How This Presents in Practice
Consider an organisation where a leadership team reaches unanimous agreement on a significant change. The discussion is not rushed. Everyone contributes, the direction is stated clearly, nobody dissents, and the meeting closes with a shared sense that a decision of consequence has been made. Over the following months, implementation is uneven and slow. Some parts of the organisation move quickly. Others make no visible move at all. A few begin work that others assumed had been ruled out.
The explanation that presents itself first is passive resistance: people agreed in the room and objected once they left it. It is a familiar reading, and it has the advantage of accounting for the pattern without troubling the decision itself. The change was sound; some people were not committed to it.
What complicates that explanation is that the agreement appears to have been sincere. Asked separately, members of the team describe the decision in warm terms and believe they have acted on it. What differs is what each of them thought would happen first. One understood the change to begin with a reallocation of effort inside an existing structure. Another understood it to begin somewhere else entirely. Each acted on their own reading, in good faith, and because each reading was compatible with the words that had been agreed, nothing in the following months forced the difference into the open. It surfaced only when the results failed to add up.
The deeper observation is that the discussion established agreement on direction and never tested agreement on specifics. Unanimity at the level of an abstraction is easy to reach and tells a leadership team very little. It is the specifics that are acted on, and the specifics were never in the room. Nobody withheld anything; nobody was overruled. The meeting produced a real agreement about a general statement, and each participant supplied the missing detail privately, from their own vantage point, without knowing they were doing it.
This scenario is an invented illustration. It is included because the shape of it is common, not because it happened.
What would have surfaced a disagreement that nobody knew they had?
Why This Matters
The costs of this pattern are unusually well hidden. A leadership team that cannot agree knows it cannot agree. The disagreement is visible, uncomfortable and, precisely because it is uncomfortable, addressed. A leadership team that agrees on a form of words while holding different pictures of what those words mean has no such signal. The meeting felt productive. The minutes are accurate. Everyone left believing the matter was settled, and so nobody looks at it again.
What follows is not inaction but divergent action. Work begins in several places at once, each version internally coherent and each proceeding on the assumption that the others are doing something compatible. Effort is spent. Budgets are committed. People are moved. For a period that can run to months, the organisation appears to be executing, because from any single vantage point it is. The problem is only legible from a position that can see several parts at the same time, which is generally not where anyone is standing.
By the time the divergence becomes visible, it has usually acquired a second identity. It arrives on the leadership team’s agenda as an execution problem, a capability problem, a resourcing problem, or a problem with a particular function or individual. All of these are more tractable-sounding than the alternative, and all of them are supported by real evidence, because divergent work does produce late delivery, strained capacity and interpersonal friction. The original ambiguity is several months upstream by then and does not present itself as a candidate.
There is a further cost that is harder to price. Each cycle of this kind teaches the organisation something about what agreement in leadership meetings is worth. Where a stated direction is followed by visible incoherence, people below the leadership team do not usually conclude that the direction was ambiguous. They conclude that direction from the top is unreliable, and they adjust — waiting for a second signal, hedging their commitment, or continuing with what they were already doing until the position stabilises. That adjustment is rational at the individual level and corrosive in aggregate. It also makes the next agreement harder to act on, which produces more incoherence, which confirms the adjustment.
None of this requires anyone to behave badly. It requires only that a group of experienced people, discussing something at the level of abstraction at which leadership discussions are usually held, should each fill in the unstated detail differently and have no occasion to discover it.
The organisation does not act on what was agreed. It acts on what each person understood was agreed.
— Zoom Riddle
Agreement on Direction Is Not Agreement on Specifics
The word alignment is used to describe several different states that behave very differently in practice. It is used for agreement on direction — a shared answer to where the organisation is going. It is used for agreement on priority — a shared answer to what comes before what when both cannot be done. It is used for agreement on interpretation — a shared answer to what a stated commitment actually requires of a particular team on a particular Monday. And it is used for something closer to compliance, in which people act on a decision they do not agree with but have accepted. These are not points on a single scale. An organisation can hold the first in abundance and none of the others.
Agreement on direction is the cheapest of the four to obtain and the least informative. Directional statements are, by construction, abstractions: they are written to be true across the whole organisation, which means they are written at a level of generality that leaves the specifics to the reader. A commitment to serve larger clients better, to move faster, to raise quality, or to work more closely across functions is not a false statement. It is an underdetermined one. Two executives can endorse it wholeheartedly while holding incompatible views about which existing activity it displaces — and displacement is the part that has consequences.
