation draws a conclusion about why, and that conclusion sets the direction of everything it does next. If the shortfall is read as insufficient commitment, the response is usually more oversight. If it is read as insufficient clarity, the response is usually more communication. If it is read as an unrealistic plan, the response is usually a more conservative one. Each of these is a substantial organisational investment, and each is being made on a diagnosis that is rarely tested.
There is a further reason this territory is worth examining carefully rather than confidently. It is unusually well supplied with claims and unusually poorly supplied with evidence. Widely repeated figures about how often strategies fail circulate without traceable methodology behind them, and are quoted in board papers as settled fact. A field where the headline statistic cannot be sourced is a field where most of what leaders believe about the problem has been inherited rather than examined.
This publication takes the position that the description is the difficult part. Organisations do not generally lack effort, intelligence or intent. What they lack is an accurate account of what is actually happening between the strategy being agreed and the work being done — and the accounts most readily available to them are the ones least likely to be right.
“The strategy did not fail on a date. It stopped being the same strategy somewhere along the way, and nobody was in the room when it happened.”— Zoom Riddle
Agreement on the words is not agreement on the meaning
A strategy is a compression. Months of argument, context, judgement and discarded alternatives are reduced to a small number of statements — sometimes to a single page. That compression is necessary; nothing travels through a large organisation uncompressed. But compression is lossy, and what is lost is the reasoning that made the words mean something specific to the people who wrote them.
Everyone who receives the strategy then decompresses it. They do so against their own function, their own constraints, their own understanding of what the organisation rewards, and their own reading of what leadership actually cares about as distinct from what it has written down. Two capable people reading the identical sentence will produce two different sets of implications, and both will be defensible readings of the text.
This is not a communication failure in the usual sense. The message was transmitted accurately. The difficulty is that a strategic statement is not self-interpreting: it acquires meaning only when someone applies it to a decision, and the decisions differ by function. “Improve client retention” implies one thing to a delivery team weighing whether to accept a difficult piece of work, another to a commercial team weighing a pricing decision, and something different again to a finance team weighing an investment case. All three are executing the priority. They are not executing the same priority.
Why ambiguity survives the moment of agreement
The most uncomfortable aspect of this pattern is that ambiguity is often what made agreement possible. A priority stated with enough precision to exclude interpretations will also exclude some people’s interests, and that produces open disagreement in the room. A priority stated loosely enough that everyone can locate their own work inside it produces consensus. Leadership teams under time pressure, working towards a publishable plan, are systematically pulled towards the second — and the resulting agreement is real, but it is agreement on wording.
Beer and Eisenstat’s work in Sloan Management Review on the barriers to strategy implementation is directly relevant here. Among the mutually reinforcing conditions they identify are unclear strategy with conflicting priorities, and communication that does not travel well vertically. Their central observation is that these conditions are rarely discussed openly, which means they persist in organisations whose leaders would sincerely deny having them.
Written strategy and enacted strategy
There is a related distinction from an unexpected source. Hollnagel, Wears and Braithwaite, writing in a white paper published jointly by the University of Southern Denmark, the University of Florida and Macquarie University, separate work as it is imagined from work as it is actually done. Their argument is that the gap between the two is normal rather than deviant — things go well largely because “people make sensible adjustments according to the demands of the situation”. Their subject is safety, not strategy, but the structural point carries: what is written is a description of intended work, and the real operating rules of an organisation are visible only in what people actually do when the description does not fit.
Applied to strategy, this reframes something that is usually treated as a compliance issue. When local behaviour diverges from the written strategy, the standard reading is that people are not following it. The more useful reading is that people are resolving an ambiguity the strategy left open, using local information the strategy did not contain — and that their resolution may well be better informed than the document.
How it varies by context
Size changes the mechanism rather than the intensity. In a small organisation, interpretation converges through repeated contact: the same handful of people discuss the same decisions often enough that divergent readings surface and get corrected within weeks. That corrective mechanism does not scale, and — importantly — it does not announce its own failure. Nothing signals the point at which there are too many people, too many decisions and too few shared conversations for interpretation to self-correct.
Distance compounds it. Multi-site organisations lose the incidental contact that surfaces divergence. Organisations operating across languages lose it further, because translation forces an interpretation to be committed to before anyone has agreed which interpretation was intended. Organisations that have grown through acquisition carry the sharpest version: two groups using identical vocabulary to describe genuinely different operating assumptions, with the shared vocabulary actively concealing the difference.
