25 Questions to Ask a Subject Matter Expert for an Article

Ask a subject matter expert questions that map to the sections your article needs: one set that frames the reader's task, one that retrieves a recent, specific instance of the work, one that produces evidence you can publish — numbers, artifacts, named tradeoffs — one that captures caveats and boundaries, and one that settles what you may quote and how. The 25-question bank below is organized exactly that way, because the test of an SME question is not whether it sounds smart on the call; it is whether the answer has somewhere to go in the draft.

To be clear about intent: these are questions for writing an article with a subject matter expert — a practitioner lending experience to your page. If you searched this while hiring an SME or preparing for a job interview, this is the wrong page.

Who this is for

This is for writers, editors, and content leads who have an SME booked — a colleague, customer, or practitioner — and a blank question doc. It pairs with two neighbors in this cluster: the interview run sheet covers how to structure and sequence the hour, and the interview content brief covers what the article must prove before you schedule anything. This page owns the questions themselves.

The premise behind every question here comes from Google's people-first guidance: the finished article has to demonstrate experience a reader can use, so the questions must retrieve that experience in publishable form — specific, attributable, and checkable.

What questions should you open with?

Open with the reader, not the expert's bio. These calibrate the SME to the audience and feed your introduction and framing sections:

1. "When someone searches this topic, what are they actually trying to get done that day?"
2. "What do people consistently misunderstand before they try it?"
3. "What decision should a reader be able to make after this article?"
4. "Who should not follow the standard advice here?"

Red-flag answer: a restatement of the topic ("they want to understand best practices"). Push with "what were you trying to get done the last time this came up?" — it converts the abstraction into a candidate story.

What questions surface first-hand experience?

These feed the core how-to and story sections, and they all point at one recent, real instance rather than habits or averages. Nielsen Norman Group's interviewing guidance is the basis for that design: people report specific episodes far more reliably than they report what they "usually" do.

1. "Walk me through the last time you did this — from the first step, not the summary."
2. "What took longer than you expected, and where did the time actually go?"
3. "What broke, and how did you notice it had broken?"
4. "What did you try first that you later abandoned, and why?"
5. "What would you do differently if you started again on Monday?"
6. "What detail would a generic article about this always miss?"

Red-flag answer: the composite anecdote — "typically a client will…". Redirect to one real, anonymizable case. If three attempts produce no specific instance, the problem is the source, not your list; rescreen against the sourcing scorecard.

What questions produce evidence you can publish?

Experience only becomes an article when it survives fact-checking and clearance. These feed your tables, examples, and claim ledger:

1. "Roughly what numbers are we talking about — and how precisely did you measure them?"
2. "Is there a checklist, doc, or dashboard from that project you could share, even redacted?"
3. "Which parts of what you have told me are your judgment, and which could I verify against documentation?"
4. "Are any of these claims time-sensitive — tied to a version, price, policy, or market that changes?"
5. "You mentioned a third party's product — is that first-hand observation or something you heard?"
6. "Can you give me an anonymized example we are allowed to print, and what must be removed from it?"

Red-flag answer: confident claims about a third party's product, pricing, or policy. The quote being real does not make the claim true; route those through the fact-checking half of the clearance workflow before anything ships.

What questions capture caveats and boundaries?

Asked after the story — never before — these feed the mistakes, tradeoffs, and "when not to" sections, which are usually the parts a summary cannot replace:

1. "When is the standard advice here actively wrong?"
2. "What would you never do, even though people recommend it?"
3. "Who disagreed with your approach, and what was their strongest argument?"
4. "What is the smallest situation change that would flip your recommendation?"
5. "What should a smart reader still expect to get wrong on the first try?"

Red-flag answer: symmetric hedging ("both approaches have merits"). Ask for the case where they personally chose a side and what it cost.

What questions settle quotes and attribution?

These close the call and feed nothing visible — they protect everything else. Reuters' interview guidance treats ground rules and exact quote handling as agreements, not assumptions, and the same discipline applies to content interviews:

1. "Which parts of this are on the record under your name, and which should be paraphrased or anonymous?"
2. "How should we describe you — role, company, or something more general?"
3. "Does anyone else — comms, legal, your employer — need to see your quotes before we publish?"
4. "Do you want to review your quotes in context before this goes live?"

Red-flag answer: "whatever you think is fine." Pin the default explicitly anyway; "they seemed relaxed about it" is not a permission record.