This is why unanimity is a weak signal, and occasionally a misleading one. Rapid agreement in a leadership discussion may indicate that the matter is genuinely settled. It may equally indicate that the proposition was pitched at an altitude at which nothing was at stake, so nobody had cause to object. The two look identical in the room and diverge sharply afterwards. Where a decision has real trade-offs — where something must be stopped, delayed or paid for — those trade-offs are the substance of the decision, and a discussion that does not reach them has not tested the agreement it produced.
The mechanism by which the divergence stays hidden is worth stating precisely. Each participant, on leaving the meeting, must convert a general statement into particular action within their own area. That conversion requires information the general statement did not contain: which of the current commitments yields, what counts as a legitimate exception, how much slower the existing work is allowed to become. Each person supplies that information from their own context, their own history, and their own reading of what the organisation values. The results are coherent individually and incompatible collectively. Critically, none of them registers as an interpretation. To the person making it, it is simply what the decision obviously meant.
Michael Beer and Russell Eisenstat, writing in Sloan Management Review in 2000, examined a set of mutually reinforcing barriers to strategy implementation and organisational learning — among them unclear strategy and conflicting priorities, and poor communication up and down the hierarchy — and observed that these barriers tend to remain invisible to the senior team, which is generally the group least likely to hear about them directly. Their account is not that leaders are unaware of the difficulty in the abstract. It is that the specific instance is not available to them, because the people positioned to name it are not the people in the room.
A related distinction runs through the research on execution. Donald Sull, Rebecca Homkes and Charles Sull, in a study of more than 250 companies and several thousand managers published in Harvard Business Review in 2015, reported that the vertical alignment processes most firms use — cascading objectives from the top downward — generally work as designed. What does not work reliably is coordination sideways, between functions and units that must act on the same commitment at the same time. Their finding is uncomfortable for the standard diagnosis: it suggests that the machinery leaders most often reach for when alignment fails is the part that is already functioning, and that the failure sits where there is usually no machinery at all.
Two misconceptions follow from conflating these states. The first is that repetition resolves interpretation. If the direction is restated more often, more clearly, or with more conviction, the reasoning goes, understanding will converge. Repetition of an abstraction produces greater confidence in the abstraction; it does not supply the missing specifics, and increased confidence can make the divergence harder to surface, because everyone is now certain they know what was meant. The second is that the presence of disagreement indicates a problem and its absence indicates health. On the evidence of how these situations develop, visible disagreement is frequently the more benign condition. It is at least addressable. Its absence may mean the question was settled, or may mean it was never asked.
Why the Distance Widens as an Organisation Grows
The gap between what a leadership team agrees and what an organisation understands is not a fixed property of leadership quality. It is substantially structural, and it grows with size in ways that are predictable in shape if not in timing.
Every layer between a decision and the work is a point of translation. The translation is necessary — a general commitment cannot be acted on until someone converts it into particular tasks for particular people — and each conversion introduces interpretation. With one layer, the interpretation is visible: the person who made the decision can usually see what was done with it. With four or five, the version arriving at the front line has passed through several hands, each of which added local context, resolved ambiguities in its own favour, and dropped whatever did not seem relevant to the audience below. No single act of translation is unreasonable. The composition of them can produce something the original group would not recognise.
Jay Galbraith’s work on organisation design, published in Interfaces in 1974, framed this as an information-processing problem rather than a communication problem. As task uncertainty rises, the volume of information that must be processed during execution rises with it, and organisational form is largely a response to that load. This reframing matters for alignment because it separates two things that are usually merged. A leadership team that believes its problem is one of communication will invest in transmission — more forums, clearer decks, better cascade. Where the underlying constraint is the volume of interpretation the structure is being asked to absorb, better transmission of the same abstraction adds load without resolving the ambiguity.
Growth changes the problem qualitatively, not only in scale. Larry Greiner’s account of organisational growth, first published in Harvard Business Review in 1972 and republished as a classic in 1998, made the argument that the arrangements which resolve one phase of growth become the constraint in the next — managers, as he put it, see “a major solution in one period become a major problem in a later period.” The mechanisms through which a leadership team stays aligned are a clear case. In a small organisation, alignment is largely a by-product of proximity: the same people are in most conversations, corrections happen in corridors, and shared understanding is maintained continuously without anyone maintaining it. That mechanism is not weakened by growth so much as quietly outgrown. It stops working at a size at which nobody has decided to stop relying on it, and its failure is initially indistinguishable from a run of individual lapses.