The misconception worth naming
The most common response to interpretive divergence is repetition — the same strategy communicated more often, through more channels, with more senior visibility. In practice, organisations often find that this raises confidence without raising convergence. Repeating a sentence does not add the reasoning that was compressed out of it; it adds emphasis. What frequently follows is a leadership team that is more certain the message has landed, working with an organisation that is no more aligned than before, which makes the eventual shortfall harder to explain rather than easier.
The tension underneath has no clean resolution. A strategy specified tightly enough to eliminate interpretation is a strategy that cannot absorb conditions its authors did not foresee, and it will be either broken or ignored within the year. A strategy loose enough to remain useful under changing conditions is loose enough to fragment. Both failure modes are real, and every organisation is somewhere on that line whether or not it has chosen its position deliberately.
Execution problems are mostly lateral
The most persistent assumption about strategy execution is that it is a vertical problem — that the difficulty lies in getting intent to travel from the top of an organisation to the front line, and that solving it is a matter of cascade, clarity and follow-up.
The most substantial evidence available suggests the assumption is largely wrong. A multi-year study of more than 250 companies and several thousand managers by Donald Sull of MIT Sloan, Rebecca Homkes of London Business School and the LSE Centre for Economic Performance, and Charles Sull — published in Harvard Business Review in 2015 — found that only 9 per cent of managers said they could rely on colleagues in other functions and units all of the time, with around half saying they could rely on them most of the time. Their conclusion is that vertical alignment processes in most firms are broadly sound. The failure sits sideways.
That figure repays a moment’s attention, because “most of the time” sounds acceptable and is not. A commitment that holds most of the time is a commitment that fails predictably at the point of maximum load, since that is precisely when the colleague you are relying on is also failing to meet someone else’s demand. Any piece of work that depends on four such commitments in sequence has a completion probability that no individual manager in the chain would recognise from their own reliable-looking contribution.
Why lateral commitments are structurally weaker
Vertical commitments have machinery behind them. They appear in objectives, in reporting lines, in review meetings, in remuneration. There is a named person to whom the commitment is owed and who notices its absence.
Lateral commitments generally have none of that. They are made between peers, often verbally, often in the margins of a meeting called about something else. They are owed to someone with no authority over the person who made them, no visibility of that person’s competing obligations, and no forum in which to raise the shortfall that does not read as an escalation. When a lateral commitment slips, the most common outcome is not correction; it is absorption. The receiving team works around it, and the failure never enters any record.
This has a compounding consequence that matters more than the individual delay. Because lateral failures are absorbed rather than reported, the reporting that reaches leadership is systematically cleaner than the operating reality it describes. Each function reports on the work it controls, which is genuinely progressing. Nobody reports on the seams, because the seams belong to nobody.
Coordination cost is a function of growth, not of competence
Jay Galbraith’s 1974 paper in Interfaces offers the structural explanation. His argument, drawing on Herbert Simon’s work on cognitive limits, is that task uncertainty is what drives organisational form: the greater the uncertainty in the work, the more information has to be processed during execution rather than settled in advance.
Growth increases both terms simultaneously. More people means more specialisation, which means more work whose completion depends on someone in a different unit. More markets, products and locations mean more uncertainty, and therefore more decisions that cannot be pre-specified and must be resolved in flight. The volume of coordination an organisation must perform rises considerably faster than its headcount does.
The important implication is that a growing organisation’s coordination load increases even if nothing about its people, culture or management gets worse. This is the single most useful reframe in the territory: the deterioration is structural. An organisation that has become harder to coordinate has not necessarily declined in any respect. It has become larger, and largeness has costs that were not present at the size where its habits were formed.
Silos, and the misreading of them
Silos are usually described as a cultural failing — the product of territorial managers protecting their own patch. Occasionally that is accurate. More often the pattern is the predictable output of a structure that was deliberately built.
Organisations specialise because specialisation works. A dedicated function accumulates depth, develops its own standards, and is measured on outcomes it can control. Each of those is a benefit, and each of them also widens the distance between functions: separate vocabulary, separate priorities, separate evidence about what matters. The behaviour that gets labelled siloed is, from inside the function, the behaviour the organisation designed the function to produce.
Treating this as a values problem generates responses aimed at attitude — collaboration initiatives, shared events, cultural messaging — for a difficulty that is generated by structure. In practice, organisations often find the effect of such initiatives decays within a quarter, because the underlying incentives were never in question.
