Building a knowledge advantage that lasts

In a service business, most of what you sell is knowledge, and most of it is undocumented. It sits in the heads of the people who have been there longest, and it walks out of the building at five o’clock every day with no contractual guarantee of returning. That is not a failure of discipline. It is the normal state of things, and it is worth doing something about.

The goal is not to write everything down. It is to make the knowledge that gets reused easier to find, and harder to lose.

What you actually know

Firms tend to overestimate how much of their knowledge is written down, and underestimate how much of it matters.

The knowledge that matters is rarely the formal kind: the methodology deck, the policy document. It is accumulated judgement: which clients need which handling, why a particular approach failed two years ago, what the real timeline is when somebody quotes six weeks, the explanation an experienced person has now given a new starter five times. None of that is in the shared drive.

Why documentation projects fail

The standard response is a documentation push. It gets announced, a template is circulated, and for about three weeks people write things down. Then a busy period arrives and it stops, permanently.

This is not a motivation problem. Documentation projects ask your busiest people to do extra work for a benefit that accrues to somebody else, later. That trade is unattractive to everyone involved, which is why the approach fails almost every time it is attempted.

Capture from the exhaust

The alternative is to stop asking people to write things down and start collecting what they already produce. Your firm generates a great deal of recorded knowledge as a by-product of doing the work:

• Email threads where a tricky question finally got resolved
• Meeting recordings and the notes taken during them
• Project retrospectives, including the uncomfortable ones
• The answer somebody typed into a chat at four in the afternoon

That material is unstructured and inconsistent, which used to make it close to useless. It is now the most cost-effective source of institutional knowledge you have, because the effort required to turn it into something searchable has collapsed.

Findable beats complete

A knowledge base that covers half of what people ask and returns an answer in ten seconds will beat one that covers everything and takes four minutes to navigate. The second will be abandoned inside a month.

Optimise for the question as it is actually asked, in the words the asker used. Completeness is a goal for later, and possibly never.

Ownership, or it rots

Every knowledge base decays. The only question is whether anybody notices.

The organisations that keep one alive have two things: a named owner for each area, and the habit of correcting the source rather than the recipient. When somebody gets an out-of-date answer, the fix is not to tell them the right one. It is to update the thing that told them the wrong one. That habit is worth more than any tool you could buy.

What ownership is really protecting is trust. One wrong answer that nobody corrects is enough for people to go back to asking a person, and a knowledge base nobody believes is worse than none at all, because it still costs something to keep.

Show where the answer came from

Knowledge assembled out of threads, recordings and retrospectives has to arrive with its source attached: the note it came from, the decision it records, the project it belongs to. An answer with nothing behind it asks to be believed. An answer with the material beside it asks to be checked, and can be.

That matters more once a language model sits between the question and the material, because fluency stops being evidence of anything. Provenance is what makes the difference between a system people rely on and one they quietly stop opening.

The advantage

The reason to do this is not efficiency, although you will get some. It is that a business whose knowledge is institutional rather than personal is a materially more valuable one. It onboards faster, it survives departures, its experienced people spend less of the week answering the same question, it can take on work that would otherwise hinge on one person’s availability, and it is worth more if you ever come to sell it.