How to Interview a Subject Matter Expert (Questions That Actually Work)

Most subject matter expert interviews fail the same way: the interviewer opens a blank document, says walk me through what you do, and types up a tidy summary of a job nobody actually does. Knowing how to interview a subject matter expert comes down to three moves: walk in already knowing what a complete answer looks like, treat frustration and judgment calls as the real content, and record the expert doing the work instead of describing it.

We run these interviews for a living. When we studied 16 small businesses across 68 roles, the work produced 3,718 subject matter expert interview questions. The sessions that turned into documentation people actually follow shared one thing: none of them started from a blank page.

Never Interview a Subject Matter Expert From a Blank Page

An unprepared interviewer spends the expert's hour on questions any benchmark could have answered: what is this role, what are the main tasks, who do you hand off to. The clock runs out, and the real knowledge never surfaces.

Before the session, build a picture of what complete looks like for the role. Take the best training you can find, from a sister location, an industry peer, or your strongest performer, and strip it to its standard sections. Compare that benchmark against what the company has. The gap is your interview.

Walk in knowing what complete looks like, and spend the interview only on what is missing. The question kit we build from that gap sizes each question to the hole it fills, ordered the way the expert thinks, not the way an auditor would. When you are documenting a process that only lives in someone's head, the interview is your only source, so the hour cannot go to warm-up questions.

Pick the Right Expert, Not the Highest Title

A subject matter expert is the person who runs a process day to day and can explain the judgment calls inside it, not the person who owns it on an org chart. Interviewing the manager is the default mistake: they have forgotten what the practitioner still handles every day, or never knew it. Record the practitioner.

The second mistake is a committee. Committees fight, coordinate badly, and eat the clock; one committed expert races to a first 80 percent, then the peer group tears it apart, tunes it, and turns into advocates. Give the work to one committed practitioner, not a committee. Score your candidates on four things:

• wants to participate
• will actually make the time
• carries influence with the crew
• runs the process the way you would want it copied

The third is the sleeper: the galvanizer, the one who can get everyone out for a beer on Friday. Title does not equal fit.

Interview both ends of the tenure scale: the veteran gives you fluency, the newcomer gives you the problems everyone else has stopped seeing. If your best candidate resists the exercise, see what to do when your expert will not document their work.

How Do You Start a Subject Matter Expert Interview?

Start by taking the threat out of the room. An expert who expects an audit gives you the official version; an expert who expects help gives you the truth. Three moves set that up:

1. Set the frame. Tell the expert the session is about how the work could go better, how problems could be avoided, and what frustrates them today. Not compliance, not a performance review.
2. Hand over the screen. The expert shares their screen and does the real work while narrating. We run these sessions with two interviewers, one mapping live and one probing, but the principle survives solo: the expert drives.
3. Give them permission to correct you. Our standing instruction in every session: we want you to say, no, it does not look like that, it is different, I am actually doing this. The map on the screen is a draft for the expert to attack.

The moment the team sees that somebody wants to learn what they do, the documentation stops being management's project and becomes theirs. People support what they helped build.

Treat Frustration as Data

Most interviewers steer away from complaints to keep the session upbeat. Steer into them. Frustration clusters where processes break: the workaround, the double entry, the report rebuilt by hand because an export mangled it. We end every session with the same question: what other frustrations have you noticed, things out of your control, places where other people are inconsistent?

Frustration is not a detour from the interview; it is the interview.

A departing bookkeeper at a home service company once walked us through her weekly commission run on a screen share, at 1am her time. It took 85 minutes just to describe: two reports merged by VLOOKUP because the export destroys job IDs, and screenshots of chat threads to find who the helper was. The owner sat in, watching his own process end to end for the first time in years. His quiet conclusion: a three-day job, every week, and nobody had known.

We closed that walkthrough by telling her, honestly, that we were impressed she could manage it at all. That is not flattery; it is technique. Experts open up when the mess they hold together is honored, not judged.

