Wisdom Interviews: How to Capture Why Your Best People Get It Right

Somewhere on your team is a person who just gets it right. Same SOP as everyone else, same tools, better results, every time. If you are working out how to capture expert knowledge before employees leave, their steps are the wrong target: the real asset is the set of unwritten principles behind their judgment, and the way to get those out is a wisdom interview. It runs across two recorded sessions a week apart, and what it produces is the judgment layer, not another set of steps.

Two people follow the same SOP. Only one gets it right.

We keep meeting the same pair inside client companies. Two people follow the same procedure. One gets it right every time. The other cannot get it right to save his life.

The steps are identical, so the steps are not the explanation. What separates them is a short list of principles, usually 10 or 12, running unconsciously in the strong performer's head. They were not learned from a manual. They were formed by stories: a grandmother in a kitchen, a first boss, a mistake that hurt enough to become a personal law.

None of that is in your documentation, and adding more detail will not put it there, because the performer does not know the list exists. You are photographing the visible half of the job.

Your best performer is not running a better process; they are running a set of laws nobody has ever seen written down.

What is a wisdom interview?

A wisdom interview is a recorded conversation that surfaces the principles behind an expert's judgment: the unwritten rules that decide what they do when the procedure runs out. It is how you capture tacit knowledge, the stage practitioners also call harvest filming.

It sits one level below normal process documentation. When we document a process, we capture three things: the purpose, the decision points, and the step-by-step from the person actually doing the work. The wisdom interview digs underneath the decision points: why this expert makes the call that way, and which experience taught them to.

Timing is the part owners get wrong. When you capture expert knowledge before employees leave, judgment is the perishable half, because anyone can reconstruct clicks from a screen recording after the fact. If your expert has a leaving date, capture the judgment first: steps can be rebuilt later, judgment cannot.

In practice the capture runs as a short protocol across two sessions:

1. Map the territory. Walk the role end to end with the expert, listing the processes they touch and the calls only they make.
2. Flag the sparks. Mark every moment they light up, bristle, or detour into a story. You are placing flags, not chasing completeness.
3. Take a week off. Send them back to the job with the inventory fresh in their mind.
4. Film the what, then the why. Record a clean walkthrough of the work, then a separate interview about every flagged moment.
5. Ask recursively. Push each story down to the principle underneath it, and read the list back before you leave.

The week between sessions is engineered, not a scheduling accident. The expert returns a richer interview candidate, and practitioners who run these for a living call the collapsed single-session version grossly inefficient. The full mechanics live in the two-interview method.

Emotion is the metal detector

You cannot ask an expert to list their principles, because they do not know them. What they do have is emotion. If there is a feeling attached to a way of working, it has hardened into a law for that person, and the feeling is how you find it.

So the first session is less about notes and more about watching. In our mapping sessions every expert gets the same standing instruction: tell us what frustrates you, tell us where other people get it wrong, and correct us the second the map does not match reality. Frustration is data. The place they bristle is a buried rule.

Three words cover it: wisdom leaves scars. The scars are where you dig.

This is also the honest workaround for a resister. An expert who will not document their work will still talk at length about the shortcut that drives them crazy. You do not need their enthusiasm for paperwork. You need their reactions.

Ask recursively until the story becomes a principle

In the second session, every flag becomes a why. The first why almost never lands on a principle; it lands on a story.

So you keep going. Is it because of this, or because of that? Would you still do it if the deadline moved? What happened the time it got skipped?

Stay in the recursion until the story compresses into a sentence that starts with always or never. The tech who triple-checks a fitting is not careful by temperament; there was an incident, and the incident became a law. The law, not the anecdote, is your deliverable.

Then read the list back. A practitioner who built a career on these interviews describes what happens when you mirror an expert's principles back at them: they look at you like you are a psychic. That reaction is your quality check. A polite nod means you stopped one why too early.

Persistence matters more than phrasing here, though the wider craft of how to interview a subject matter expert makes the digging faster.