Where it differs
Functional structures concentrate the difficulty at a small number of well-known seams, which at least makes it visible. Divisional and regional structures distribute it more widely and disguise it better, since each division looks internally coherent and the losses sit between them. Professional and knowledge-based organisations carry a particular version, in which the person on whom a lateral commitment depends is also the person delivering billable work under their own utilisation pressure — and the commitment loses that contest reliably. Multi-site operations add physical distance to functional distance, so that a dependency failure may not surface for weeks.
The tension here is genuinely unresolved. Reducing internal dependency by making units more self-contained lowers coordination cost and raises duplication, fragmentation and drift. Increasing integration lowers duplication and raises coordination load and decision latency. There is no position on that line that is free, and the organisation’s current position is usually an accumulation of past decisions rather than a current choice.
Commitments accumulate; almost nothing is withdrawn
Ask a leadership team how many strategic priorities it has and the answer will usually be three, four or five. Ask how many named initiatives are formally live across the organisation, with a sponsor, a budget line and a slide, and the answer is frequently an order of magnitude higher — and it is common for nobody to be certain of the number.
That asymmetry is the pattern. Organisations are structurally equipped to start work and structurally unequipped to stop it.
The asymmetry between adding and removing
Starting something has a clear owner, a moment, a business case and a person whose standing rises with it. It is a decision that produces visible energy and is easy to describe positively.
Stopping something has none of these properties. There is rarely a defined owner for closure. The decision produces no visible output — the organisation looks exactly as it did, minus one workstream. It implies that the original commitment was mistaken, which touches the judgement of whoever made it, and it removes work from people who have built expertise and identity around it. The result is that initiatives are far more often abandoned than closed: they lose their meeting, then their sponsor’s attention, then their resourcing, and continue to exist on paper for years.
Capacity, meanwhile, is not elastic. It expands considerably more slowly than the commitment list does, and it is shared: the small population of people capable of leading demanding cross-functional work is the same population every new initiative wants. Each addition is individually justifiable. The aggregate is a portfolio that no realistic assessment of available attention could support, arrived at without any single unreasonable decision.
Why prevention loses to firefighting
Repenning and Sterman’s 2001 paper in California Management Review, based on system dynamics modelling and more than a dozen case studies, sets out the reinforcing dynamic underneath this. Managers under delivery pressure default to working harder rather than working smarter, because effort produces immediate visible output while capability improvement produces benefit later and diffusely. Worse, the people who prevent problems receive no visible credit, precisely because the problem they prevented never occurred and cannot be pointed to. They named this the capability trap, and its self-reinforcing character is what makes it durable: the more time is spent on recovery, the less is available for the improvement that would reduce the need for recovery.
This is why strategic work loses to operational work with such consistency. It is not a failure of discipline. Operational shortfalls are visible immediately, attributable to a person, and painful within the week. Strategic under-investment is invisible, unattributable and painful in eighteen months. Anyone making that trade repeatedly under pressure will make it the same way, and the organisation’s reporting will confirm they were right to.
Measurement makes it worse in a specific way
Two long-standing pieces of work explain why measuring execution more closely often distorts it further.
V. F. Ridgway, writing in Administrative Science Quarterly in 1956 — published by Cornell’s Johnson Graduate School of Management — examined single, multiple and composite performance measures and found that all three could produce undesirable consequences for overall performance. His documented examples included public employment interviewers assessed on the number of interviews conducted, who conducted fast interviews and placed few applicants; and law-enforcement investigators working to a monthly case quota, who selected easy cases as the month closed. His summary judgement was that “the cure is sometimes worse than the disease”. It is worth noting that this paper is the origin of the idea now circulated as “what gets measured gets managed” — and that Ridgway was warning against over-reliance on measures, not recommending them. The popular usage inverts his argument.
Steven Kerr’s 1975 paper in the Academy of Management Journal examined reward systems across a manufacturing company, an insurance firm, and settings in politics, medicine and academia, and found the same structure repeatedly: organisations state a hope and reward something else, and behaviour follows the reward. Applied here, an organisation that states a strategic priority while continuing to recognise, promote and report on operational throughput has not communicated a mixed message. It has communicated a clear one, and it is being followed.
Activity read as progress
The combined effect is an organisation that generates a large volume of genuine, well-intentioned activity attached to strategic language, and a reporting layer that cannot distinguish that activity from progress towards the intended position. Status is reported by initiative rather than by outcome, because initiative-level status is available and outcome-level status is not. Everything is advancing. Nothing has arrived.