Watch for the Moments the Expert Lights Up

Somewhere in every good interview, the expert's voice changes. A story arrives with heat behind it: an old mistake, a proud save, a rule they will defend. Emotion means the thing has hardened into a law for that person, and those laws are why your best performer beats everyone else following the same SOP.

Capturing tribal knowledge takes more than a numbered list of steps. Two people can follow the same procedure and only one gets it right every time, because the good one is running 10 or 12 principles nobody ever wrote down, formed by a first boss or a mistake that became a rule. In the first interview, only flag those sparks. Do not chase them yet.

The second pass is where you dig, asking recursively (is it because of this, or because of that?) until the story resolves into a principle you can teach. That pass is a wisdom interview, and it has its own method.

Why Record the Work, Not the Memory of the Work?

We never ask someone to write down how they do their job. We record them doing it: screen captures, narrated walkthroughs, real work in real systems.

The truth about a process lives in execution, not in anyone's memory of it.

Ask an expert to describe a task and they will compress 15 clicks into then I process it. Watch them do it and every click, pause, and judgment call lands on the recording. The more automatic the work, the wider that gap, which is why your best expert is often your worst narrator: that is unconscious competence, and watching beats asking.

Do not map and film in the same sitting: film about a week after mapping. The expert goes back to work with the inventory fresh in mind, notices things they could not name before, and returns a richer candidate. Collapsing the two sessions is the expensive kind of efficient; the two-interview method covers the mechanics.

The honest cost is calendar time: two sessions a week apart put a finished draft two to three weeks out, and the prep falls on you, not the expert. For a task one person can screen record in ten minutes, skip all of this.

Every session fishes for the same three catches: the purpose of the process, the decision points where the right answer depends on context, and the step-by-step as the practitioner performs it. Steps alone make a checklist.

All three make documentation people follow.

---

Close Every Interview With the Next One Booked

An interview that ends with we will follow up is an interview that ends. Book the next session before anyone hangs up. A session on the calendar can be rescheduled; a vague follow-up dies quietly.

Between sessions, turn the recording into a draft while it is fresh. Anything the transcript cannot answer becomes a gap flag, and gap flags become the next session's opening questions, so every interview leaves your guide smarter than it found it. The expert's total cost stays small: 2 to 4 hours per process.

That loop of interview, draft, and correction is the whole engine of The Systems Effect: we interview the people who hold a business's undocumented knowledge and turn what they say into SOPs, process maps, and training their teams actually follow. Run the first lap this week: pick the process people interrupt you about most, find whoever actually runs it, and book an hour with a screen share. Just do not show up with a blank page.

Frequently Asked Questions

What questions should you ask a subject matter expert?

Ask three kinds of questions: purpose (why does this process exist), decision points (where do you make a judgment call, and what does the right call look like), and step-by-step (show me exactly how you did this last time). Open with an improvement frame: how could this go better, and what frustrates you about it today. Steps give you a checklist; judgment calls and frustrations give you the knowledge.

Who should you interview when documenting a process?

Interview the practitioner, not the manager: the person who has done the work for years, still does it, can speak fluently about it, and wants to participate. Influence matters as much as skill, because the expert who helps build the documentation becomes its advocate. Then add the newest person on the team, still close enough to see problems veterans have stopped noticing. One committed expert beats a committee every time.

How long should an SME interview take?

Budget 90 minutes for a working session and expect to use all of it: one commission walkthrough took 85 minutes just to describe the process once. Most processes need 2 to 4 hours of the expert's total time, spread across two or three sessions. Keep mapping and filming about a week apart; the expert returns from the break noticing things they could not name before.

How do you interview someone who is bad at explaining their work?

Stop asking them to explain and start watching them work. Put the task on a screen share, record it, and interrupt with why questions at each pause. Experts with years of repetition compress steps without knowing it, so the recording catches what the explanation skips. If they freeze when asked to describe the job, that is normal: it usually means the knowledge is automatic, not absent.