Why principles travel further than steps

Steps are local. A documented closeout procedure improves closeout and nothing else. Principles generalize: surface 10 of them from one performer and they will apply to processes you never discussed in the interview.

The math is why this matters. Across 16 small businesses we gap-analyzed, covering 68 roles and 461 process areas, an average of 27% of the work was documented and half the role areas had nothing at all. No team closes that much ground quickly. Surfacing 10 or 12 principles from your strongest operator takes an afternoon, and they keep paying off in corners the project never touches.

Steps solve the process in front of you; principles solve the ones you have not met yet.

This is also why SOPs alone are not enough to capture tribal knowledge: the format has no slot for judgment.

It explains something else too. Your best person is usually your worst explainer, because the principles run below awareness, which is the whole problem of documenting the expert who cannot explain their own work.

Not every process earns this treatment. Two sessions and the engineered week between them take your best performer off the floor twice and push the finished documentation out by a week or more. Run a wisdom interview where judgment decides the outcome; where the work is mechanical, like data entry or order picking, one screen recording captures everything a wisdom interview would.

Put the wisdom layer into training, not a binder

What you do with the principles decides whether the interview was worth running. The familiar failure is capture, format, file, forget. One field services client paid five figures for a complete training platform build; two years later the team had never opened it, and the owner called unused documentation his worst nightmare.

Principles die in a binder even faster than steps do, for the same reason SOPs collect dust: a rule with no story attached reads as an order, and orders get compliance at best.

So put the origin stories into the training itself. Let the expert tell the incident on camera during the filming session, then cut that clip into the module next to the rule it created. The new hire inherits the law with the scar still attached, instead of reading the rule for the first time after the fire.

---

Can AI do this part?

Not this part. AI is genuinely useful downstream of the interview, and we use it there: transcription, formatting, structure. But automated documentation of deep knowledge has a specific failure mode: it is efficient and incorrect, which is treacherous. It looks complete and is missing the one layer that made your expert worth interviewing.

There is a second problem. People do not own what they did not build. An expert who talked their principles into existence across two sessions will defend them in training; a team handed a machine-written manual will treat it like weather. The knowledge has to come from your people, because AI cannot capture what it has never seen.

Surfacing that layer is the core of what we do at The Systems Effect: interviewing the people who hold a business's undocumented knowledge and turning what they say into training their teams actually follow. You can start smaller this week. Book one hour with your best performer, hit record, and ask one question: what is the mistake that changed how you work? Then keep asking why until the story becomes a rule.

Frequently Asked Questions

What is a wisdom interview?

A wisdom interview is a recorded conversation designed to surface the unwritten principles behind an expert's judgment rather than the steps they follow. It runs as two sessions about a week apart: the first maps the role and flags the moments the expert shows emotion, and the second films the work and asks recursive why questions until stories compress into principles. The output is a short list, usually 10 or 12 rules, that explains why the expert gets it right.

Why do two people follow the same SOP with different results?

Because the SOP captures steps, and the difference between the two people is not in the steps. The stronger performer runs on internalized principles formed by mistakes, mentors, and incidents, and applies them at every judgment call the document cannot cover. Until those principles are surfaced and taught, adding detail to the documentation will not close the gap.

How do you capture tacit knowledge?

Watch the work instead of asking for a write-up, record everything, and treat emotion as the signal for where the deep knowledge sits. When an expert bristles, lights up, or insists on doing something one specific way, ask why recursively until the story behind the habit becomes a stated principle. Then verify by reading the principles back: if you captured them accurately, the expert will recognize themselves immediately.

Can AI extract expert knowledge from interviews?

AI is excellent after the interview: transcription, structuring, and formatting are real accelerators. It cannot run the extraction itself, because the wisdom layer only surfaces through live recursive questioning and reading human emotion, and automated documentation of deep knowledge tends to be efficient and incorrect at the same time. Use AI to speed up producing what a human interviewer captured.