The misconception that keeps this in place is that prioritisation is a ranking exercise. Ranking a list changes nothing on its own, because the items below the line are not thereby stopped — they continue to consume the same shared capacity while carrying a lower label. In practice, organisations often find that the ranked list and the resourcing pattern have only a loose relationship to one another, and that this is visible to everyone below the leadership team well before it is visible to the leadership team.
Whether it is strategy or execution is usually contested
When a plan does not produce the intended result, someone has to decide what the failure was. That decision is rarely presented as a decision. It arrives as an observation — the strategy was sound, we did not execute it — and it is the most consequential unexamined judgement in the whole territory.
The evidence at the point of the shortfall is generally compatible with both explanations. A strategy that assumed conditions that did not hold and an organisation that could not deliver a sound strategy produce the same visible outcome: the position was not reached. Distinguishing them requires knowing what the strategy assumed and whether those assumptions held, and organisations very often did not record the assumptions in a form that permits the check.
Why the default lands on execution
There is a structural reason the attribution tends to land on execution. The people best placed to make the judgement are usually the authors of the strategy. Attributing failure to execution locates the problem in the layer below and preserves the quality of the original thinking. Attributing it to the strategy does the reverse. This is not cynicism about individuals; it is an ordinary consequence of who is in the room when the question is asked.
The cost of getting it wrong compounds. An organisation that misreads a strategy problem as an execution problem responds with tighter oversight, more reporting and more review of work that was correctly executed against a flawed premise. The additional load reduces the capacity available for the work, performance declines further, and the diagnosis appears confirmed. The organisation ends up with an execution problem it did not originally have, arrived at through the correction.
Growth changes what execution requires
Larry Greiner’s account of organisational growth, published in Harvard Business Review in 1972 and republished as an HBR Classic in 1998 with new commentary, contains an observation that applies squarely here. Managers, he wrote, experience the irony of seeing “a major solution in one period become a major problem in a later period”.
The mechanisms an organisation uses to turn intent into action are subject to exactly this. Direct founder involvement, informal escalation, a weekly conversation between the people who matter, decisions made by whoever is nearest — these are not deficiencies at small scale. They are efficient, and they are why the organisation succeeded enough to grow. Each of them then degrades quietly as scale increases, and none of them fails visibly. The organisation continues to rely on a mechanism that no longer covers its territory, and reads the resulting shortfall as a performance problem in the people using it.
This is why the same leadership team, with the same capability and the same commitment, can execute well at one size and poorly at another without anything having gone wrong in any conventional sense.
Decision clarity and where it stalls
Paul Rogers and Marcia Blenko, both then partners at Bain, published a survey of executives at more than 350 global companies in Harvard Business Review in 2006. Only around 15 per cent of those surveyed said they had an organisation that helped the business outperform its competitors. Their descriptive account of where decision-making stalls is useful without being prescriptive: they locate the recurring bottlenecks at four boundaries — global against local, centre against business unit, function against function, and inside against outside partners.
What is notable is that all four are boundaries rather than levels. This is consistent with the coordination finding discussed earlier: the friction sits where authority is genuinely shared and neither party is straightforwardly wrong, not where someone has simply failed to follow an instruction.
Related to this, and widely observed, is what happens when ownership of a strategic outcome is unclear. Work does not stop. It drifts — continuing in a direction set by whoever is closest to it, entirely reasonably, without anyone holding the whole. Drift of this kind produces no alarm, because every individual part of it is being done conscientiously.
Why the review does not catch it
Leadership teams generally believe the review cycle protects them from all of this, and it is worth being precise about why it frequently does not. A review inherits the ambiguity of the strategy it reviews: if four people hold four readings of a priority, all four can report progress against it truthfully. It inherits the reporting asymmetry described above, in which functional progress is visible and the losses between functions are not. And it inherits a well-documented reluctance to raise information that reflects badly on the person raising it.
Beer and Eisenstat’s finding is precisely that these barriers tend to be invisible to leadership — not because leaders are inattentive, but because the information that would make them visible is held by people for whom raising it carries a cost. A review conducted with complete integrity by everyone in the room can still produce a picture that is materially wrong.
“Every review is conducted using the organisation’s own account of itself. That account is the thing in question.”— Zoom Riddle
What the research shows
The research on this territory is more consistent about the shape of the problem than about its causes, and considerably less complete than the confident tone of most writing about it would suggest.
Where the evidence converges
Three findings recur across independent bodies of work.
