How much company knowledge actually sits in one expert's head?
Honestly: nobody can tell you a credible percentage, and you should distrust pages that do. Knowledge doesn't come in countable units, and the studies behind the popular figures measure self-reported estimates. What you can do — better — is measure your own exposure directly: it takes a month, costs nothing, and produces a number that actually describes your company.
Why the popular numbers mislead
Statistics like 'X% of critical knowledge is undocumented' circulate widely and mean little: undocumented relative to what total? Knowledge that was never articulated cannot be counted, by definition — the expert himself doesn't know what he knows until a situation calls for it. The honest formulation: the share is large, it concentrates in judgment rather than facts, and its size in your company is an empirical question you can answer locally.
The measures that actually work
Three proxies, all measurable in a month. The interruption count: how many times per week is this person asked something only he can answer? (Track it — the answer surprises.) The absence test: list concretely what stalled during his last two-week holiday, then extrapolate to three months. The paper-trail test: take three real recent cases he handled and check whether a competent colleague could have resolved them from documents alone. The full diagnostic turns these into a scored risk register.
What the proxies typically reveal
Companies that run this exercise tend to find the same shape: the volume of factual knowledge at risk is moderate — most facts are somewhere — but the judgment layer is almost entirely concentrated: the exceptions, the diagnostics, the client calibrations, the failure memory. That layer is also the hardest to see, because it only activates in situations, and the most expensive to lose, because it cannot be rebuilt from files.
From measurement to decision
The point of measuring is the decision after: for each concentrated domain — spread it, capture it, or accept the risk explicitly. What the measurement removes is the comfortable vagueness: 'we probably should do something about Jan's knowledge someday' becomes a scored line item with an owner and a date.
Where MentX comes in
MentX is a personality-aware knowledge base: it captures the knowledge of a company's key person — from meetings and a guided intake track — into a living, temporal knowledge graph with full source attribution. MentX is the capture branch of that decision: the interruption list becomes the intake agenda, and the judgment layer — the part no percentage ever counted — lands in a living, temporal knowledge graph with full source attribution, askable through DigiME. The diagnostic above is, unchanged, the first conversation of the Founding Partner Program.
