Why is this urgent now?
Because the people who carry the most context are leaving in large numbers. In US manufacturing alone, a 2024 Deloitte and Manufacturing Institute study projects up to 3.8 million jobs to fill between 2024 and 2033, with about 1.9 million at risk of going unfilled (Deloitte press release). Every one of those seats that turns over hands a successor a role without its history.
What actually gets lost?
Mostly the knowledge nobody thought to write down. Procedures are usually documented somewhere. What isn't:
- The people outside the team the role quietly depends on: the supplier contact who answers on a Saturday, the engineer in another plant who knows the old firmware.
- The history behind open commitments: why a customer was promised a price, what was traded for it.
- The workarounds in the systems: which report is wrong and how to correct for it.
- The judgment calls the expert makes without noticing they are making them.
When should knowledge transfer start?
Months before the last day, not in the final two weeks. The useful material is the expert's normal working rhythm, and it is only visible while they are still doing the work. A rushed handover in the last fortnight captures what the expert can recall under pressure, which is the smallest part.
What are the usual methods, and their limits?
Each common approach is useful. None of them sees the work directly.
| Method | What it captures well | Where it falls short |
|---|---|---|
| Exit or knowledge-transfer interviews | Stories, reasoning, the expert's own view of what matters | Only what the expert remembers to say; depends on the interviewer knowing what to ask |
| Documentation sprints and wikis | Procedures and formal steps | Time-consuming for the busiest person; the informal network and history rarely make it in |
| Overlap and shadowing | Tacit skill, seen in action | Needs both people at once for weeks; covers only what happens to come up |
| Video and Q&A capture tools | Answers to known questions, searchable later | The questions still come from memory |
| Understudy | A record of the role rebuilt from the work itself: projects, commitments, cross-team contacts, systems | Works where the work leaves a digital trace (email, calendar, systems); needs the role-holder's consent, by design |
These combine well. An interview run against a record of the actual work asks better questions than one run from a blank page.
How does Understudy do it?
It reconstructs the role from the work itself into a living record, then generates the handover document, training material, and job description a successor needs. The record belongs to the person whose role it is; they decide what is shared. It maps the work, not the worker. You can see the full output on fictional data in the live demo.
For a practical starting point, use the 100 handover questions checklist.
Are exit interviews enough?
They help, but an interview only captures what the expert remembers to say. It works best when the interviewer starts from evidence of the actual work, so the questions cover what the expert would not think to mention.