In short: Organisations rarely lose their reasoning gradually. They lose it at transitions: the first hire, the first manager layer, an acquisition, a regulator's first visit, a founder's departure, a reorganisation, a platform migration. Each is predictable, and each has a specific defence.
Key points
- Seven transitions break the record: first hire, first manager layer, acquisition, first audit, leader departure, reorganisation, platform migration.
- Treat an acquisition as the largest handover the company will ever do, and keep the acquired team's graveyard.
- Before a founder leaves, start extraction interviews months ahead and prioritise the graveyard and relationships.
- In a reorganisation, reassign record ownership in the same plan that reassigns teams and budgets.
- For platform migrations, test the export before signing and reconstruct sample decisions afterwards.
The rungs in this movement describe steady states. The damage happens in between. Each transition changes who decides, who remembers, or where records live, and each tends to happen under time pressure, when nobody is thinking about the record.
Seven transitions
The first hire. Breaks: the assumption that everyone was there. Defence: make the ledger readable by strangers before they arrive, not after.
The first manager layer. Breaks: founders as the default decision-makers. Defence: write decision rights and the first decision classes the month managers are appointed.
An acquisition. Breaks: everything at once. Two organisations with different definitions, different decision histories, and leaders who may leave within a year. Defence: treat integration as the largest handover the company will ever do. Run handover packs for every acquired leader who owns decisions. Map the acquired glossary to yours rather than overwriting it, and open a conflicts list for definitions that diverge. Keep the acquired company's decision history intact and searchable, marked with its origin. Its graveyard is especially valuable: it records what the acquired team already tried in markets you are about to enter.
The first regulator or major customer audit. Breaks: informal decision-making in any process that is now in scope. Defence: identify in-scope decision classes early, and introduce traces and evidenced review there first, before the request arrives.
A founder's or long-tenured leader's departure. Breaks: the largest single store of unwritten reasoning in the company. Defence: start the extraction interviews months, not weeks, before. Prioritise the graveyard and relationships, which are least likely to exist anywhere else.
A reorganisation. Breaks: ownership of records. Decision classes move between teams, and the old owner's records become orphaned. Defence: reassign record ownership explicitly as part of the reorganisation plan, alongside headcount and budgets.
A platform migration. Breaks: history and links. Records move, but supersession links, sources and outcomes often do not. Defence: test the export before signing the new contract, and verify after migration that a sample of decisions can still be fully reconstructed.
The common pattern
Every one of these is known in advance. The defence is always cheaper before the transition than after it. The failure is almost always the same: the transition is planned in terms of people, money and systems, and the reasoning is assumed to follow on its own. It does not.
In practice
- Add "what happens to the decision record" to the checklist for every acquisition, reorganisation, leadership change and platform migration.
- For acquisitions, run handover packs for every acquired leader and keep the acquired graveyard.
- For reorganisations, reassign record ownership in the same document that reassigns teams.
- For migrations, run a reconstruction test before and after.
What would prove this wrong. Organisations that take no specific action on records during these transitions showing no measurable loss of decision quality or institutional memory afterwards.
Common questions
When do companies lose institutional knowledge?
Mostly at transitions: the first hire, the first manager layer, an acquisition, a first regulator or customer audit, a leader's departure, a reorganisation, and a platform migration.
How do you preserve decision history during an acquisition?
Run handover packs for every acquired leader who owns decisions, map the acquired glossary to yours rather than overwriting it, and keep the acquired company's decision history intact, searchable and marked with its origin.
How do you avoid losing records in a software migration?
Test the export before signing the new contract, and after migrating verify that a sample of decisions can still be fully reconstructed, including supersession links, sources and outcomes.
Why do transitions cause knowledge loss?
They are planned in terms of people, money and systems, and the reasoning is assumed to follow on its own. It does not.