Why This Matters
Most organisations that describe themselves as “growing pains” are not, in fact, watching something break. They are watching a mechanism that was correctly sized for an earlier version of the organisation stop being correctly sized for the current one, while everybody involved continues to behave as though it still works.
This is the argument at the centre of one of the most cited pieces of research on organisational growth. Larry Greiner, writing originally in Harvard Business Review and revisiting the argument twenty-six years later for the same publication, described organisations moving through five identifiable phases — creativity, direction, delegation, coordination and collaboration — each of which ends when the very practice that made the previous phase work becomes the source of a crisis in the next one. Greiner’s own phrase for this is instructive: managers, he wrote, “experience the irony of seeing a major solution in one period become a major problem in a later period.”
That irony is the subject of this guide. It does not attempt to tell a reader which phase their organisation is in, or what to do about it — the research reviewed here describes a pattern, not a prescription, and this page stays on the description side of that line. What it does is go stage by stage through three things that reliably change as organisations grow: how decisions get made, how work gets coordinated, and how leadership behaves. Each of these can continue running on its original design well past the point where that design is still serving the organisation, and each fails in a way that is easy to misread as a different problem entirely.
What Changes in How Decisions Get Made
In a small organisation, concentrating decisions in a founder or a small founding team is not a compromise — it is usually the fastest and most coherent way to decide anything. The same people hold the context, the same people hold the authority, and very little is lost in translation between the two.
As headcount grows, the number of decisions that want to route through those same few people grows faster than their attention does. Research from Bain & Company, based on a survey of executives at more than 350 global companies, found that only about 15% of respondents described their organisation as one that helps the business outperform its competitors on execution. The same research identified where decision-making typically stalls: not in one place, but at one of four recurring bottlenecks — between global and local levels, between the centre and a business unit, between one function and another, or between an organisation and its outside partners. The bottleneck moves as the organisation grows; the underlying pattern, of decisions queuing at a boundary rather than being made at the point closest to the information, tends to recur.
Consider an organisation where two hundred people now report, directly or indirectly, into a structure that still routes an unusually high number of decisions through its founder. This is not because the founder has refused to delegate — by every account, the founder has repeatedly and explicitly asked people to decide without him. But several decisions made in his absence were later revisited by him: not overturned, but reopened and discussed again. Nobody in the organisation experiences this as a contradiction. What they experience is a pattern — reopening reads as reviewing, whatever the founder intended it to mean — and they adjust their own decisions accordingly, whether or not he asks them to.
The observation worth sitting with is not that the founder said one thing and did another. It is that the organisation is responding to the behaviour, not the stated policy, and that this happens quietly enough that the stated policy can remain technically true for a long time. What signal does revisiting a decision send, regardless of the outcome of that particular decision? That question does not resolve itself the same way in every organisation. It looks different depending on whether growth has been organic or driven by acquisition, whether the founder is still the majority owner or answering to outside investors, and whether the business was professionally managed from its earliest days or has never operated any other way.
The pattern also reads differently by sector and by maturity. In a business built around a small number of large, high-stakes decisions — a manufacturer approving capital investment, an asset-heavy operator sanctioning a major contract — concentrated decision-making can persist much longer without visible strain, simply because there are fewer decisions competing for the same attention. In a business built around a high volume of smaller, faster decisions — a multi-site operator, a professional services firm juggling client work — the same concentration reaches its limit sooner, because the queue forms faster than any one person can clear it. Neither pattern is a verdict on the organisation; both are consequences of where decisions sit relative to how many of them there are.
What Changes in How Work Gets Coordinated
Early-stage coordination is frequently informal, and informal coordination is frequently excellent — not despite being informal, but because of it. People overhear things. Context travels sideways without anyone deliberately routing it. A team of a dozen people rarely needs a coordination mechanism, because proximity already is one.
Jay Galbraith, in a widely cited paper on organisation design, offered an explanation for why this stops working that has nothing to do with culture or effort. Drawing on Herbert Simon’s work on the cognitive limits of decision-makers, Galbraith argued that organisational form is driven by task uncertainty: as the information-processing demand on an organisation rises, it eventually exceeds what informal, proximity-based coordination can carry, and the organisation has to find some other way to move information — through more self-contained units, more formal reporting systems, or more deliberate lateral relationships between functions. None of these replacements is as cheap as overhearing a conversation. All of them become necessary anyway, because the volume of information that needs to move has outgrown the channel that used to move it.
Research from MIT Sloan and the London Business School, based on a multi-year study of more than 250 companies, points to where this shows up first. The central obstacle to execution in most organisations, the researchers found, is not weak vertical alignment between senior leadership and the front line — most organisations’ alignment processes are reasonably sound. It is cross-unit coordination: the lateral relationships between functions and business units, which rarely receive the same deliberate design attention as the vertical reporting line, and which are precisely what Galbraith’s account predicts will come under the most strain as an organisation grows.
This is not a problem unique to any one function. It shows up in how a growing organisation shares information between sales and delivery, between a head office and a site that opened more recently, or between a product team and the people who support customers using what it built — anywhere two groups whose work depends on each other stop being close enough to coordinate by accident.