Several other structural changes compound this. Functional specialisation means the people who must act together on a commitment increasingly do not share a vocabulary, a set of measures, or a working definition of what good looks like. Geographic distribution removes the informal correction that co-location supplies. The multiplication of forums means that any given person now hears the direction from several sources, at different times, in different words. And the growth of the leadership team itself changes the character of its meetings: a group of five can hold an argument to its conclusion, while a group of twelve tends to converge on formulations broad enough to accommodate everyone, which is precisely the failure mode described above.
Context alters the shape of this without removing it. In professional and knowledge-based organisations, where authority is distributed and expertise is the basis of standing, a leadership decision is more often treated as an input to local judgement than as an instruction, and divergence is both more legitimate and harder to see. In multi-site operations, distance is physical and the divergence tends to present as inconsistency between locations rather than as disagreement, which routes it toward operational explanations. In production environments, the tighter coupling of work makes some forms of divergence surface faster and more visibly, though this is not the same as making the underlying ambiguity visible — a fast operational failure can conceal a slow interpretive one.
Ownership structure and history matter too. Organisations that grew from a founding group frequently retain informal mechanisms long past the size at which they function, partly because the people who benefit from them are the least positioned to notice their absence. Organisations assembled through acquisition inherit multiple prior interpretations of similar-sounding commitments, which can persist for years beneath a common set of terms.
There is a genuine tension here with no clean resolution. Reducing interpretive distance generally means adding specificity, and specificity has costs: it constrains local judgement in situations the centre cannot see, it dates quickly, and past a point it substitutes compliance for the discretion that made distributed decision-making valuable in the first place. Organisations that respond to divergence by specifying more tend to encounter a different failure — capable people executing instructions that no longer fit their circumstances, and declining to say so because saying so is not their role. The trade-off is real, it moves as the organisation changes, and it does not have a settled answer.
Why the Correcting Information Travels Badly Upward
If divergence of this kind were straightforwardly visible, it would be short-lived. Someone would notice that two teams were acting on incompatible readings and say so. That this often does not happen for months is the part of the pattern that most requires explanation, and it has less to do with candour as a personal quality than with the conditions under which information moves upward.
Amy Edmondson’s field study of 51 manufacturing work teams, published in Administrative Science Quarterly in 1999, found that team psychological safety predicted learning behaviour — asking questions, seeking feedback, discussing error — while team efficacy did not, once safety was controlled for. The finding is often cited loosely, so its specific shape is worth keeping. It is not that confident teams perform better, nor that comfortable teams perform better. It is that where people believe the interpersonal cost of raising a problem is low, the behaviours through which a group learns what is actually happening occur; where they do not, those behaviours are suppressed, and the group’s picture of its own situation degrades without anyone deciding to conceal anything.
Applied to alignment, this has an uncomfortable implication. The person best placed to notice that a stated direction is being read two different ways is usually a manager somewhere in the middle, who can see both readings. Raising it requires that person to suggest that a decision the leadership team believes it made clearly was not clear — a statement that is easy to hear as a criticism of the people who made it. Where the cost of that is uncertain, the rational move is to resolve the ambiguity locally and say nothing. This is not a failure of nerve. It is a reasonable response to an unclear payoff, made independently by a number of people who do not know the others are making it.
The filtering that follows is largely structural. Information moving upward is summarised at each stage, and summarisation is selection. What is easy to state, quantifiable, and consistent with the direction already agreed travels well. What is ambiguous, interpretive or awkward — the observation that two teams appear to mean different things by the same commitment — travels badly, because it does not fit the format and cannot be resolved at the level where it is noticed. By the time a report reaches the leadership team, it has typically passed through several rounds of this. Nobody suppressed anything. The channel simply has a shape, and some information is the wrong shape for it.
Erik Hollnagel, Robert Wears and Jeffrey Braithwaite, in a 2015 white paper published jointly by the University of Southern Denmark, the University of Florida and Macquarie University, drew a distinction that generalises well beyond its safety-science origins: between work as it is imagined by those who specify it and work as it is actually done. Their observation is that things go well not because people follow procedures exactly but because “people make sensible adjustments according to the demands of the situation.” The adjustments are usually invisible upward — they are how the work gets done, not exceptions to it — which means the picture available at the top is a picture of the specification rather than of the practice. When something eventually fails, the gap between the two is discovered all at once, and often mistaken for a recent deterioration rather than a longstanding condition.