The section-mapped question bank at a glance

The artifact on this page is the section-mapped SME question bank — the 25 questions above, grouped by destination. Use this summary table to check coverage before the call:

Question group	Feeds which article section	Questions	Red flag to push past
Opening context	Introduction, reader-task framing	1–4	Topic restatement instead of a task
First-hand experience	Core how-to, anchor story	5–10	Composite "typical client" anecdotes
Publishable evidence	Tables, examples, claim ledger	11–16	Unverifiable third-party claims
Caveats and boundaries	Mistakes, tradeoffs, "when not to"	17–21	Symmetric hedging with no side chosen
Quotes and attribution	Nothing visible — protects everything	22–25	"Whatever you think is fine"

If a planned article section has no question feeding it, add one before the call. If a question feeds no planned section, it is trivia — cut it and spend the minute on a follow-up instead.

Worked example: trimming the bank for one article

Assumption: this is our own example, not a real interview. An editor is preparing a guide on "migrating a help center without breaking SEO" with a support-operations SME. From the bank, they keep: reader-task questions 1 and 3, experience questions 5, 7, and 9, evidence questions 11, 12, and 14, caveat questions 17 and 20, and all four permission questions — fourteen questions for a 60-minute call, which is already ambitious. Everything else waits for follow-up email. The trimmed list maps one-to-one onto the article's planned sections, which is the whole point: coverage of the outline, not coverage of the topic.

Claim ledger

Claim	Source	Status	Review rule
People report specific recent instances more reliably than generalized habits, so episode-anchored questions yield better material.	Nielsen Norman Group user-interview and interviewing-users guidance	verified on 2026-09-14	Recheck within 180 days
Interview discussion guides should avoid leading questions and keep the conversation on what the interviewer needs to learn.	Nielsen Norman Group interviewing-users guidance	verified on 2026-09-14	Recheck within 180 days
Interview ground rules, attribution terms, and quote handling should be agreed explicitly rather than assumed.	Reuters Handbook interview guidance	verified on 2026-09-14	Recheck within 180 days
Helpful, people-first content should demonstrate first-hand experience where relevant.	Google Search Central helpful content guidance	verified on 2026-09-14	Recheck within 90 days or if guidance changes
The 25-question bank, red-flag answers, and worked example are our own editorial artifacts.	Our own, dated 2026-09-14	disclosed	Present as workflow advice, not measured outcomes

FAQ

What questions should you ask a subject matter expert when writing an article?

Ask five kinds, in order: reader-task framing questions, recent-instance experience questions ("walk me through the last time"), evidence questions that produce numbers and artifacts, caveat questions that capture boundaries, and permission questions that settle attribution. Skip any question whose answer has no section to land in.

How many questions should I prepare for an SME interview?

Prepare the full bank but plan to ask twelve to fifteen in an hour. Depth beats coverage: a fully-specified story with numbers and permissions outperforms twenty-five skimmed answers. The run sheet's exit conditions decide what gets dropped, not the clock alone.

Should I send the questions to the expert in advance?

Send the reader task and the question groups, not the verbatim list. Rehearsed answers converge on existing talking points; the publishable specifics come from live retrieval plus follow-ups. The exception is the permission group — flagging those in advance speeds up employer sign-off.

What if the SME only gives high-level answers?

Anchor on recency and repeat: "the last time this happened, what did you actually do first?" If three episode-anchored attempts still produce generalities, capture their judgment as labeled opinion and find a second source for evidence — do not dress the generalities up as experience.

Are these the same questions as a user research interview?

They share a foundation — episode anchoring, no leading questions — but diverge on purpose: user research questions explore behavior, while these retrieve publishable material, so evidence precision and quote permissions get their own question groups that research interviews do not need.

Where this fits in the interview pipeline

Questions are the middle of the pipeline. Upstream, find the right practitioner and scope the article with the interview content brief so you know which sections need feeding. Run the hour against the interview run sheet, which sequences these questions and holds the exit conditions. Afterwards, estimate how many articles the answers support with the interview yield worksheet, and clear every quote through the permissions and fact-checking workflow. The standard the finished page must meet is what experience means when you are not a famous expert.

Sources and last-reviewed notes

Last reviewed: 2026-09-14. Sources checked: Nielsen Norman Group guidance on interviewing users and running user interviews, the Reuters Handbook guidance on interviews, and Google Search Central guidance on helpful, people-first content. The section-mapped question bank, red-flag answers, and the worked example are our own editorial artifacts; the example is a labeled hypothetical and no real expert, company, or interview is referenced.