A register records what a room concluded. What the next team needs is what the room argued about.
Ask an enterprise how it carries learning from one initiative to the next and it will show you an artefact. Usually a register: a row per lesson, columns for description, impact, action required and responsibility, filed at closure and archived. It exists, it is complete, it satisfies the requirement.
The interesting question is not whether the register is any good, but what the enterprise believed it was buying when it funded one.
J.S. Busby's 1999 Project Management Journal study of post-project reviews suggests the answer is a mechanism pointed in the wrong direction. He watched four review meetings differing in structure, chairing and thoroughness, and reports the successes and failings seemed common to all. One characteristic differentiated them in a way that, in his words, seemed to matter: the presence of outsiders — managers of other projects, invited to attend. His sixth recommendation is to invite them, and his reason is a criticism of the alternative: doing so was probably far more effective at dissemination than written summaries would have been [FACT CHECK REQUIRED] [SOURCE DETAILS REQUIRED].
The Strategic Context
Most enterprises have never separated two questions that behave very differently.
The first is how knowledge gets into a review. Here the teaching orthodoxy is not documentary at all, and it is worth saying so, because the usual criticism of lessons-learned practice attacks a position nobody holds. The standard material prescribes meetings, surveys, interviews, observations, audits and informal discussion, drawing on the project manager, the team, consultants, contractors, clients and end users. An older course handout adds that the project team must always be included and that the project should be viewed from several perspectives — including that of other projects. Nothing there is naive about how knowledge is assembled.
The second is how knowledge gets out — from the initiative that produced it to whoever faces the same problem next. Here the orthodoxy narrows to one mechanism: record it, document it, archive it for future use. The register is the whole outward path.
The divergence between the study and the teaching material sits entirely on the second question, and so does the enterprise consequence. An organisation can run an exemplary gathering process, produce an accurate record, and transfer almost nothing.
What a Written Summary Cannot Carry
Busby's criticism of summaries is about content rather than effort.
A summary is composed from one standpoint, so a reader cannot tell how contentious an issue was — whether the finding was obvious to everyone or was the surviving side of a long argument between two functions who still disagree. And summaries, in the study's phrasing, often lack the detail that adopting a new practice depends on. A conclusion is portable and cheap to write; the conditions under which it holds are neither.
Compare what an outsider took away. The managers of new projects who sat in did not only see the headlines; they saw the reasoning that produced the conclusions, and got a sense of the context in which the project had taken place — which Busby treats as usually vital to understanding how things succeed or fail.
A second limit is named precisely. Where practices were disseminated, what moved was what he calls propositional knowledge — knowledge you can articulate but not necessarily practise — and he is careful to say it is impossible to know whether those managers would in fact behave differently. Keep that caution. An enterprise measuring transfer by whether people can state a lesson is measuring the weaker half of what it needs.
Reframing the Issue
The reframe is to treat transfer as a design problem with a medium, and to notice that the enterprise has been funding one medium by default.
Almost every organisation's transfer budget is spent on writing, formatting, filing and searching. Almost none is spent on attendance — the hours of a person about to face the same problem, sitting where it is being explained. That allocation was never decided; it followed from the fact that a document is auditable and an hour of attendance is not. For an enterprise whose work has largely become a sequence of temporary undertakings, that default is expensive in a particular way [Related article: From Producer to Orchestrator].
Two neighbouring questions belong to other arguments, and stating the boundaries keeps this one useful.
Whether the loop closes at all — whether post-completion work is funded, and whether the next estimate is measurably different — is a question about funding structure, answered elsewhere [Related article: The Estimating Loop Nobody Closes]. This article assumes the loop is funded and asks what travels along it. Neither substitutes for the other: an enterprise can fund the record properly and still transfer conclusions rather than practice.
And the loss described here is lateral, between peer initiatives, and it happens at the moment of writing. The compression that occurs as a report climbs a reporting hierarchy is a different loss with a different mechanism [Related article: Whose Knowledge Does Your Governance System Actually Hear?].
The Knowledge Nobody Knows They Have
The mechanism that keeps enterprises funding the weaker channel is not budgetary. It is that the people holding the knowledge do not think of themselves as holding any.
Busby found that most participants who had worked on the project under examination said something like I already knew that — and that it typically does not occur to them to tell anyone on another project. He offers three explanations as conjecture rather than finding: they may not realise it matters to others, may not have known they knew it until it was said aloud, or may be too busy. His conclusion is that people generally underestimated the dissemination function of these reviews.
Read that against the register and the shape appears. A register is populated by people who have just concluded that the meeting's value was for someone else, writing down what they suppose a future reader needs. The two most valuable categories — what was contested, and what somebody discovered they knew only when challenged — are precisely what a person in that frame will not think to record.
Ian Whittingham, writing in 2008, put the difficulty from the other end. Because every project is unique and temporary, he argues, each fails or succeeds in its own unique and temporary way — one reason, he suggests, that the success of one project is so hard to carry into another. His article is paywalled and the argument behind that opening is not visible, so it is taken here as the observation it is [SOURCE DETAILS REQUIRED]. It explains why belief in the register survives: if each project really is sui generis, a record of conclusions is all that could transfer, and the disappointing results are nobody's fault.
