A meeting can be well attended, well chaired and entirely sincere, and still contain no diagnosis at all.
In 1999 J.S. Busby, then at Cranfield University, sat inside four post-project review meetings at three companies, analysed the transcripts sentence by sentence, and published the result in the Project Management Journal. Twelve hours of conversation among experienced people, all of whom had agreed the review was worth holding. Across those twelve hours, on his account, nobody asked a diagnostic question. Participants asked plenty beginning with "why", but each was a request for clarification — why was it poor meaning in what way was it poor — rather than for a cause.
The temptation is to dismiss that as four badly run meetings. Busby forecloses it. He states plainly that he cannot claim the same pattern would appear in other organisations; he is equally plain that he comes out strongly in favour of holding these reviews; and he notes that most of the firms had never run one, so judging the practice by a first attempt would not be fair. What he describes is not a failure of the meeting but a property of how capable people reason about events that have already happened [FACT CHECK REQUIRED] [SOURCE DETAILS REQUIRED].
The cost is not the hours in the room. It is that an organisation reviewing without diagnosing spends the money, produces the document, closes the governance item and learns almost nothing — while believing the opposite.
The Strategic Context
Nearly every large organisation requires a review after a significant initiative closes. The rule is uncontroversial and cheap to comply with: it produces a meeting, headings and a record. What it does not reliably produce is an explanation — a statement that this happened because of that, and here is why we believe the link is real rather than coincidental. That is the only output which changes anything, because a remedy without a diagnosis is a guess with a budget attached. Everything else — what went well, what frustrated people, resolutions for next time — is atmosphere.
The gap between the two is invisible from outside. Both look like a meeting and a document; both close the item. An executive reading the completion pack cannot tell them apart, and usually neither can the people who were in the room, because talking at length about a difficult project feels like understanding it.
Why the Absence Is So Hard to See
Three things conceal it.
The room does real intellectual work, in the wrong direction. Busby records three modes of reasoning. Participants argued dialectically, one explanation meeting a contradictory one until a third absorbed both. They replayed sequences to establish what happened when. And they simulated — working out what would have followed had a different supplier been chosen. All three are demanding; two run forward, from a candidate cause to its consequences.
Reasoning forward comes naturally; reasoning backward, from an outcome to the causes behind it, is harder. His summary is blunt: instead of tracing the chain of causes and effects, people jump to remedies and work forward to simulate them. The room feels analytical because it is. It is analysing the future.
The group is learning, but not what anyone intended. Participants discovered how constrained a colleague's role was, and how much less definite an explanation was than each had assumed. Where several had blamed a delay on a single cause, the meeting established the truth was a combination — which is unlearning, and less satisfying than knowledge that adds.
Responsibility drifts outward automatically. He records a strong tendency to explain problems by referring to other parties, and reports that on only two occasions across four meetings did an individual admit an error or a need to change how they worked. He does not read this as bad faith, and supplies the counterweight: successful individuals and firms tend to believe events are within their control, which is why it is worth asking what could have been done even when the causes look external.
Reframing the Issue
The missing diagnosis is not a skills gap. It is the output of two forces, one cognitive and one cultural, and only the second is within a leader's gift. The cognitive force is the directional asymmetry above; exhortation does not remove it, though structure can compensate.
The cultural force deserves an executive's attention, because it is a virtue operating out of context. Busby observes that organisations of the kind he studied strongly promote the norm of being constructive, and that managers prefer people who come to them with solutions rather than problems. In ordinary operating conditions this is a good rule. It suppresses complaint, filters noise and pushes responsibility to where it belongs.
Inside a retrospective it does something else. His finding is that the norm makes people hold back from exploring causation unless they already know they can supply an answer — not primarily because they dislike criticising, but because criticising without a remedy looks bad. A rule requiring a solution alongside every problem is, in a retrospective, a rule against diagnosis. It is enforced most vigorously by leaders who would describe themselves as encouraging candour, and it does its work before anyone speaks.
The generalisation beyond the firms Busby studied is ERANORTH's, not his. What a leadership team can establish is whether the same rule runs in its own organisation.
What the Room Does With a Remedy
The pattern completes itself at the end of the meeting, and here the cost becomes measurable. Busby's finding is that remedies were not analysed at all — only proposed and briefly contested — with side effects unexplored and implementation unplanned. Two meetings were hurried at the close; the pattern held in the others too.
Then he interviewed the participants. None dismissed the reviews as worthless; none gave them unqualified support. He identifies two causes and affirms both: part of the scepticism, he writes, undoubtedly lay in a cynicism about organisations conditioned by long experience of managerial activity that produced no visible improvement — and part lay in the unconvincing quality of the remedies these reviews produced.
That is a compounding loop, and its first half is already present in most enterprises. People arrive believing this sort of exercise changes nothing. The review proposes remedies nobody has examined, which quietly fail or are never attempted, and the belief is confirmed cheaply. The review does not merely fail to teach; it teaches that reviews do not work.
Two Organisations, One Pattern
Consider, hypothetically, a national postal and parcel operator two months after a network modernisation lands nine months late. The review is well organised. Participants replay the sequence, argue about whether the sorting-technology vendor or the property programme drove the slip, settle on a synthesis both sides accept, then discuss what would have happened had the workstreams been sequenced differently. Nobody asks why they were assumed independent, who held that assumption, or what in the organisation's planning practice generates assumptions of that kind. The remedy — sequence dependent workstreams explicitly — is proposed, agreed and never examined. Nobody asks what it would cost, what it would delay, or who would enforce it.
