Knowledge Infrastructure
What Happens to Institutional Knowledge When Your Best Person Leaves?
2026-07-15
42% of institutional knowledge is unique to the individual holding it (Atlan). When that person leaves (resigns, retires, or is simply unavailable), colleagues can't perform that portion of the role until the knowledge is painstakingly rebuilt, a process that can take up to two years. Losing a single experienced employee can cost up to 213% of their salary once that lost productivity is counted, and institutional knowledge loss is estimated to cost businesses $1.3 trillion annually in the US alone.
Why this is worse than it looks on paper
Most businesses track the visible costs of someone leaving: recruitment, onboarding, training time. What doesn't show up on a spreadsheet is the knowledge that was never written down anywhere: which supplier actually delivers on time, why a particular client needs to be handled a certain way, the shortcut for a recurring problem that took years to figure out, the context behind a decision made three years ago that nobody else was in the room for.
None of that is malicious or a failure of process. It's simply how expertise accumulates in most organisations: inside people, not systems. It works fine until the person isn't there anymore.
The risk compounds with size and specialism
The more specialised a role, the higher the proportion of knowledge that exists only in that person's head. A generalist's knowledge overlaps with colleagues; a specialist's often doesn't. Businesses with a small number of highly experienced people carrying disproportionate institutional knowledge (common in trades, professional services, and technical teams) are the most exposed, because there's no redundancy if that one person is unavailable.
This isn't limited to people leaving permanently, either. The same gap opens up whenever that person is on leave, off sick, or simply hard to reach at the moment a decision needs making.
What actually closes the gap
The instinct is usually "we should document things better," and that's true, but documentation alone tends to rot: it goes out of date, nobody maintains it, and it's hard to search when you actually need it under time pressure. What holds up over time is knowledge infrastructure built with retrieval in mind from the start: a system that captures what people actually know, connects it to how the business really works, and can be queried in plain language when someone needs an answer, not just stored and forgotten.