“We Don’t Even Know What He Knows”: The Bus Factor in Mid-Sized Companies

A call I will not forget any time soon. The managing director of a mid-sized machinery manufacturer, 33 years a family business, calm voice, and then this sentence: “Christian, we don’t even know what he knows.” His best design engineer was retiring. The company was heading into its next audit, a new product generation and customers demanding ever more complex configurations, and the knowledge holding all of it together lived in the head of one person who would leave the company in four months.

This is not an isolated case. In over twenty years I have seen it dozens of times: the sales director who is the only one who can configure the special edition; the project manager whose head contains the entire costing logic; the quality manager who is the only person who really knows the inspection history at every audit.

A question of when, not if

What these situations have in common is called the bus factor: how many people can your organisation lose before critical knowledge is gone? For most mid-sized companies the honest answer for at least one core process is: one. That is not a theoretical risk, because retirement waves make it a question of when, not if.

Knowledge in heads vs. knowledge in systems

The difference between companies that have this problem under control and those that do not is always the same: whether knowledge lives in the heads of individuals or in systems. Today this can be solved with AI-assisted approaches in weeks, not years: configuration logic is captured digitally, project history becomes searchable, experience is recorded in a structured way. But it only works if you start before the fire.

Weeks, not years

The classic answer was the documentation project: multi-year, wiki-based, abandoned. Today the same task is solvable in weeks: configuration logic captured digitally, twenty years of project history made searchable, experience recorded through structured AI-assisted interviews. The expert answers questions instead of writing manuals, and the result is a system colleagues can query, not a binder nobody opens.

FORGE Document Intelligence builds exactly this kind of searchable knowledge base from existing documents and project files.

Where to start

• Map the dependencies. For every core process, ask: who else can do this? A one-name answer puts that process on the risk list.
• Prioritise by impact and time. Retirement dates are known years in advance, so treat them like project deadlines, not surprises.
• Capture in weeks, not years. Structured interviews plus AI-assisted documentation beat the multi-year wiki project that nobody maintains.
• Capture the retrievable, not the complete. Eighty percent searchable this quarter beats one hundred percent documented never.

The retiring engineer will not stay a day longer because the documentation is missing. The only variable you control is how much of those 33 years is still in the building on day one after he leaves.

Retirement is only the most predictable way to lose a key person. How companies lose them to slow decisions instead, we described in The Process Was Correct. The Person Was Gone.

If you recognise your own organisation in this story, get in touch. The first 90 minutes of consultation are free.