There is a second, quieter version of the same problem. A white paper produced jointly by researchers at the University of Southern Denmark, the University of Florida and Macquarie University draws a distinction between work-as-imagined — the version of a process that exists in documentation, in policy, in the org chart — and work-as-done, the version that people actually perform once they have adapted it to the demands of a specific situation. “Things do not go well because people simply follow the procedures and work as imagined,” the authors write. “Things go well because people make sensible adjustments according to the demands of the situation.” In practice, organisations often find that the gap between these two versions widens as they grow — not because documentation gets worse, but because informal adaptation, once invisible and low-stakes at a small scale, keeps happening at a larger one, where it is far less likely that any one person can see the whole picture of what has actually diverged from what was written down.
What Changes in How Leadership Behaves
The leadership style that builds an organisation is rarely the same style that scales one. Direct, hands-on, highly visible leadership suits a small team moving quickly, because the leader is close enough to see problems as they happen and to act on them personally. That same directness, unchanged, can start to work against the organisation once its scale means the leader is no longer close enough to see most of what is happening — but is still behaving as though they are.
Michael Beer and Russell Eisenstat, writing in Sloan Management Review, studied what they called the “silent killers” of strategy implementation: a set of mutually reinforcing organisational conditions — among them a top-down or laissez-faire senior management style, unclear strategy and conflicting priorities, and poor vertical communication — that block execution and learning largely without leadership noticing. “Without transforming that barrier into a capability,” they wrote, “none of the other barriers can be turned into capabilities either.” Their point was not that any one of these conditions is unusual. It is that they compound, and that a leadership style suited to an earlier stage of growth is one of the more common ways they take hold, because nobody experiences a style as having changed — it is the organisation around it that has.
Amy Edmondson’s research at Harvard Business School offers one explanation for why senior leaders are often the last to know. In a field study of manufacturing work teams, Edmondson found that team psychological safety predicted learning behaviour — the willingness to surface mistakes, ask questions and raise concerns — while team efficacy did not, once psychological safety was accounted for. “Team psychological safety is associated with learning behavior, but team efficacy is not,” she concluded, “when controlling for team psychological safety.” At a small scale, a problem that a team does not feel safe raising formally will often still reach a leader informally, through proximity. At a larger scale, that informal route is exactly the one growth has already removed, which means a team’s willingness to say something out loud stops being a nice-to-have and becomes close to the only route a problem has left.
The founder in this guide’s opening illustration and the leadership pattern described here are two views of the same underlying tension. There is no research reviewed for this guide that resolves it cleanly: direct, centralised leadership genuinely is faster in the moment, and distributed judgement genuinely does scale further. What changes as an organisation grows is not which of these is correct, but how much the organisation can still afford the first one before it starts quietly costing more than it saves.
What the Research Shows
Taken together, the research drawn on above — spanning organisational strategy (Sull, Homkes and Sull, MIT Sloan and London Business School), organisation design (Galbraith), decision architecture (Rogers and Blenko, Bain & Company), strategy implementation (Beer and Eisenstat, Harvard Business School), team learning (Edmondson, Harvard Business School) and safety science (Hollnagel, Wears and Braithwaite) — agrees on one thing more consistently than on any other: a mechanism that suits an organisation at one stage does not fail because it was poorly designed. It fails because the organisation around it changed while the mechanism did not, and the resulting problem is very often misattributed to something else — culture, individual performance, insufficient effort — because the mechanism itself is rarely the visible part of the story.
Where this body of work is less settled is on the shape of the change itself. Greiner’s model describes growth as a sequence of relatively distinct phases, each ending in an identifiable crisis. Galbraith’s information-processing account implies something closer to a continuous rise in coordination demand, without a sharp line marking where one stage ends and another begins. The research reviewed for this guide does not resolve which picture is more accurate, and it is worth being honest that it may not need to: the two accounts are asking slightly different questions, one about organisational form and one about information flow, and an organisation experiencing this pattern may recognise elements of both.
What remains genuinely unresolved, across all of this research, is scale. None of the studies cited here — nor, as far as this review found, the wider literature they sit within — offers a generalisable threshold at which a given mechanism stops working. The Bain research reports what proportion of executives feel their organisation supports outperformance; it does not report the headcount at which that changes. That threshold appears to be organisation-specific, shaped by sector, ownership structure and pace of growth rather than by a number that travels between organisations.
Looking for your own sector? The Other Sectors page points to the closest match.
Questions worth considering
- At what point did a decision-making pattern that used to work start requiring more of certain people’s attention than they actually have?
- Which parts of how work gets coordinated are still travelling through channels that assume everyone can see and hear the same things?
- Where does documented practice diverge from what people actually do — and has anyone checked recently?
- What would it take for a leader to notice they are sending a different signal than the one they intend?
- Which mechanisms in this organisation were sized for a version of it that no longer exists?
This guide sits within Growth. A related guide: Why Nothing Ever Leaves the Priority List.
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.