SME Extraction: Interviewing Your Own Experts So Their Knowledge Becomes Content

The most differentiated content a B2B company can publish is the stuff only its people know: what actually breaks in implementations, which buyer objections are legitimate, what the sales team hears that the website never says. That knowledge already exists in your solutions engineers, support leads, founders, and senior reps. The reason it never becomes content is not that nobody knows things, it is that the people who know things do not write, and the people who write do not know things.

Asking an expert to write a draft is usually where this dies. Writing is a separate skill, drafting is slow for people who do not do it daily, and a busy expert will deprioritize a blog draft forever without ever quite refusing. The fix is to stop asking experts to write and start asking them to talk. Thirty minutes of a good interview typically yields more usable raw material than a month of waiting for their draft, because talking is cheap for experts and structure is cheap for writers.

Come in with a specific angle, not an open topic. Tell me about onboarding produces rambling; walk me through the last onboarding that went sideways and what you saw first produces a story with detail. Questions that reliably surface expertise: what do customers get wrong before they ever talk to us, what advice do you give in calls that we have never published, what would you tell a competitor's customer to check, where does the standard best practice actually fail. Each is designed to pull out opinions and specifics, the two ingredients generic content lacks.

Record and transcribe, always, so the interviewer can listen instead of stenographing. Follow the energy: when the expert speeds up, gets opinionated, or says honestly, most people get this wrong, you have found the piece, even if it was not the piece you planned. Push past first answers with what does that look like in practice and can you give me a real example, because the first answer is usually the polished public version and the second answer is usually the content.

The writer's job is structure and compression, not translation into marketing language. The specific phrases experts use, the blunt heuristics, the odd analogies, are exactly what makes the piece read like a human with experience rather than a content team with a keyword target. A common failure is sanding the transcript into smooth corporate prose, which throws away the differentiation you ran the interview to capture. Keep the expert's verbs and their judgment calls; fix only the meandering.

Send the draft back to the expert with a tight ask: mark anything factually wrong or anything you would not say. Do not invite general wordsmithing, which reopens the writing bottleneck you designed around. Most experts return this review quickly because reviewing for accuracy is fast and flattering, whereas reviewing for style is the slow homework they were avoiding in the first place. Byline the expert where it fits; it costs nothing and makes the next interview easier to book.

One good interview is a nice piece; a standing cadence is a moat. Book a recurring slot, even just one interview every couple of weeks, rotating across sales, support, product, and leadership. Each function knows different things: support knows failure modes, sales knows objections, product knows tradeoffs. A single interview also rarely yields a single piece. One rich conversation typically breaks into a main article plus several smaller assets, because experts answer questions you did not ask on the way to the ones you did.

Protect the experts' side of the bargain. Keep interviews to thirty minutes, arrive prepared, and show them what their time produced. An expert who sees a published piece with their name on it, generating comments or getting used by the sales team, becomes an advocate who volunteers material unprompted. An expert who gives an hour and never sees an outcome learns the opposite lesson. In practice, the second interview with the same person is often better than the first, because trust lowers the polish filter.