The first is that horizontal reliance, not vertical alignment, is where execution most often fails. This is the central conclusion of the Sull, Homkes and Sull study at Harvard Business Review, and it is structurally consistent with Galbraith’s much earlier argument in Interfaces that rising task uncertainty raises information-processing demand within organisations. Two very different methods, four decades apart, locate the difficulty in the same place.
The second is that measurement and reward systems shape execution behaviour more powerfully than stated intent does. Ridgway’s work in Administrative Science Quarterly and Kerr’s in the Academy of Management Journal were written two decades apart in different research traditions and reach a compatible conclusion: behaviour follows what the system actually registers.
The third is that the information leadership teams need in order to see the problem is systematically the information least likely to reach them. Beer and Eisenstat’s Sloan Management Review work makes this argument directly. Amy Edmondson’s 1999 study in Administrative Science Quarterly of 51 manufacturing teams at Harvard Business School approaches it from a different direction, finding that team psychological safety predicted learning behaviour while team efficacy did not once safety was controlled for. A capable team and a team willing to surface what it knows are not the same thing, and only the second produces usable information.
Where it diverges, and where it stops
Convergence on symptoms should not be mistaken for agreement on causes. Structural accounts locate the difficulty in design — dependency, information load, boundary conflict. Behavioural accounts locate it in what people are willing to say and what they are rewarded for. Both are supported. Neither is sufficient, and there is no settled account of how much of the observed variance belongs to each.
The methodological limits are worth stating plainly rather than leaving implicit. Much of this evidence is self-reported by managers, which measures perception of reliability rather than reliability. Much of it is cross-sectional, which supports association rather than causation. Much of it is drawn from large organisations, and the generalisation to mid-sized ones is assumed more often than demonstrated. And there is no agreed definition of execution failure, which means there is no meaningful base rate — a serious gap for a field in which base rates are quoted constantly.
The figure this publication does not use
The most repeated claim in this territory is that some large proportion of strategies or change programmes fail, most commonly given as 70 per cent. Mark Hughes of the University of Brighton examined this directly in the Journal of Change Management in 2011, tracing five separate published instances of the claim and finding no valid or reliable empirical evidence supporting any of them.
The figure’s earliest traceable origin describes itself as an unscientific estimate, and its author subsequently walked it back. It has nonetheless been cited for three decades, frequently by organisations selling a response to it.
Zoom Riddle does not use it, and treats its persistence as evidence about the field rather than about organisations. A number that survives thirty years without a methodology is telling you how rarely the sources in this territory are checked.
How this presents in practice
The following is an illustrative composite. It describes no actual organisation.
Consider an organisation where the leadership team publishes three strategic priorities for the year. Six months later, a review finds that most departments have made progress on something, but that almost nothing has moved on the second priority.
The obvious reading is weak follow-through: people agreed and then failed to act. It is the reading the situation invites, and it points directly at accountability.
What complicates it is that each department can point to genuine work done, and each believes it was working on the priorities. Asked to describe the second priority in their own words, four department heads give four materially different descriptions. None is careless. Each is a coherent reading of the same published sentence, arrived at by a competent person applying it to their own decisions.
What the situation suggests is that agreement was reached on the wording rather than on the meaning. Nobody disagreed, because nobody realised they were disagreeing. The disagreement was present from the first day and invisible for six months, and the review is the first occasion on which the four readings have been placed side by side.
It is worth sitting with how ordinary this is. No one behaved badly. No one was unclear in a way they could have detected. The mechanism that would have surfaced the divergence — people close enough to each other’s decisions to notice a different interpretation early — is exactly the mechanism that thins out as an organisation grows, and it thins out without announcing that it has.
The question it raises is not easily answered: how would an organisation know the difference between shared understanding and identical wording?
The same pattern appears across many kinds of organisation; see Other Sectors for how it is read elsewhere.
The plan
- Why this matters
- Agreement on the words is not agreement on the meaning
- Execution problems are mostly lateral
- Commitments accumulate; almost nothing is withdrawn
- Whether it is strategy or execution is usually contested
- What the research shows
- How this presents in practice
- Questions worth considering
Questions worth considering
- Which of the organisation’s strategic priorities would different members of the leadership team describe in materially different terms if each were asked to explain it without notes?
- Of the initiatives formally live today, how many could be named — with a sponsor and a reason for existing — by someone outside the team running them?
- When a plan falls short, whose account of the failure is being relied on, and what would an account from outside that room look like?
Related perspectives: Operations and Growth.
Your team is facing similar challenges?
Subscribe to The Riddle Ledger
Perspectives are sent as they are published, with no promotional email, and you can unsubscribe at any time.