The structure of decision-making adds a further layer. Paul Rogers and Marcia Blenko, drawing on a Bain survey of executives at more than 350 global companies published in Harvard Business Review in 2006, reported that only around 15 per cent said they had an organisation that helped them outperform competitors, and described four recurring points at which decision-making stalls: between global and local, between the centre and business units, between one function and another, and between the organisation and its outside partners. These are the seams at which two parties can each believe a matter is settled — differently. They are also the seams least likely to be represented in any single reporting line, which is why the divergence they produce tends to reach the top late, if at all.
A consequence worth noting is what happens when a leadership team responds to a late surprise by asking why it was not told. The question is reasonable and its effect is frequently the opposite of what is intended: it raises the perceived cost of being the person who reports an ambiguity, particularly one that cannot yet be stated crisply. What arrives next is more information of the kind that was already arriving, which is to say more of what summarises well. The channel narrows precisely at the moment it is asked to widen.
In practice, organisations often find that the surprises which are hardest to explain afterwards were known somewhere in the organisation for a considerable period, and were not withheld so much as never given a form in which they could travel.
What the Organisation Is Actually Aligned To
There is a version of this problem in which everyone understands the stated direction perfectly well and the organisation still does not follow it. It is worth separating from the interpretive case, because it looks similar from the top and has a different cause.
Steven Kerr’s paper in the Academy of Management Journal in 1975, later republished as an Academy Classic, examined reward systems in a manufacturing company and an insurance firm alongside settings in politics, medicine, war and academia, and found a recurring divergence between what organisations said they hoped for and what their systems actually rewarded. His formulation of the underlying mechanism is deliberately plain: “most organisms seek information concerning what activities are rewarded,” and then do those things, often to the exclusion of activities that are not rewarded. The behaviour is not cynical. It is a rational reading of the environment, and the reading is generally accurate.
This is where a great deal of apparent misalignment actually sits. A leadership team states a priority. The measurement, reporting, promotion and attention structures continue to reward what they rewarded before, because those structures were built for the previous priority and changing them is slow, contested and frequently overlooked in the enthusiasm following a decision. People below read both signals — the stated one and the structural one — and act on the structural one, because it is the one with consequences attached. From the top, this presents as resistance, or as a failure of communication. From below, it presents as an obvious choice between what is said and what is counted.
V. F. Ridgway’s paper in Administrative Science Quarterly in 1956 examined the dysfunctional consequences of performance measurement and found that single, multiple and composite measures could all produce undesirable effects on overall performance. Among his documented examples: public employment interviewers evaluated on the number of interviews conducted ran fast interviews and placed few applicants; investigators given a monthly case quota selected easy cases as the month closed. His conclusion — that quantitative measures are useful tools whose indiscriminate use produces harm, and that “the cure is sometimes worse than the disease” — is worth holding alongside the fact that his paper is the traceable origin of the maxim about measurement and management that is now most often quoted in support of the opposite position. The popular usage inverts the source, which is itself an instructive observation about how organisational ideas travel.
For alignment specifically, the implication is that a measurement system is a statement of priority whether or not it is intended as one, and it is a more credible statement than anything said in a meeting, because it is attached to consequence and repeated continuously. Where the two conflict, the organisation is aligned — accurately and efficiently — to the measurement. The leadership team observes that its priority is not being followed and reasonably concludes that people did not understand it. In many cases people understood it precisely and also understood what would happen to them if they acted on it.
A third pattern belongs here because it is so often misread as a lack of commitment. Nelson Repenning and John Sterman, of MIT Sloan, examined in California Management Review in 2001 why process improvement efforts frequently fail to take hold, using system dynamics modelling alongside a dozen or more case studies. Their account centres on a self-reinforcing trap: under pressure, managers default to working harder rather than working smarter, and because prevented problems are invisible, the people whose work stops failures from occurring receive no visible credit for it. The organisation’s attention flows to whatever is currently on fire, which is always more legible than the work that would have stopped the fire starting.
