Why Knowledge Repositories Capture Documents but Lose the Judgement
Knowledge becomes an organisational asset only when using it is easier than finding the person who remembers it.
A Repository Full of Evidence, Empty of Judgement
A programme office announces that its new intranet library is live. In six weeks, the team has loaded 1,800 documents, indexed 43 lessons and issued contribution targets to every workstream. The numbers look reassuring. Yet when a supplier delay threatens a critical migration weekend, the workstream leader does not search the library. She telephones the commercial manager who handled a similar dispute nine months earlier.
That composite scene captures the central weakness in the present enthusiasm for knowledge management. We have become adept at capturing the residue of work while missing the judgement that made the work succeed. Minutes, reports and presentations are plentiful; the reasoning behind a concession, the warning hidden in a schedule and the credibility of the person offering advice are not.
At the turn of this century, large organisations are investing heavily in intranets, groupware, document-management systems and corporate directories. After the millennium changeover, many also want to preserve the knowledge assembled in exceptional programmes before contractors disperse and experienced staff move on. The objective is sound. The prevailing model is not.
Knowledge management fails when it treats contribution as a clerical duty and retrieval as a technical problem. The real problem is an operating bargain: why should a busy practitioner expose hard-won judgement, and why should another practitioner trust and use it?
The Capture Error
A document can be stored because it is complete. Knowledge is useful because it is situated. Consider the difference between a close-out report saying that a data conversion required “additional reconciliation” and the explanation a programme manager gives over the telephone:
- the source ledger closed two days later than assumed;
- the reconciliation team needed authority to reject incomplete extracts;
- the apparent technical defect was actually an unresolved ownership dispute;
- the successful workaround depended on a finance manager being available before 7 a.m.
The report records the outcome. The conversation reveals the mechanism.
Repositories strip away precisely the signals practitioners use to judge whether experience travels: who produced it, under what pressure, with which constraints, and whether the lesson was tested or merely agreed at a meeting. The more aggressively an organisation standardises submissions, the easier they become to file and the harder they become to believe.
The repository is not the memory. It is useful only when it preserves enough context to let a future decision-maker reconstruct the judgement.
This is why capture campaigns often produce an impressive first harvest and then decay. The programme office requests lessons at closure, people submit what is easy to approve, a librarian classifies it, and usage falls because the material answers yesterday’s reporting requirement rather than tomorrow’s decision.
Culture Is Not a Communications Exercise
Leaders often diagnose low participation as insufficient awareness. They launch roadshows, publish contribution guides and add “knowledge sharing” to objectives. These actions may increase submissions, but they do not repair the bargain.
Three forces usually suppress useful contribution.
- Status: scarce expertise confers influence. Asking an expert to “put everything in the system” can sound like a request to surrender that influence without recognition.
- Risk: a candid lesson exposes decisions, compromises and errors. A sanitised lesson is safer to publish but less useful to read.
- Time: writing for an unknown future reader competes with visible delivery work. Unless capture is part of the work itself, it becomes an unpaid tax on the conscientious.
The retrieval side has its own culture. Practitioners prefer people to repositories because people can be questioned. They can qualify an answer, challenge a false analogy and admit uncertainty. A file cannot. Knowledge becomes an organisational asset only when using it is easier than finding the person who remembers it.
The implication is uncomfortable: organisations cannot purchase sharing while rewarding hoarding, candour while punishing every disclosed error, or reuse while celebrating only original work. Culture here is not atmosphere. It is the accumulated logic of incentives, permissions and daily decisions.
The Strongest Case for Better Technology
The serious objection is that these failures reflect immature systems, not a flawed idea. Current search is uneven. Taxonomies are inconsistent. Many intranets are little more than electronic filing cabinets. Better indexing, expertise directories, common document formats and more reliable networks should reduce friction materially.
That case is correct as far as it goes. Poor tools can certainly defeat good behaviour. A repository that returns 200 undifferentiated results is not neutral; it trains people not to search again. A corporate directory that lists job titles but not current experience cannot route a useful question.
But technology can solve only the friction it can see. It can locate a document; it cannot make the author candid. It can count contributions; it cannot distinguish a politically safe lesson from a tested one. It can connect two employees; it cannot give either permission to spend an hour helping the other.
The choice is therefore not between culture and technology. It is between technology that automates the filing of supply and technology that supports the flow of demand. Most current programmes begin with the first because repositories are visible, countable and purchasable. The second is harder because it requires leaders to change how work is governed.
Replace the Capture Campaign with a Use System
The better approach begins at the moment of need, not at project closure. A transformation programme should identify the recurring decisions where prior experience has economic value: selecting a migration sequence, negotiating a supplier change, approving a data exception, or deciding whether a troubled workstream can recover.
Around each decision, four practices matter.
- Name the question. Capture material against a recognisable decision, not a broad subject heading. “Criteria for accepting incomplete customer records” is more useful than “data migration lessons.”
- Keep provenance and context. Record the authoring role, date, constraints, evidence and circumstances in which the judgement held.
- Provide a steward, not merely a librarian. The steward tests relevance, connects the seeker with an experienced practitioner and retires advice that no longer reflects current operations.
- Close the reuse loop. Ask the recipient what was used, what changed and whether the advice held. That feedback turns an assertion into organisational evidence.
A practical measure follows. Do not begin by counting documents deposited. Count decisions supported, introductions made, hours avoided and pieces of guidance amended after reuse. In the earlier composite programme, suppose the library’s 1,800 documents produce six downloads in a month, while the commercial manager receives 27 requests for advice. The 27 requests are not evidence that knowledge management is unnecessary. They are the demand map the system failed to capture.
The first improvement is not to force that manager to write 27 documents. It is to record the recurring questions, host short facilitated reviews, attach concise decision notes to the relevant programme gates, and make the manager’s successors visible in the expertise directory. The system should learn from the traffic already flowing through the organisation.
The Leadership Test
The present market for knowledge management encourages leaders to ask which platform will hold the organisation’s knowledge. That question is too easy. No platform holds the tacit judgements, working relationships and permission to speak honestly on which useful knowledge depends.
A more demanding question is this: which decisions should become better because this organisation remembers? Once those decisions are named, tools can support them, stewards can curate the evidence, and practitioners can see a credible return for contributing.
The organisations that make progress will not be those with the largest repositories. They will be those that turn asking, answering, testing and revising into part of ordinary work. Capturing knowledge is not the objective. Improving the next decision is.