The short version
- Do not start from what the leaver thinks they know. Start from what people keep asking them. Those two lists overlap less than anyone expects.
- People are bad at listing their own expertise, because the things they know best have stopped feeling like knowledge.
- Two weeks is enough if you spend week one gathering questions and week two answering them. It is not enough if you spend both weeks writing documents.
- Capture the reasoning behind decisions, not just the decisions. The successor's expensive mistakes come from re-choosing options that were already ruled out.
Someone hands in their notice on a Monday. They are the only person who understands how the reconciliation job works, or which clients are on the legacy contract, or why the deployment order matters. You have two weeks, and most of those two weeks will be eaten by their own wind down.
"Write up everything you know before you go." Three weeks later, the successor is stuck on something that is not in the document.
This happens on almost every departure and it is not because the leaver was careless. It is because the instruction was impossible.
The wrong question to start with
"What do you know that nobody else does" is the wrong opening question. People cannot answer it well. Expertise is mostly invisible to the person holding it: the things they know best feel obvious, so they do not make the list. What ends up in the document is the material they find interesting, which is rarely the material the next person will be stuck on.
The right question is asked of everyone else: "What do you ask this person, and what happens when they are not there?" That list is external, specific, and grounded in real friction. It is also short enough to finish.
A leaver given a blank page writes a system overview. A leaver given forty real questions answers forty real questions, and does it in a fraction of the time.
Week one: collect the questions
Do not put the leaver on documentation duty yet. Spend the first week gathering, from three sources.
- Ask the team. One message to everyone who works with them: what do you go to this person for, and what would you not know how to do without them? Give them a day. You will get thirty to sixty items.
- Read the leaver's messages. With their agreement, scroll back three months in the channels they are active in. Every question directed at them is a candidate. This catches the things nobody thinks to mention because they have stopped noticing they ask.
- Ask the leaver what breaks. Not what they know. What goes wrong, how often, and what they do about it. People can answer this one, because failures are memorable in a way that routine expertise is not.
Deduplicate into one list. Then rank it: how often does this come up, and what happens if nobody can answer it. Most lists have a top fifteen that matter and a long tail that does not.
Week two: answer them in public
Now give the leaver the list. Answering specific questions is fast, and the answers land in a usable shape because they were written against a real need.
Two rules make the difference. First, answer in the place the question will be asked again, not in a separate document tree. Second, have the successor or the covering person write the draft from a conversation, and let the leaver correct it. The person who does not yet understand something is far better at spotting the gap than the person who has understood it for four years.
| Day | Focus |
|---|---|
| Days 1 to 2 | The top fifteen questions, answered in writing where they get asked |
| Day 3 | Decisions and their reasoning, especially the options ruled out |
| Day 4 | Relationships: introduce the successor by name to every external contact |
| Day 5 | Successor works alone, leaver available. Log every question they hit. |
Day five is the one nobody schedules and the one that finds the real gaps. A day of the successor working unaided, with the leaver still in the building, produces a better gap list than any amount of planned documentation.
Capture the reasoning, not just the outcome
The expensive failures after a departure are almost never "we did not know how to do X". They are "we did X the way it looked like it should be done, and there was a reason it was not done that way."
So for each significant choice in the leaver's area, capture four things: what was decided, when, what the alternative was, and why the alternative lost. The fourth is the one that saves money and the one that vanishes when someone walks out. A format that fits on half a page is in the decision log template, and the wider cost is in the hidden cost of undocumented decisions.
What you will still be missing
Even a good two weeks leaves a gap, and it is always the same gap: current state. Where each live thing stood on the leaver's last afternoon, what was half agreed with which client, which thread was waiting on whom. It is too specific to document in advance and too perishable to survive a handover week.
The durable fix is upstream of the departure. This is what StandIn is built for. Each day, a person spends about ninety seconds on a short brief: what moved, what is open, what is blocked, what is next, mostly drafted from the work that already happened. When they are unavailable, permanently or for an afternoon, their StandIn answers from that brief, in their words, with a source under every answer. It never guesses, and when the answer is not in the record it says so and names who to ask. A team that has been doing that for six months does not face a blank page when someone resigns. See how teams use it, or read knowledge transfer before someone leaves.
Common Questions
What if the leaver is not cooperative?
Then the question list matters more, not less, because specific questions are harder to deflect than a blank documentation task. Keep it short, keep it concrete, and do it in writing so the gaps are visible. Recognising the handover in the reference or the final conversation helps more than escalating.
Should we record video walkthroughs?
For procedures with a lot of clicking, yes. For anything else, no. Video is slow to search, and three weeks later the successor needs one sentence rather than a twenty minute recording they have to scrub through.
Two weeks is not enough. Can we pay for more?
A few days of paid consultancy in the month after departure is usually better value than extending the notice period, because by then the successor knows what they do not know. The questions are real rather than hypothetical.
How do we stop this happening with the next person?
Ask the same question quarterly rather than at resignation: what do you ask this person, and what would stall without them? Doing it while nobody is leaving is cheap and the results are better, because nobody is in a hurry. See how to find key person dependency.
90 seconds, then it's on.
Engineers publish a brief before they log off. The next timezone starts with full context, not a reconstruction of what happened while they slept.