Applied to a stated leadership priority, this explains a common and puzzling trajectory. A commitment is made and genuinely meant. The work it requires is developmental — it pays off later, and its success consists of things not happening. Meanwhile the operational demands of the current period are immediate, visible and rewarded on sight. Nobody cancels the commitment. Attention drains from it week by week, each week for a defensible reason, until it is discovered some months later to have quietly stopped. The subsequent explanation usually reaches for insufficient buy-in, which has the effect of locating the cause in the attitudes of people rather than in the pattern of what the organisation notices.
The compounding consequence is a matter of credibility. Each stated priority that is not reflected in what is measured or attended to teaches people something about how to read the next one. Over several cycles, an organisation can arrive at a settled and largely unspoken interpretive practice in which stated priorities are treated as signals of emphasis rather than as commitments, and the real priority is inferred from the reporting pack. Leadership teams are frequently unaware that this practice exists, because it is not something anyone would say aloud, and because the visible behaviour it produces — polite acknowledgement followed by continuity — is easy to read as agreement.
What the Research Shows
Taken together, the published work points somewhere more specific than the general observation that alignment is difficult.
Two findings are worth putting side by side. The study by Sull, Homkes and Sull, drawing on more than 250 companies and several thousand managers, reported 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. The Bain survey reported by Rogers and Blenko, covering executives at more than 350 global companies, found only around 15 per cent saying their organisation helped them outperform competitors. These measure different things and neither is a verdict on organisational health, but both point away from vertical transmission as the primary constraint and toward the lateral seams — between functions, units, geographies and the centre — where commitments must be jointly interpreted and where no single reporting line has a full view.
Edmondson’s work supplies the mechanism by which such problems remain unreported. Beer and Eisenstat’s account adds that the barriers concerned are typically least visible to the senior team. Hollnagel and colleagues supply the general form of the gap between specification and practice. Kerr and Ridgway, from very different starting points, converge on the finding that the signals attached to consequence outrank the signals attached to intention. Greiner supplies the reason the problem recurs in organisations that have previously solved it, and Galbraith supplies the reason it intensifies with size independently of anyone’s competence.
Where the evidence is thinner is worth stating plainly, since the field is not short of confident claims. The widely repeated assertion that a fixed and large proportion of change initiatives fail — most often given as 70 per cent — does not survive examination. Mark Hughes of the University of Brighton, writing in the Journal of Change Management in 2011, traced five separate published instances of the claim and found no valid or reliable empirical evidence supporting any of them. The figure circulates because it is memorable and because it flatters the reader’s sense that the difficulty is universal. Any argument on this subject that leans on it is leaning on nothing, and a leadership team presented with such a figure is entitled to ask where it came from.
Several genuine limitations apply to the rest. Much of this work is decades old; the organisations studied differ from contemporary ones in size, geography and the speed at which information moves. Survey-based findings report what managers say about their own organisations, which is not the same as what is happening in them. Case-based and modelling work identifies mechanisms convincingly but says little about how commonly they occur or in what proportions. And almost all of it identifies conditions associated with the problem rather than establishing what causes it in any particular organisation, which is a distinction easily lost when research is used to explain a situation already in progress.
What the research does not settle is at least as important as what it does. It does not establish how much interpretive divergence is normal, or costly, or tolerable in a given context. It does not say where the line falls between specificity that helps and specificity that hollows out judgement. It does not tell a leadership team whether a particular late surprise is evidence of a systemic condition or an ordinary event. Those questions are answered locally or not at all, and the research is most useful for sharpening them rather than for closing them.
Continue reading
Up: Perspectives
In this area
- Alignment Between Leadership and the Front Line — the guide in this area
- Why a Strong Quarter Sometimes Borrows From the Next One
- Why Symptoms and Root Causes Get Confused in Growing Organisations
- Why Some Organisational Problems Can Only Be Solved by Leadership, Not Systems
For how this pattern presents in other kinds of organisation, see other sectors.
The plan
- How This Presents in Practice
- Why This Matters
- Agreement on Direction Is Not Agreement on Specifics
- Why the Distance Widens as an Organisation Grows
- Why the Correcting Information Travels Badly Upward
- What the Organisation Is Actually Aligned To
- What the Research Shows
- Continue reading
- Questions worth considering
Questions worth considering
- If a decision made at the top were restated by three layers of the organisation in their own words, how closely would the versions match — and who would be in a position to notice if they did not?
- What does the organisation actually reward in practice, and is it the same thing its stated direction asks for?
- When a surprise arrives late, who already knew, and what form would the information have needed to take in order to travel?
Related perspectives: Governance and Strategy Execution.
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.