Two Programmes, Same Loss
Consider, hypothetically, a hotel group refurbishing properties on a rolling programme — six or seven a year, each run by a different combination of internal project manager, designer and builder. The register from the last refurbishment records that the lift upgrade delayed handover. True, and useless. What the team learned, and nobody recorded, was that the delay came from a sequencing convention that works in a property with a service lift and fails in one without — and that two people argued about it for a month before one conceded. The next refurbishment is in a property without a service lift. The conclusion transferred; the condition under which it applies did not.
A national museum's collection digitisation programme, hypothetically, runs in tranches by collection type. Each tranche produces a closure record naming the metadata standard adopted and the throughput achieved. What the cataloguers know — which categories of object defeat the standard, and why the workaround they invented is safe for one class of item and dangerous for another — never enters the record, because nobody asked and because the cataloguers assume everybody knows. Three tranches later, the same workaround is applied to the wrong class of item by a team that read the record and reasonably concluded it was standard practice.
Neither organisation has a bad register. Both have a register carrying the transferable half of what they learned.
Decision Framework
The dissemination design. A short attachment to the review of any material initiative, settled when the review is scheduled rather than when it is written up. Four parts:
- Who from outside will attend, selected by what they are about to do rather than by seniority or availability. The criterion is the instrument's whole value; a governing seat filled by exposure is the same idea in a different setting [Related article: Who Sits on the Board for the Benefits?]. Presence during the work, and what it obliges of someone holding a right of refusal, is a separate matter [Related article: The Right to Reject and the Duty to Look].
- What those attendees may change afterwards, because a person who observes and can alter nothing is an expensive audience.
- What the written record will carry that a person cannot — dates, figures, decisions, the points that were contested and who held them. This is the clause that fixes the register, and it fixes it by narrowing it.
- What the record will not attempt to carry, stated explicitly, so nobody mistakes its silence for completeness.
The outsider test. At our last three reviews, who attended who was not on the initiative, and what did they do differently afterwards? An answer of nobody is common and is the finding.
The contention test. Does our record show what was disagreed about, or only what was concluded? Count the disagreements preserved in the last three closure records. In most enterprises the number is zero, because a record showing disagreement looks like a record of a badly run project.
The standpoint test. Whose account is this record, and who else was in the room who would describe the same events differently? The question is answerable in a minute and rarely asked.
From Strategy to Execution
Immediately. Take the closure record of your last significant initiative to two people who worked on it and ask each, separately, what it leaves out. The gap between their answers measures the standpoint problem, and it costs an hour.
Over the next two quarters. Attach a dissemination design to every review above a threshold, and fund attendance as a named line rather than as goodwill — an hour given by someone with delivery targets is the first thing to go. Narrow the register at the same time to what a document carries better than a person, and stop asking it to carry practice. Then decide what the enterprise retains from this spend as against what leaves in an individual's experience, because that split governs whether any of it is an asset [Related article: What Does the Enterprise Own After a Capability Investment?].
Over years. The durable change is that attendance becomes ordinary — a manager starting a similar initiative expects to sit in on the review of the last one, and nobody treats it as an intrusion. That arrives only if attendance was useful the first several times, which is why the authorisation clause matters more than it looks.
Two further boundaries. What governance is entitled to do with a signal that reaches it intact is argued elsewhere [Related article: What Your Projects Know That Your Strategy Doesn't]. And the depth of access a party buys into work it has commissioned, and whether it can use what it sees, is a different decision with a different cost [Related article: Visibility You Cannot Use].
Signals to Monitor
- The proportion of the transfer budget spent on attendance rather than on writing, formatting and filing. Most enterprises have never computed it.
- Whether anyone from outside attended the last three reviews, and whether they were selected by what they were about to do.
- The number of disagreements preserved in your last ten closure records. Zero is the usual answer and it is not evidence of harmony.
- Whether a practice was adopted elsewhere because someone attended, and whether anyone can name it. A named instance beats a compliance rate.
- How often a register entry is read by someone who was not on the initiative. Where the system can tell you, the number is usually small.
- Whether the reviews that generate the record are producing explanations at all, because a dissemination design attached to a meeting with nothing to disseminate simply moves the problem [Related article: The Review That Cannot Ask Why]. And whether the events reviewed were selected for what they could teach [Related article: Which Failures Does Your Enterprise Decide Not to Investigate?].
Questions for the Leadership Team
- When an insight moves from one of our initiatives to the next, what actually carries it — and can we name a case?
- What share of what we spend on organisational learning is spent on people being present, rather than on documents?
- Do our closure records show what was contested, and would we be comfortable if they did?
- Who selects the attendees at our reviews, and on what criterion?
- If the three people who ran our last major initiative left tomorrow, what would the record be able to tell their successors?
- Have we ever asked someone who read a closure record whether it changed what they did?
Closing Perspective
The register is not the problem, and removing it would not help. It is a residue — an accurate account of what a room concluded, written by someone who believed the conclusions were the valuable part.
They are not. The valuable part is the reasoning that produced them and the conditions under which they hold, and that material has properties a document handles badly: it is contested, it is contextual, and much of it is held by people who do not know they are holding it until somebody in the room challenges them. The only medium in the study's evidence that moved it was a person who was there.
An enterprise choosing here is not choosing between rigour and informality. It is deciding whether its learning budget buys a record of answers or an hour of somebody's attention while the answer is being argued. The second is cheaper, harder to audit, and — on the only evidence available — the one that works. To find out which you have bought, ask the last person who read a closure record what they did differently, and listen to the pause.
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.