The Knowledge That Leaves

Organizations lose knowledge continuously and absorb it without difficulty, because a single departure leaves colleagues who can fill the gaps.

The damaging pattern is different. People who joined at similar times leave at similar times, and a cohort that accumulated its understanding together departs together, taking a shared body of knowledge that nobody remaining ever held.

Four categories, and only one is documented

Procedural knowledge. How the work is done. This is generally written down, and it is the least valuable of the four because it is also the easiest to reconstruct.

Causal knowledge. Why arrangements exist. Why invoicing runs on fourteen days, why that supplier is used despite the price, why the committee includes an outside member. Each was a reasonable response to something, and the something is documented nowhere.

Relational knowledge. Who to call at the county and what they actually need in order to move something. Which supplier will flex on terms in a bad year. Which neighbouring district will cooperate and which will not. This is often the most valuable and it is entirely personal.

Pattern knowledge. How things tend to go here. What a quiet crew usually precedes. Which kinds of complaint escalate and which resolve. This is judgment in the sense used throughout this Library, and it took decades to build.

Why nobody records it

Not negligence. Three ordinary mechanisms.

Tacit knowledge is invisible to its holder. Somebody who has known why the fourteen-day cycle exists for twenty years does not experience it as knowledge. It is simply how things are, and they will not think to mention it because it does not present as information.

Nobody asks. The person who knows is rarely interviewed about what they know, because their colleagues also know it and the gap only becomes visible after they leave.

And the moment for asking arrives at the worst time. A retirement is a scheduling event handled by whoever is covering the transition, and the last month is consumed by handover of tasks rather than transfer of understanding.

What loss looks like from inside

It does not present as a knowledge problem, which is why it goes unaddressed for years.

Decisions take longer because nobody is sure whether something has been tried. A practice is removed as inefficient and the original problem returns eight months later, apparently unrelated. Relationships that used to be easy become formal and slow. Mistakes recur that the organization made before and corrected.

Each of these gets attributed to the new people being less capable. Occasionally that is true. More often the new people are operating without an inheritance nobody realised was being transmitted.

The exit interview is the wrong instrument

Standard exit questions ask about the experience of working somewhere. What is needed is entirely different material and different questions.

What here would look strange to somebody arriving fresh? Which arrangements exist for reasons nobody would guess? Who do you call when something needs to move, and what do they actually need from you? What has this organization tried that did not work? What are you worried will be lost when you go?

That last question is remarkably productive. People generally know exactly what they hold, have never been asked, and are frequently uneasy about it.

Record cases, not procedures

The most efficient form is not a manual, which nobody writes and nobody reads.

It is a dozen situations in the person’s own words. What was the situation, what did you know at the time, what did you decide, what happened. A page each, recorded by anybody willing to sit with them for an afternoon and type.

Cases carry all four categories at once. The reasoning is embedded in the situation, the relationships appear as they were used, and the pattern is demonstrated rather than asserted. A manual carries only the first category and does it worse than the existing procedures.

An afternoon is the entire cost, and the window for it closes on a date the organization can see coming years in advance.

Attach reasoning as you go

The version that costs nothing is prospective rather than remedial.

Every time a practice is adopted, a single line records why. This check was added in 2027 after the notification failure. Attached to the practice itself rather than filed elsewhere, so a successor encounters it without knowing to search.

One sentence, written at the moment the reasoning is obvious, prevents two failures. It stops a capable successor removing something load-bearing because it looks like habit, and it stops an obsolete practice being preserved indefinitely by people who dare not touch what they do not understand.

The Idaho timing

A large cohort across this state’s districts, cooperatives, family firms, and volunteer bodies is reaching the end of its working life at roughly the same moment.

These are people who managed water through drought years, ran manufacturers through two recessions, and held county offices through booms and reversals. What they hold exists in no document, and the current default is that it retires with them.

The recovery mechanism is not sophisticated. It is somebody sitting down with them, before the retirement date, and asking. Most will give the afternoon willingly, and most have never been asked.