Expert content without writing: a 20-minute interview
Expert content without writing: a 20-minute interview with five questions becomes a blog post, a LinkedIn post and a page section; the expert checks only facts.
How can an expert at a service firm publish if they have no time to write? They don’t write. They give 20 minutes of conversation with five questions on one topic. The person in charge of marketing records the conversation and turns it into a blog post, a LinkedIn post and a section of a service page. The expert checks only the substance: facts, numbers and examples, not the style. Their voice stays in quotes from the recording. To keep this going through busy months, each expert has one conversation a month, booked in advance.
In this post I’ll show how to prepare the conversation, which questions to ask, what comes out of it, and how to split the work between the expert and the person who writes.
Buyers read vendor content, and experts are hard to reach
In the Content Marketing Institute’s research for 2025 (980 B2B marketers, mostly from North America, including agencies, professional services and consulting), 33% of respondents said accessing subject matter experts was a challenge. The most often cited challenge not related to content creation was lack of resources (54%), and only one in three say they have a scalable model for content creation.
At the same time, buyers still read what vendors publish. According to 6sense’s 2025 study, 94% of B2B buyers use language models, but they still rely on vendor content and third-party experts. In my view, the most valuable content at a service firm is the content that shows the knowledge of a specific person on the team. Yet that person has the least time, because they run projects and talk to clients.
The conversation: 20 minutes, one topic, five questions
It’s easier for an expert to find 20 minutes in their calendar than two evenings for writing. For those minutes to be enough, the conversation covers one topic, picked from the firm’s list of topics. I described how to build that list from buyers’ questions in the post on content for service firms.
The person running the conversation doesn’t have to be an expert. In 2021, the Content Marketing Institute (CMI) advised that they should know enough about the topic and the expert to ask the necessary questions, and ask for an explanation when they don’t understand something. The conversation is recorded with the expert’s permission and then transcribed, by hand or with a transcription tool (“How to Woo and Win Subject Matter Experts (Almost) Every Time”).
I recommend these five questions:
- What client problem do you see most often in this area, and how do you recognize it?
- What do clients usually do first, and why doesn’t it work?
- How did you solve it in your last project, step by step?
- When does this approach not work, or what would you not recommend?
- Is there anything I haven’t asked about that would help a client?
I take the last question from CMI, which advises ending the conversation with an open-ended question. The third question gives you an example and the fourth gives you the limits, two things that in my view are most often missing from content written without an expert. Send the expert the topic and the questions before the conversation, so they can recall the project.
One conversation, a blog post, a LinkedIn post and a service page section
A 20-minute recording usually gives you three things:
- a blog post: the problem, the approach, an example from a project and the limits, meaning the answers to questions one to four,
- a LinkedIn post: one idea from the conversation, ideally a sentence from the expert that surprises or contradicts what clients usually do,
- a service page section: an example from a project or a description of when the service doesn’t fit.
The blog post comes first, because it has the most content. The LinkedIn post and the service page section take ready-made sentences from it, so the expert checks them together with the post.
The expert checks substance, not style
The expert gets the finished text with marked sentences and, for each one, answers three questions: are the facts and numbers right, is the example described the way it happened in the project, and is this how they would say it. They don’t fix commas, word order or headings. The person who writes is responsible for style.
To keep the check to a few minutes, mark the sentences that need confirmation in the text: numbers, names, how the project went and quotes from the recording, for example with comments in the document. If the text goes out under the expert’s name and in the first person, as a LinkedIn post usually does, the expert also reads it once in full, still without editing the style. You don’t publish the client’s name or project results without the client’s permission, even if they came up in the conversation.
The expert’s voice stays in their own sentences
The expert’s voice disappears fastest when the person writing rewrites their sentences to make them “nicer”. I recommend keeping two or three sentences from the recording almost word for word, including the colloquial ones, along with their examples, even if they’re not very impressive.
A rhythm that survives a busy month
You book conversations in advance, for example for the whole quarter, one a month for each expert. When an expert has their worst week on a project, the conversation moves by a few days but doesn’t drop. With two or three experts, that gives you two or three blog posts a month, and each of them spends less than an hour a month on it: 20 minutes of conversation, checking the marked sentences and reading the texts under their name. It’s also worth keeping one recorded conversation in reserve that hasn’t been used yet.
Example: a conversation with an expert at a software house
This is an illustration, not a real firm: take a software house where an architect has been running migrations of old systems to the cloud for years but has never published anything. The topic from the firm’s list: “why cloud migration costs more than the client expected”.
| Question | What it covers in this topic |
|---|---|
| 1. Problem | how the architect recognizes that the client got the costs wrong |
| 2. What clients do first | they move the system unchanged and are surprised by the bills |
| 3. How they solved it in the last project | the order of steps, without the client’s name (no permission) |
| 4. When the approach doesn’t work | systems not worth moving |
| 5. Anything we missed | a topic for the next conversation |
The recording becomes a blog post, a LinkedIn post with one of the architect’s sentences about the bills, and a paragraph on the migration service page about when migration doesn’t make sense. The architect checks the marked sentences about costs and the order of steps, and reads the LinkedIn post, which goes out from their profile, in full. The blog post goes out under the firm’s name, with a note that it is based on a conversation with the architect.
Frequently asked questions
Is it honest to put the expert’s name on the text?
In my view, yes, if the substance comes from them and they read the whole text before publication. If they don’t want to sign it, you can say the text is based on a conversation with them.
What if the expert doesn’t want to talk?
Start with one conversation and show them the finished text. In my view, what convinces them best is that the check takes little time and the text sounds like them. If they still don’t want to, pick another expert.
Can AI write the draft from the recording?
It can, if it gets the transcript and writes only from what the expert said. Every sentence from outside the recording and sources is a risk that the expert signs something they don’t claim. That’s why the expert’s check stays, no matter who writes.
Summary
If your firm’s knowledge sits in the heads of a few people who have no time to write, start with one expert and one topic. Book 20 minutes, ask five questions, write a blog post and a LinkedIn post from the recording, and give the expert the marked sentences to check, plus the texts under their name to read in full. In our open repo (files and instructions for an AI agent such as Claude Code), a blog draft may contain only claims that have a row in the sources file next to it, and each paragraph with a claim shows which sources it rests on (open source).
Every method on this blog also runs as a skill in the open-source B2BForce repo. About the repo