# Customer interview script for ad research

> A 25-minute customer interview script for ad research. Fifteen questions surface triggers, doubts and the exact words buyers use, so you can turn them into hooks.
- **Author**: Marxx
- **Published**: 2026-10-31
- **Category**: Research
- **URL**: https://marxx.ai/library-templates/customer-interview-script-for-ad-research

---

A customer interview for ad research is a 20 to 30 minute chat about the last time someone bought, not about what they might buy one day. Use the script below, then pull out the exact words. Those words become your hooks.

## Who to talk to

Talk to five to eight recent buyers. If you can, add one or two people who looked and did not buy. Offer a small thank-you, such as a INR 300 voucher, that does not depend on what they say. Ask before you record.

## The script

```
INTRO (read aloud)
"Thanks for your time. I am not selling anything. I want to learn how
you chose [product]. There are no wrong answers. Is it OK if I record?
I will only use what you say to improve our ads and our product."

WARM-UP
1. Tell me a bit about yourself and your day.

THE TRIGGER
2. Think back to when you first started looking for [category].
   What was happening in your life then?
3. What had you tried before? What went wrong with it?

THE SEARCH
4. Where did you look? Who did you ask?
5. What did you compare? What made you cross some options off?

THE DECISION
6. When did you decide? What was the last thing that tipped it?
7. What almost stopped you? What worried you?
8. What would have made you say no?

AFTER BUYING
9. What was the first week like?
10. What surprised you, good or bad?
11. Who did you tell, and what did you say?

THE WORDS
12. How would you describe it to a friend in one sentence?
13. If it disappeared tomorrow, what would you miss?

CLOSE
14. What should I have asked that I did not?
15. Who else should I talk to?
```

## What to write down

Make a sheet with one row per remark. Tag each row as one of five types.

| Type | What it is | Where it goes |
|---|---|---|
| Trigger | The event that started the search | Call-out hooks |
| Objection | What almost stopped them | Proof and FAQ lines |
| Alternative | What they used before | Contrast hooks |
| Word | A phrase they use | Hook and headline wording |
| Proof | What convinced them | Proof in the ad |

Say a buyer of a INR 1,499 water bottle tells you she bought it after her old bottle leaked in her bag. The trigger is the leak. The use is carrying it with a laptop. Her word is "leak-proof". Two hooks from that: "Your bottle should not be the reason your laptop is wet." and "I tested it upside down in my bag." Use the second only if someone did.

## Rules for a good interview

- Ask about the past. "Would you buy this?" gets polite answers. "What did you do last time?" gets facts.
- Do not lead. Ask "what worried you?" not "was the price a worry?".
- Stay quiet after an answer. People add the useful part when you wait.
- Do not pitch. If you start selling, you stop learning.
- Stop when you hear the same reasons again and again.

When you cannot reach enough buyers, Marxx builds personas from real ad comments and app-store reviews. That gives you words to start with. It does not replace a conversation. See [audience personas](/features/audience-personas).

Turn the sheet into ideas with [how to come up with ad angles](/guides/how-to-come-up-with-ad-angles), then pick patterns from the [ad hook examples](/templates/ad-hook-examples) and [how to write ad hooks](/guides/how-to-write-ad-hooks).

## FAQ

### How many interviews do I need?

Start with five. Stop when new interviews stop adding new reasons to buy.

### Can I interview over WhatsApp?

A call works best, because you hear the pauses and can ask a follow-up. A voice note or chat works when someone cannot talk, but you lose some of that.

### Should I pay people?

A small thank-you is fair. Keep it the same for everyone, so it does not shape what they say.

---
- [More Research articles](https://marxx.ai/library-templates/category/research)
- [All articles](https://marxx.ai/library-templates)