How founders can turn what only they know into team knowledge

Most UK small businesses with staff still run on the person who started them. In 2024, 73% of UK SME employers were majority-owned by the person or family who set them up, and in 98% of those the owner was still involved in managing the business (Department for Business and Trade, 2025). ISO 9001:2026, published on 16 September 2026 (BSI, 2026), gives certified businesses a dated reason to check what knowledge they depend on.

In technical businesses specifically, that knowledge lives in the founder’s head as hundreds of small judgements, from which supplier to ring when the usual one is late to the workaround that keeps an ageing instrument in tolerance. A founder planning to step back in the next few years carries all of it. Few have it written down.

This article argues that founder knowledge is captured best from the work itself: record decisions as they happen, then have a named deputy run each one. The method is the same one used in operational excellence work on any undocumented process. You will leave with a one-page knowledge register and a five-question scorecard to check your progress.

Founder knowledge loss is a planning issue

Founder knowledge loss is a planning issue because the concentration is visible years before anyone leaves. The structure of most small businesses puts the knowledge in one place, and the founder’s own timetable says when that place will empty.

• Ownership is narrow. Around 41% of UK SME employers have at most one owner, partner or director, rising to 45% among micro businesses (Department for Business and Trade, 2025).
• Tenure builds the store. Some 48% of SME employers have traded for 21 years or more, which means two decades of decisions, often made by the same few people.
• Experts struggle to say what they know. A 2026 study found that guidance drawn from experts’ recorded work outperformed the framework the experts wrote themselves (Cho and colleagues, 2026).

In summary, the gap shows early enough to be scheduled.

Why does recorded work matter for knowledge transfer?

Recorded work matters because it holds the judgements a founder skips when asked to explain. University of Minnesota researchers had an AI model draw decision rules from real expert conversations, then trained 456 novices on them. Those novices came closer to expert performance than novices given a framework the experts had written through interviews (Cho and colleagues, 2026).

A 2025 study points the same way. An AI agent questioning simulated staff reached 94.9% full-knowledge recall across 864 simulations, recovering what the sole specialist knew without questioning that person directly (Zuin and colleagues, 2025). The two studies test narrow settings, maths tutoring and simulated companies, so both need care. The direction for a founder is still clear: capture from the work, then check it.

What goes wrong when founders write their own manual?

When founders write their own manual, they document the steps they think about and leave out the judgements they have stopped noticing. The binder looks complete and usually passes an audit. The risk sits in the spaces between the documents.

Call this the blank-page trap: faced with an empty document, the founder writes the routine and leaves out the exceptions that only surface when something breaks. Routine procedures are the easier part to rebuild, because the team performs them every week. The real loss arrives with the rare scenario, a fault that turns up once every two years and that only the founder remembers solving.

A cyclist’s manual: knowing more than you can say

Imagine asking an experienced cyclist to write instructions for staying upright. They would describe pedalling and steering, and skip the small corrections they make every second without thinking. Film them riding and the corrections are there to see.

A founder’s judgement works the same way. A written description gives you the pedalling. A recording of a real supplier failure or a difficult customer call shows the corrections as well, because the founder makes them on the recording without being asked.

Three actions for founders to move what they know into the business

Mapping — listing the decisions only you make

Start with decisions, because a task can be taught from a procedure while a decision needs the judgement behind it. For ten working days, note every question that reaches you and who asked it. Then sweep two to three years of incidents, complaints and supplier failures, noting who knew the fix each time. The log catches frequent decisions and the sweep catches rare, expensive ones, so rank the register by cost if the knowledge left, then by frequency.

Recording — capturing decisions while you make them

Capture each decision while you make it, because a recording keeps the corrections a written account drops. With everyone’s consent, record the supplier call or the walk-through of a fault on a phone. An AI transcription tool can turn each recording into a first draft of the steps and the reasons, which you then correct. Correcting is usually quicker than writing from a blank page. Each corrected item moves to “recorded and checked” on the register.

Handing over — letting a named deputy run each item while you watch

Give each item to a named deputy and let them run it while you stay silent, because a silent run shows whether the knowledge has moved. Start with the highest-cost items. After two clean runs, date the item in the register. A stall points to the missing judgement, which goes back for another recording. If you hold ISO 9001, the dated register also gives you something concrete to show during the transition to the 2026 edition (BSI, 2026).

How ready is your business to run on what you know?

Score each question 0, 1 or 2. Add the five scores together, then read the band underneath to see where to start.

Question	0	1	2
Do you have a written list of the decisions only you make?	No	Partly	Yes, documented
Is a second holder named for each high-cost item?	No	Partly	Yes, documented
Have you swept past incidents for rare problems only you solved?	No	Partly	Yes, documented
Have you recorded yourself handling your top five items?	No	Partly	Yes, documented
Has a deputy run each top-five item while you stayed silent?	No	Partly	Yes, documented

0–3: Start the question log this week. 4–7: Record your top five items before you add more to the register. 8–10: Move to silent runs, and put the register on the agenda of every management review so it stays current.

What to do next

This method stops working in two places. Relationships built on personal trust move through introductions over months, so a recording can only describe them. In a business with no plausible deputy, the register makes the gap visible and leaves it open, which turns it into a hiring decision.

Founder knowledge moves into the business when you capture it from the work and prove it with a named deputy. This week, start the question log and note every question that reaches you alongside the person who asked it.