If you have ever sat down with an expert to capture what they know, you will recognise this. You ask them to tell you about their job. They talk for twenty minutes. You leave with a description of their role that could have come from the job advert, and none of the things that make them good at it.
The expert was not holding back. The question was too big.
Why "tell me about your job" fails
Broad questions get summaries. Asked to describe their work in general, people reach for the version they would give in an appraisal: the main responsibilities, the systems they use, the people they work with. That is the official version, and you could have found it elsewhere.
The useful knowledge is in the detail of particular cases: what they noticed, what they checked, what they did differently from the procedure and why. People do not store that as a summary. They store it as stories. To get it out you have to ask for a story.
There is a second problem. Experts have done the work so often that the steps no longer feel like steps. Ask "what do you do" and they skip the parts that feel obvious, which are often the parts a new person most needs.
Questions that work
These work in any kind of team, from a plant room to a finance office. None of them needs special software.
1. "Talk me through the last time you did it." The single most useful question. A specific, recent case has real detail in it. Let them go step by step, and do not let them slip back into "usually". If they say "normally I would", ask what happened this time.
2. "What was the hardest one you have had?" The difficult case is where the judgement shows. Ask what made it hard, what they tried first, and how they knew it was not working.
3. "What do you check first, and why?" Experts almost always have a first move: a figure they look at, a sound they listen for, a report they open before anything else. It is often the most valuable single thing they know, and they rarely mention it unprompted.
4. "What would you tell your replacement on their first day?" This gives people permission to say the things they think are too obvious or too informal for a procedure: who to ring, what to ignore, which system cannot be trusted.
5. "Where is the manual wrong?" Every expert has a list. Asking for it directly shows that you want the real version, not the official one. Expect a pause, then a lot.
6. "What does a new person usually get wrong?" Experts have watched others make the same mistakes for years. Their answer tells you where the procedure needs the most care.
How to ask them
The questions are half of it. The rest is how you listen.
- Ask why, often. When they describe a step, ask why they do it that way. The reason is what lets someone else handle a case nobody planned for.
- Ask for an example when they generalise. "It depends" is the start of an answer, not the end. Ask what it depends on, and for a time it went one way and a time it went the other.
- Do not correct them. If what they describe differs from the procedure, write it down and ask why. The difference is usually the point.
- Keep it short and narrow. Twenty minutes on one task beats an hour on a whole role. People tire, and the detail fades.
- Use their words. Write down what they said, not your tidy version of it. Their phrasing often carries the meaning: "you can tell it's going to fail" says more than "monitor closely".
Ask more than one person
One expert gives you one version. Ask three people the same questions and you will see what everyone does, what only one person does, and where they disagree. The agreement is your procedure. The differences are your next set of questions.
Try it this week
Pick one task that only one or two people really understand. Book twenty minutes with each of them. Ask for the last time, the hardest case, the first check, and what they would tell their replacement. Write the answers down in their words, and put them side by side at the end of the week.
You will have more usable knowledge than most handbooks hold, and a short list of things to ask next.
If this is on your mind, we would like to hear how your team learns. Thirty minutes with our team, no slides.