---
title: "AI Customer Interview Examples: 12 Real Scripts and Prompts for 2026"
date: "2026-06-17"
description: "AI customer interview examples are copy-ready scripts and opening prompts that tell an AI interviewer how to start a conversation, what to listen for, and which follow-ups to fire when a customer says something vague."
keywords: ["ai customer interview examples", "ai interview script", "customer interview template"]
author: "Perspective AI Team"
category: "AI Conversations at Scale"
slug: "ai-customer-interview-examples-2026-real-scripts-and-prompts"
excerpt: "AI customer interview examples are copy-ready scripts and opening prompts that tell an AI interviewer how to start a conversation, what to listen for, and…"
image: "/images/blog/0a501cfb-ee87-49ae-ac62-e4d418e6de4f.png"
tags: ["product management", "guides", "customer research", "ai interview script", "how-to"]
lastModified: "2026-06-17"
definition: "AI customer interview examples are copy-ready scripts and opening prompts that tell an AI interviewer how to start a conversation, what to listen for, and which follow-ups to fire when a customer says something vague. The 12 scripts below are organized by use case — churn, product-market fit, win-loss, onboarding, pricing, and concept testing — and each includes the goal, a sample opener, key follow-up prompts, and what a good transcript looks like. The strongest scripts share three traits: they ask about specific past behavior instead of hypotheticals (the core principle of Rob Fitzpatrick's The Mom Test), they instruct the AI to probe every vague answer once, and they end with a single forward-looking question. Sean Ellis's 40% product-market-fit benchmark and the Van Westendorp pricing model both translate cleanly into AI-moderated formats. Perspective AI runs these scripts as live AI interviews that follow up automatically, so you can field hundreds of conversations in parallel. The difference between a useful interview and a wasted one is almost never the topic — it's whether the script tells the AI to dig."
faqs: [{"question": "What is an AI customer interview example?", "answer": "An AI customer interview example is a copy-ready script or opening prompt that instructs an AI interviewer how to conduct a specific type of customer conversation. It typically includes a goal, a sample opener anchored in the customer's real experience, follow-up prompts the AI fires based on the answer, and a description of what a useful transcript looks like. The 12 examples in this guide cover churn, product-market fit, win-loss, onboarding, pricing, and concept testing."}, {"question": "How do I write a good AI interview script?", "answer": "Write a good AI interview script by anchoring the opener in a specific past event, not a hypothetical. Ask \"Walk me through the last time you did X\" instead of \"Would you use X,\" because past behavior is factual and hypotheticals are not. Then give the AI explicit follow-up logic — at minimum, instruct it to probe any vague answer once with \"What specifically?\" — and end with a single forward-looking question. Keep each script focused on one decision."}, {"question": "How is an AI interview different from a survey?", "answer": "An AI interview adapts to each answer in real time, while a survey collects fixed responses and stops. When a customer gives a vague or surprising answer, an AI interviewer follows up to capture the \"why,\" whereas a form records the shallow answer and moves on. This is why AI interviews surface root causes — like the real reason behind a churn or a lost deal — that survey dropdowns miss entirely."}, {"question": "How many AI customer interviews should I run?", "answer": "Run as many AI customer interviews as it takes to see patterns repeat, which is usually more than the five users traditional budgets allowed. Because AI interviews field in parallel at near-zero marginal cost, teams commonly run 50 to several hundred per study to confirm a finding is a real pattern and not noise. For a product-market-fit read using the 40% test, aim for a statistically meaningful sample of active users."}, {"question": "Can AI interviews replace human-moderated research?", "answer": "AI interviews replace human moderation for most structured, scalable qualitative research — discovery, churn, onboarding, pricing, and concept testing — where consistency and volume matter more than rapport with one high-stakes participant. Human moderators still add value for sensitive executive interviews or highly exploratory studies. For the bulk of recurring research, AI-moderated interviews remove the scheduling and synthesis bottleneck that kept teams from doing research consistently."}]
---

## TL;DR

AI customer interview examples are copy-ready scripts and opening prompts that tell an AI interviewer how to start a conversation, what to listen for, and which follow-ups to fire when a customer says something vague. The 12 scripts below are organized by use case — churn, product-market fit, win-loss, onboarding, pricing, and concept testing — and each includes the goal, a sample opener, key follow-up prompts, and what a good transcript looks like. The strongest scripts share three traits: they ask about specific past behavior instead of hypotheticals (the core principle of Rob Fitzpatrick's *The Mom Test*), they instruct the AI to probe every vague answer once, and they end with a single forward-looking question. Sean Ellis's 40% product-market-fit benchmark and the Van Westendorp pricing model both translate cleanly into AI-moderated formats. Perspective AI runs these scripts as live AI interviews that follow up automatically, so you can field hundreds of conversations in parallel. The difference between a useful interview and a wasted one is almost never the topic — it's whether the script tells the AI to dig.

