Every established business carries invisible infrastructure: the way quotes actually get priced, the sequence for closing a month, the list of things to check before a crew leaves the yard. None of it is written down, because for years it didn't need to be — the people who knew it were in the room. Then someone leaves, or takes leave, or gets sick, and the business discovers how much of its operation was stored in one person's head. We have watched this play out dozens of times, and the pattern is always the same: the cost of documentation was trivial compared with the cost of its absence.
The real price of tribal knowledge
The direct costs are obvious once itemised: weeks of reduced output while a replacement learns by asking, the errors made before the questions were thought of, and the senior hours consumed answering them. The indirect costs are larger and quieter. Work that only one person understands cannot be delegated, so the business cannot grow past that person's capacity. It cannot be improved, because a process that lives in memory cannot be measured or redesigned. And it cannot be sold — buyers and valuers discount heavily for processes that leave with the staff. Documentation is the antidote to all four, which is why we call it insurance rather than administration.
Our position: document the four things that matter
We are not advocates of documentation for its own sake. Writing down everything is a form of procrastination, and most of it is never read. In our experience, four documents carry almost all of the value:
- Core workflow maps. One page per critical flow — order to invoice, job to completion, month-end close — showing the steps, the owner of each, and the system it happens in. If a step has no owner, that is the first finding.
- The recurring task runbooks. Written instructions for the tasks that happen weekly or monthly: the payroll checklist, the stock order routine, the funder report. These are the ones that hurt most when the person who does them is away.
- The credentials and systems register. What software the business runs, who administers it, and where the recovery information lives. We have seen businesses locked out of their own systems for weeks because this register did not exist.
- The exceptions log. The non-standard situations and how they are handled — the customer terms, the supplier quirks, the jobs that don't fit the normal flow. This is the knowledge that is genuinely impossible to reconstruct.
How to do it without stalling the business
The mistake to avoid is the documentation project: a heroic attempt to write everything down that consumes a quarter and produces a folder nobody opens. The approach that works is smaller and steadier. Pick the process whose failure would hurt most, and document it properly — by the person doing it narrating the steps while someone captures them, which takes an hour, not a week. Then pick the next one. Store everything where the work happens, not in a folder nobody visits. And give every document an owner and a review date, because documentation that is not maintained rots into misinformation — which is worse than none at all.
Where our opinion is unfashionable
We think most software-first answers to this problem are backwards. The market sells knowledge management platforms to businesses that have not yet written a single runbook, and the platform becomes a folder of good intentions. Write the four documents first, in ordinary office tools, and let real use tell you whether you need a system at all. In thirteen years we have recommended a knowledge platform perhaps three times — and each time it was to document an already-documented business, not to substitute for the writing.
If the assessment of your own business just produced an uncomfortable silence, that is the insurance gap. A scoped engagement can close the four critical documents in a few weeks — before the situation writes the case for you.
Close the gap before it tests itself
We document core workflows and runbooks as a fixed-scope engagement.