Interview a Subject Expert for Source Material
Content teams routinely cover subjects they do not practise, so the draft gets assembled from other people's articles and reads exactly like that. The obvious remedy, sending the finished draft to an internal expert for review, arrives too late: reviewers correct errors but never supply the material that would have made the page worth reading. The interview has to happen before the outline exists, and it has to be run for extractable specifics rather than general explanation.
CATEGORY
Content production
FORMAT
sme-interview-for-source-material.md
WHEN TO REACH FOR THIS
Use when a page needs expertise nobody on the content team has and the expert will give you an hour at most.
The skill file
sme-interview-for-source-material.md
---
name: sme-interview-for-source-material
description: Use when a page needs expertise nobody on the content team has and the expert will give you an hour at most.
---
# Interview a Subject Expert for Source Material
Content teams routinely cover subjects they do not practise, so the draft gets assembled from other people's articles and reads exactly like that. The obvious remedy, sending the finished draft to an internal expert for review, arrives too late: reviewers correct errors but never supply the material that would have made the page worth reading. The interview has to happen before the outline exists, and it has to be run for extractable specifics rather than general explanation.
## What you need first
- the reader decision the page must serve, agreed before the call
- the current top ranking pages, read in advance, so you can ask what they get wrong
- a recorded call with a transcript, with the expert's consent recorded too
- a booked 45 to 60 minute window, because experts do not give more than that
## Method
1. Send three questions ahead and no more. Experts prepare answers for what they see and improvise the rest, and the improvised answers are consistently the better material.
2. Open by asking what the ranking pages get wrong. Disagreement produces sharper and more specific speech than an open invitation to explain the topic.
3. Ask for the most recent real case: what came in, what they decided, what it cost. Concrete recall yields numbers and detail that a general question never surfaces.
4. Every time you hear a rule of thumb, ask for the exception. The exception is the part nobody else on page one has written down.
5. Push every vague quantity to a figure. "It takes a while" becomes "usually six to eight weeks, longer if the survey is queued" only if you ask a second time.
6. Close by asking what they would tell a friend to avoid. This question reliably produces the strongest paragraph of the eventual page.
7. Mark the transcript into three buckets before any drafting: verbatim quotes worth attributing, facts that still need checking, and background. Everything sounds usable in speech, so never draft straight from the raw transcript.
## What this produces
A marked-up transcript plus a shortlist of roughly ten attributable claims, figures and examples, enough to brief a writer with.
## Where this goes wrong
- Spending the hour asking the expert to explain the basics, which buys you material already present on every competing page
- Interviewing after the draft exists, so the expert edits wording and never contributes the substance that would have made the page different
- Publishing a quote without letting the speaker see it in context, which is how a fair quote turns into a complaint and a retraction
---
From the QuQi skill library - https://www.quqi.io/skills/sme-interview-for-source-material
Free to download · no account, no email
What you need first
-
the reader decision the page must serve, agreed before the call
-
the current top ranking pages, read in advance, so you can ask what they get wrong
-
a recorded call with a transcript, with the expert's consent recorded too
-
a booked 45 to 60 minute window, because experts do not give more than that
Method
-
01
Send three questions ahead and no more. Experts prepare answers for what they see and improvise the rest, and the improvised answers are consistently the better material.
-
02
Open by asking what the ranking pages get wrong. Disagreement produces sharper and more specific speech than an open invitation to explain the topic.
-
03
Ask for the most recent real case: what came in, what they decided, what it cost. Concrete recall yields numbers and detail that a general question never surfaces.
-
04
Every time you hear a rule of thumb, ask for the exception. The exception is the part nobody else on page one has written down.
-
05
Push every vague quantity to a figure. "It takes a while" becomes "usually six to eight weeks, longer if the survey is queued" only if you ask a second time.
-
06
Close by asking what they would tell a friend to avoid. This question reliably produces the strongest paragraph of the eventual page.
-
07
Mark the transcript into three buckets before any drafting: verbatim quotes worth attributing, facts that still need checking, and background. Everything sounds usable in speech, so never draft straight from the raw transcript.
What this produces
A marked-up transcript plus a shortlist of roughly ten attributable claims, figures and examples, enough to brief a writer with.
Where this goes wrong
-
Spending the hour asking the expert to explain the basics, which buys you material already present on every competing page
-
Interviewing after the draft exists, so the expert edits wording and never contributes the substance that would have made the page different
-
Publishing a quote without letting the speaker see it in context, which is how a fair quote turns into a complaint and a retraction
Use this skill in your own AI
The download is a plain markdown file with the name and trigger in its frontmatter. Where an assistant supports skills it can load itself, that frontmatter is what it reads to decide this one applies.
Claude Code
Save it as ~/.claude/skills/sme-interview-for-source-material/SKILL.md and Claude loads it on its own when what you are doing matches the trigger line. Put it in .claude/skills inside a project instead if the whole team should have it.
Claude
Upload the file in the skills section of your settings. Once it is there it applies itself in any conversation where the trigger fits, so you do not have to remember it exists.
ChatGPT
There is no skills format to install into, so paste the file contents into a Project instruction or a Custom GPT instead. It then applies to every chat in that project rather than only the one you paste it into.
Anything else
Paste the markdown into the chat before your question. It works in any assistant, it just has to be pasted again each time.
Questions about this skill
When do I interview the expert rather than send them the finished draft to review?
Whenever the page needs expertise nobody on the content team has. Review arrives too late: reviewers correct errors but never supply the material that would have made the page worth reading, so you get an accurate article assembled from other people's articles. The interview has to happen before the outline exists, and it is run for extractable specifics.
What do I need booked and prepared before the call?
The reader decision agreed in advance, the current ranking pages read so you can ask what they get wrong, a recorded call with consent recorded too, and a window of forty-five to sixty minutes, because experts rarely give more. Send three questions ahead and no more, since experts prepare answers for what they see and improvise the better material.
What do I come away with, and which part does the writer use?
A marked-up transcript plus roughly ten attributable claims, figures and worked examples, enough to brief a writer without a second call. The marking is what makes it usable: verbatim quotes worth attributing, facts still needing a check, and background. Everything sounds usable in speech, so drafting from the raw transcript is how unchecked assertions reach the page.
What is the mistake that wastes the hour?
Spending it asking the expert to explain the basics, which buys material already present on every competing page. Open on what the ranking pages get wrong, push every vague quantity to a figure, and ask for the exception to each rule of thumb. Publishing a quote without letting the speaker see it in context is how a fair quote becomes a retraction.
More in Content production