## What makes a good AI customer interview script?

A good AI customer interview script gives the AI three things: a clear goal, an opener anchored in the customer's real experience, and explicit follow-up logic for when answers go shallow. Unlike a survey, which collects whatever the respondent types and stops, an AI interview script is a set of instructions for a conversation that adapts in real time.

The single most important rule, drawn from [Rob Fitzpatrick's *The Mom Test*](https://www.sachinrekhi.com/p/the-mom-test-rob-fitzpatrick), is to ask about specific past events rather than hypotheticals. "Would you use this?" gets you "Sure, probably" — a hypothetical answer that is, in Fitzpatrick's words, worthless for product decisions. "Walk me through the last time you tried to solve this" gets you facts the customer cannot embellish. Every script below is built on that distinction.

The second rule is the probe. A static form takes "the pricing was confusing" at face value; a good AI interview script instructs the interviewer to ask "What specifically was confusing?" and keep going until it has a concrete example. Forms flatten customers into dropdowns, while [AI-moderated interviews follow up on vague answers](/blog/ai-moderated-interviews-how-they-work-when-to-use) the way a skilled human researcher would. For pacing and depth before you copy scripts, see the [mechanics of good AI interviewing](/blog/ai-moderated-interviews-the-mechanics-of-good-ai-interviewing-in-2026).

This guide is the copy-ready companion to a full bank of [customer interview questions that get honest answers](/blog/customer-interview-questions-2026-that-get-honest-answers) and a stage-by-stage set of [product discovery questions to ask at every stage](/blog/product-discovery-questions-2026-what-to-ask-every-stage). Where those give you questions to assemble yourself, the scripts below are complete openers and probe sequences you can drop into an AI interviewer today.

Each of the 12 examples below follows the same four-part structure — **Goal**, **Sample opener**, **Key follow-up prompts**, and **What good output looks like** — so you can adapt it fast.

## Churn interview scripts: why customers leave

Churn interview scripts surface the real reason a customer left, which is rarely the reason they give first. Cancellation surveys hand you a dropdown; a churn interview gets the story behind the click. Because churned customers have nothing left to lose, exit conversations are among the highest-ROI research you can run.

**Script 1 — The recently churned customer**

- **Goal:** Find the root cause behind a cancellation, not the surface label.
- **Sample opener:** "Thanks for the time. I'd rather not sell you anything — I just want to understand your experience. Can you take me back to the day you decided to cancel? What was actually happening that week?"
- **Key follow-up prompts:** "What were you originally hoping the product would do for you?" → "When did you first feel it wasn't going to do that?" → "What did you switch to, or what are you doing instead now?" → "If we had fixed one thing, would you have stayed — and which thing?"
- **What good output looks like:** A timeline. Great transcripts pinpoint the moment value broke (a failed onboarding, a missing integration, a champion leaving) rather than a vague "it wasn't worth it."

**Script 2 — The silent / passive churner**

- **Goal:** Understand customers who quietly stopped using the product before they formally canceled.
- **Sample opener:** "I noticed you hadn't logged in for a while before your subscription ended. No judgment at all — I'm curious what changed. What were you using us for back when you were active?"
- **Key follow-up prompts:** "What pulled your attention away?" → "Was there a specific moment it stopped fitting your workflow?" → "What would have had to be true for you to keep using it?"
- **What good output looks like:** The disengagement trigger. The most valuable answers reveal that churn was a lagging indicator — the [conversational signals that beat usage data](/blog/at-risk-customer-identification-the-conversational-signals-that-beat-usage-data-alone) appeared weeks earlier.

To turn these scripts into a repeatable program, pair them with an [early churn warning signals playbook](/blog/early-churn-warning-signals-2026-catch-at-risk-customers-before-they-leave) so you interview at-risk accounts *before* they leave, and use a ready-made [churn interview template](/templates/churn-interview) as your starting outline.