The same shape appears elsewhere. A contact-centre operator, hypothetically, consolidates three legacy systems and overruns by forty per cent. Its review reaches a confident conclusion: the requirements were poor. That is a category, not a diagnosis, and the difference is the point. A diagnosis would ask how the requirements came to be poor, who was positioned to notice, and why nobody did. The category closes the conversation; the diagnosis opens it. Busby's own warning applies — there is very little in a complex undertaking that cannot be put down to a communications problem, and doing so is the start of a diagnosis rather than the end of one.
A further caution concerns the confidence a retrospective earns. Ian Whittingham, in the visible opening of a 2009 column, grants that we collect data, reconstruct the chain of events and compare what participants said they did with what they did — then observes that catastrophic failure still provokes the question but why this particular sequence?, casting doubt on the account just built. The rest sits behind a paywall [SOURCE DETAILS REQUIRED]. The visible part is enough: a narrative can be complete, consistent, and still not a diagnosis.
Both organisations will file a review. Neither will be able to say, two years later, what changed because of it.
Decision Framework
The correction is structural, because the causes are. One instrument and three tests.
The diagnosis mandate, written before the review by whoever commissions it, containing four things.
- The question the review is authorised to answer — a specific causal one, not "how did it go": why two workstreams were treated as independent, why the requirements were accepted as they stood.
- Who may be asked a causal question, and by whom. Naming this in advance turns an accusation into a scheduled item, and exposes a common awkwardness: some people best placed to answer hold approval rights and were never close to the work [Related article: The Right to Reject and the Duty to Look].
- An explicit waiver of the constructiveness rule for the diagnostic phase. A participant may state a cause without owing a remedy in the same sentence. This is the load-bearing clause and should be read aloud, because a norm suspended in writing but not in the room is not suspended.
- A requirement that every proposed remedy is examined — for side effects and for what implementing it would take — before the meeting closes or in a second sitting. Whether a recommendation is then tracked to implementation is a separate discipline argued elsewhere [Related article: Value Management Is Not Cost Reduction]. The concern here is the prior failure: an unexamined remedy is not yet something to track.
The diagnostic-question test. In your last review, was a question asked that sought a cause rather than a clarification, and what followed? Most organisations cannot answer, because the record holds conclusions rather than questions.
The admission test. Did anyone accept a share of the cause, and what happened to them afterwards — in the meeting, and in the six months after? The second half matters most, and your own people already know the answer.
The examination test. Of the remedies your last review proposed, which were examined for side effects and for the effort implementation would take, and what did that find? A remedy surviving unchanged was either very good or never examined — a distinction available only to someone able to tell a real examination from a presented one [Related article: Enough Technical Depth to Test the Answer].
From Strategy to Execution
Immediately. Read the last review's record for questions rather than findings and count the causal ones. An afternoon's work, producing a number hard to argue with.
Over the next two quarters. Write the mandate into the closure requirement, and separate the diagnostic sitting from the remedy sitting — the constructiveness waiver cannot survive a meeting that must also produce answers. Fund the examination of remedies explicitly; it is a small line and will be cut if it sits inside a project that has ended [Related article: The Estimating Loop Nobody Closes]. Decide separately who outside the initiative attends, because a diagnosis nobody carried away stops at the door [Related article: Written Down Is Not Passed On].
Over years. The durable change is to the norm, and norms move by example. The test is whether a senior person can say I think I caused this, and I do not yet know how to fix it in a room with their peers, and have the conversation continue.
Two boundaries. This concerns the review held after the work is finished; the review running during delivery — what it may change and stop — is a different instrument [Related article: Quality Assurance Cannot Tell You the Specification Was Wrong]. And who is entitled to declare an initiative a success, against what criterion and on what date, is not asked here [Related article: Who Is Entitled to Say It Worked?].
Signals to Monitor
- The ratio of causal to clarifying questions in review records. Countable, and almost nobody counts it.
- Whether anyone has accepted a share of a cause on the record in the last year, and what visibly happened to them.
- The proportion of findings that are categories — communications, requirements, resourcing — rather than mechanisms. A high proportion signals a diagnosis that stopped at its starting point.
- What participants say privately about whether the review will change anything. The scepticism Busby found was not hidden; nobody had asked.
- Whether the events reviewed were chosen for what they could teach or for how loud they were [Related article: Which Failures Does Your Enterprise Decide Not to Investigate?], and whether a finding ever reaches someone with authority to act on it [Related article: Whose Knowledge Does Your Governance System Actually Hear?].
Questions for the Leadership Team
- In our last three reviews, did anyone ask a question that sought a cause rather than a clarification — and can we point to it in the record?
- What happens, in practice rather than in policy, to a person who says in a review that they got something wrong?
- Do we require people to bring solutions rather than problems, and have we considered what that rule does inside a retrospective?
- Which of the remedies from our last review were examined for side effects, and by whom?
- If we asked those who attended whether the review would change anything, what would they say — and has anyone asked?
- Can we name one decision, not one process, that is different today because of a completed initiative?
Closing Perspective
The organisations Busby studied were not careless. They were doing something difficult in the wrong direction, under a rule helping them everywhere else. That is why exhortation fails: nobody in the room is choosing not to diagnose. What a leadership team can change is narrow and real — name the question before the meeting, suspend the rule that suppresses causal enquiry for its duration, decide who may be asked, and refuse to let an unexamined remedy count as an output. None of it requires better people.
A review that cannot ask why is not a weak review. It is a different activity running under the same name, at the same cost, producing a document that proves it happened. The choice is whether to keep paying for that, or build something able to carry an uncomfortable answer.
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.