When an institution says that it has learned from experience, one of the simplest ways to test the claim is to imagine replacing the people who experienced the event. The officials who dealt with the crisis move on, the managers who supervised the response retire, the analysts who reconstructed what happened take other jobs and the frontline staff who developed practical workarounds are gradually replaced by people who know the episode only indirectly, if they know it at all. Once that turnover has occurred, the important question is not whether the newcomers can recount what happened, but whether the institution they inherit is different because it happened. If future personnel encounter changed procedures, different expectations, revised decision conditions or other durable consequences of the earlier experience, then some part of the lesson has travelled beyond the individuals who originally learned it. If nothing consequential survives their departure, what appeared to be institutional learning may still have been primarily individual learning taking place inside an institutional setting.

This distinction matters because organisations routinely attribute learning to themselves when the underlying change may depend heavily on particular people. A difficult programme produces experienced staff who become unusually good at recognising certain risks; a crisis creates managers with tacit knowledge of how to coordinate under pressure; repeated failures teach a team which apparently minor signals deserve attention. Performance subsequently improves, and the institution reasonably appears to have learned. Yet the improvement can conceal a fragile dependency. The people involved have changed, but the organisation around them may not have changed enough to preserve what they discovered. When those individuals leave, the capability disappears with them, revealing that experience modified human judgement without becoming sufficiently embedded in the institution that surrounded it.

The turnover test helps expose this fragility because it separates the fact that learning occurred somewhere from the stronger claim that the institution itself was altered by it. People inevitably learn while working inside institutions, and much valuable knowledge will always remain personal, professional and tacit. Institutional learning does not require eliminating that human dimension, nor would such a goal be desirable. The question is whether consequential experience leaves traces beyond the particular minds that encountered it. An institution becomes capable of learning across turnover when later personnel inherit conditions shaped by earlier experience, even though they did not personally undergo the event that produced those conditions. What survives may influence how they work, what they notice or which options are available to them, allowing the consequence of experience to persist after direct experiential continuity has been broken.

This is also why institutional learning should not be confused with institutional memory, even though the two are closely related. Memory concerns the continuing availability of the past: whether an institution can recover what happened, access previous reasoning, retrieve evidence or reconstruct earlier decisions. Learning concerns something different, namely whether experience produced a persistent modification in the institution. An organisation can remember an event extremely well without having changed because of it. Reports can remain accessible, archives can preserve extensive documentation and employees can repeatedly discuss a famous failure while essentially the same institutional conditions continue to reproduce it. Conversely, an organisation can be persistently changed by an experience even when many later employees cannot consciously recall the original episode. The past may no longer be widely remembered as an event, yet its consequences can remain active in the present.

Consider a public organisation that experiences a serious service failure because staff repeatedly underestimate an unusual category of risk. During the immediate response, a small group develops considerable expertise and becomes highly skilled at recognising early warning signs. If the organisation relies primarily on those people thereafter, performance may improve for several years, creating a convincing impression that the institution has learned. The turnover test asks what happens when those people leave. If their successors return to the earlier pattern because the institution still presents information in the same way, expects the same judgements and provides no durable mechanism through which the earlier insight affects current work, then the learning was never sufficiently institutionalised. Experience improved the people who lived through it, but the institution remained dependent on their continued presence.

The same test can reveal the opposite pattern. Suppose future staff arrive with no direct knowledge of the original failure yet encounter an organisation in which relevant signals are now treated differently, certain questions are routinely asked and particular risks receive attention before decisions are finalised. Those employees may never have read the original review or spoken with the people involved, but their behaviour takes place within conditions that exist because the earlier experience changed the institution. The lesson has acquired a life beyond personal recollection. This does not mean that everyone acts mechanically or that human judgement has become irrelevant; rather, experience has altered the environment in which future judgement occurs, allowing people who did not personally learn the original lesson to inherit some of its consequences.

Turnover is therefore useful as a diagnostic because institutions normally experience it anyway. Political leaders change, civil servants rotate between roles, contractors leave, teams are reorganised and generations of professionals gradually replace one another. An institution whose learning survives only while particular individuals remain in place is vulnerable to ordinary organisational time. It may repeatedly rediscover the same lesson because each cohort must learn again through direct experience what previous cohorts had already discovered. From the perspective of individual development, learning continues to occur; from the perspective of institutional capability, however, the organisation is paying repeatedly for knowledge it has already had the opportunity to acquire.

This problem is particularly important after crises, when learning can appear strongest precisely because individual memory is unusually vivid. People who experienced an emergency remember the failures, improvisations and consequences in ways that shape their judgement for years afterwards. They know which formal procedures were misleading, which informal relationships proved indispensable and which warning signs seemed insignificant until it was too late. While these people remain influential, the organisation may operate with impressive situational intelligence. Yet crisis-generated capability can decay silently as personnel change because the institution mistakes the continuing presence of experienced people for durable organisational modification. The turnover test forces a more demanding assessment by asking whether the lesson remains consequential after the experiential generation has gone.

The same principle applies to successful experience. Institutions do not learn only from failure, and successful programmes can generate valuable judgement about coordination, timing, public response or implementation. But success can also become dependent on the people who produced it. A team may discover an effective way of working together without making the underlying conditions reproducible for its successors, leaving later staff with the outward form of the programme but not the practical intelligence that made it successful. When personnel change and performance declines, the organisation may attribute the deterioration to weaker individuals when the deeper problem is that earlier success never became sufficiently institutional. The people learned how to succeed; the institution did not fully learn how to preserve the conditions of that success.

Artificial intelligence and digital systems make the turnover test more important rather than less. Digitalisation can create an impression of persistence because systems remain after people leave, but technological continuity alone does not establish that the relevant lesson has become institutional. A system can preserve an obsolete rule just as effectively as a useful one, while an AI-enabled workflow can reproduce previous judgement at scale without preserving the reasoning that would allow future users to recognise when circumstances have changed. The relevant question remains whether experience has produced a durable and appropriate modification in future institutional action. Technology may become one means through which that happens, but the presence of technology should not be confused with the existence of learning.

The turnover test also changes how organisations might interpret succession. Personnel replacement is usually treated as a human-resources or continuity problem, but it can also function as an empirical test of institutional cognition. When experienced people depart, the organisation discovers which capabilities belonged to them and which had genuinely become properties of the wider institutional architecture. Sudden performance deterioration after turnover is therefore not merely evidence that valuable employees were lost; it can reveal where the institution had mistaken concentrated individual competence for institutional capability. Conversely, continuity of intelligent action across successive generations provides stronger evidence that experience has altered something more durable than personal expertise.

None of this implies that every lesson must survive indefinitely or that institutions should preserve all modifications generated by previous experience. Some lessons become obsolete as circumstances change, and mature institutions need the capacity to revise what they have previously learned. The turnover test addresses a prior question: whether a lesson that the institution currently considers valid has become capable of surviving beyond the individuals who first acquired it. Persistence is not proof that the lesson remains correct, but without some form of persistence there is little basis for claiming that the organisation itself has learned rather than merely hosted people who did.

This gives institutional learning a demanding but practical criterion. After a significant experience, an organisation can ask what would remain different if everyone directly involved were gradually replaced. The answer need not be that future personnel remember every detail, reproduce every interpretation or consciously recognise the historical origin of current arrangements. What matters is whether they inherit an institution whose future cognition and action have been consequentially modified by what happened before they arrived. The strongest evidence that a lesson has become institutional is not that the people who experienced it remember what they learned, but that people who never experienced it enter an institution that is already different because the learning occurred.