Turn one customer interview into five content ideas

One good customer interview holds more content than most teams ever use. The problem is time: nobody has an afternoon to reread a transcript and pull out the angles.

AI is good at exactly this kind of work: reading a long text, spotting themes, and drafting options. In an MIT experiment with 453 professionals, including marketers, access to ChatGPT cut the time spent on writing tasks by 40 percent, and quality rose by 18 percent.¹ This guide shows how to use that speed on interviews without losing the one thing that makes them valuable: the customer’s real words.

Why interviews are worth the effort

A customer interview gives you three things no brainstorm can: real problems, real language, and real proof. The phrases customers use to describe their situation are often better headlines than anything a team writes. Their stories make claims believable.

Most of that value sits unused. The transcript is filed after the project, and the team writes the next campaign from memory. A 15-minute AI pass turns one conversation into a small content plan: a blog post, a case study angle, a few social posts, and a line for the sales deck.

Before you paste anything

Start with your company’s AI rules. Many teams don’t have clear ones yet, so people make it up as they go. Microsoft and LinkedIn’s 2024 survey found that 78% of AI users bring their own AI tools to work, and only 39% have received AI training from their company.² A lot has changed since then. In McKinsey’s 2026 global survey, nearly nine in ten respondents said their organizations regularly use AI in at least one business function, and 44% said AI is scaling across the enterprise, up from 38% a year earlier.³ In the EU, training is no longer just good practice: since Feb 2, 2025, Article 4 of the AI Act has required organizations that use AI systems to take measures to support their staff’s AI literacy, a duty the EU eased in July 2026 but kept.⁴ An interview transcript is exactly the kind of material that should only go into an approved tool.

1. Check consent. Make sure the customer agreed to the interview being used for content, and note any limits.
2. Remove names and anything confidential. Replace the person’s name, company, prices, internal figures, and anything that identifies them with neutral labels such as “[Customer]” or “[Product].”
3. Clean the transcript lightly. Fix speaker labels and delete small talk, but don’t rewrite what the customer said.
4. Add context in one line. Who the customer is (role and industry, not name), what they bought, and why you interviewed them.

The prompt

Paste the cleaned transcript into your approved AI assistant and ask:

“Read this interview. List five content ideas it supports. For each, give a working headline, the customer quote that proves it (copied word for word), the audience it suits best and the format that fits (blog post, case study, social post, sales slide or email). Use only what is in the transcript. If an idea needs facts the transcript does not contain, say so.”

A good answer looks something like this: “Idea 2: ‘We stopped losing Friday afternoons to reporting.’ Quote: ‘Friday afternoons used to be gone, every week.’ Audience: operations managers at mid-size firms. Format: short case study. Missing: how many hours were saved.” The last line is the AI telling you what to ask the customer next.

The last two sentences matter. They tell the AI to stay inside your material and to admit gaps, rather than filling them in. You can add your brand voice guidelines or a list of current campaign themes, so the ideas fit what the team is already working on.

Coming soon: Prompts that work: a brief template for marketers →

What to check

Make sure every quote is real and word for word. Assistants sometimes tidy up phrasing or invent it. Keep the ideas that sound like your customer, not like the AI.

This is not just caution. OpenAI itself describes the problem plainly: “Hallucinations are plausible but false statements generated by language models.”⁵ Its researchers explain that “standard training and evaluation procedures reward guessing over acknowledging uncertainty.”⁵ In other words, a model is often rewarded for giving an answer, even when it should say “I don’t know.”

A quick way to check is to search the transcript for each quote. If the exact words don’t appear, the quote doesn’t go out. Then check the ideas themselves:

• Is it true to the interview? The idea should reflect what the customer actually meant, not a stronger version of it.
• Is it useful to a reader? A good idea answers a question other customers have.
• Is it on-brand? It should fit your positioning and current priorities.
• Is it approved? Any quote that goes public needs the customer’s OK.

From ideas to drafts

Pick the strongest idea and ask the AI for a short outline, built around the verified quote. Then write or edit the draft yourself, or with the AI, and keep the customer’s words intact. One interview can feed a month of content if you reuse it across formats: the long story for the blog, the sharpest quote for social posts, the problem-and-result line for sales.

Keep a simple log for each interview: the five ideas, which ones were used, where and when. After a few interviews, you will have a library of verified quotes and themes. That makes the next campaign faster and better grounded.

Who on the team gains most

This is a good first task for newer team members. In a study of 5,179 customer support agents, an AI assistant raised the number of issues resolved per hour by 14% on average, and by 34% for novice and low-skilled workers.⁶ AI helps people with less experience work more like the best people on the team.

It also suits busy teams. Microsoft’s 2025 Work Trend Index found that 80% of the global workforce say they lack enough time or energy to do their work.⁷ A task that used to take an afternoon of rereading now takes 15 minutes, which means the interview actually gets used.

The checking step stays with someone who knows the customer. The AI can find the angles. Only a person who was in the conversation, or who knows the account, can tell which ones are true to what the customer meant.

Be open about how the content was made where it matters. In the EU, the AI Act requires certain AI-generated content to be clearly labeled, including deepfakes and text published to inform the public on matters of public interest.⁸ A customer story edited by your team with some AI help usually doesn’t fall into those categories, but check with whoever handles legal questions if you are unsure.

Coming soon: Build your first AI content pipeline →