## Product-market fit interview scripts: validate real demand

Product-market-fit interview scripts test whether customers would genuinely miss your product, using Sean Ellis's benchmark as the spine of the conversation. Ellis, after analyzing 100+ startups, found that companies where 40% or more of users said they'd be "very disappointed" without the product showed strong, sustainable growth — the origin of the [40% PMF test](https://learningloop.io/glossary/sean-ellis-score). An AI interview turns that survey item into a conversation that explains *why*.

**Script 3 — The Sean Ellis PMF deep-dive**

- **Goal:** Quantify PMF and capture the qualitative reason behind the score.
- **Sample opener:** "How would you feel if you could no longer use this product — very disappointed, somewhat disappointed, or not disappointed?"
- **Key follow-up prompts:** "What's the main benefit you'd miss?" → "What type of person do you think would benefit most from this?" → "What's the closest alternative you'd switch to, and why is it worse?"
- **What good output looks like:** A clean "very disappointed" rate plus a recurring benefit phrase across transcripts. When 40%+ say "very disappointed" *and* describe the same core benefit in their own words, you have a defensible PMF signal.

**Script 4 — The Mom Test discovery interview**

- **Goal:** Validate that the problem is real and painful before committing to a build.
- **Sample opener:** "Tell me about the last time you ran into [the problem]. What happened, start to finish?"
- **Key follow-up prompts:** "What did you do to solve it?" → "How much time or money did that cost you?" → "What did you try before that, and why did you stop?"
- **What good output looks like:** Evidence of past spending, workarounds, or hacks. Talk is cheap; a customer who already cobbled together a spreadsheet to solve the problem is showing you real demand. For the full method, see [the AI-first approach to jobs-to-be-done interviews](/blog/jobs-to-be-done-interviews-the-ai-first-approach-to-running-jtbd-research-at-scale) and the [product-market-fit research methodology stack](/blog/product-market-fit-research-the-2026-methodology-stack-for-pre-pmf-teams).

## Win-loss interview scripts: why deals close or don't

Win-loss interview scripts uncover the real decision criteria behind a closed deal, which the CRM "closed reason" field almost always gets wrong. Most win-loss programs only talk to the deals that pick up the phone, biasing the data toward your happiest and angriest customers; an AI interviewer reaches the quiet middle at scale.

**Script 5 — The closed-won interview**

- **Goal:** Identify what actually tipped the decision in your favor so sales can repeat it.
- **Sample opener:** "Congrats on getting started with us. I'm curious — when you were evaluating options, what made you ultimately pick us over the others you looked at?"
- **Key follow-up prompts:** "Who else was involved in the decision, and what mattered to them?" → "Was there a moment you almost went a different direction?" → "What nearly killed the deal?"
- **What good output looks like:** A specific differentiator named in the buyer's language, plus the hidden objection that almost lost the deal.

**Script 6 — The closed-lost interview**

- **Goal:** Learn why a qualified prospect chose someone else, without sounding defensive.
- **Sample opener:** "Thanks for considering us. We didn't end up being the right fit this time, and I'm genuinely trying to learn — what made the other option the better choice for you?"
- **Key follow-up prompts:** "What did they offer that we didn't?" → "Was it price, features, timing, or something about the process?" → "Is there anything that would make you reconsider us down the road?"
- **What good output looks like:** A pattern across lost deals — a missing capability, a pricing gap, or a sales-process stumble. Build the program with the [AI win-loss approach to why deals close or don't](/blog/win-loss-interviews-how-ai-uncovers-why-deals-really-close-or-don-t) and a structured [win-loss interview template](/templates/win-loss-interview).

## Onboarding interview scripts: where new users get stuck

Onboarding interview scripts capture the friction new users hit in their first sessions, while the experience is still fresh. The classic mistake is the onboarding survey that asks "How's it going?" before the user has done anything worth reporting. A conversation timed to a real milestone gets a real answer.

**Script 7 — The first-week activation interview**

- **Goal:** Find the exact step where new users stall before reaching first value.
- **Sample opener:** "You signed up about a week ago — I'd love to hear how the first few days went. Can you walk me through what you did right after you created your account?"
- **Key follow-up prompts:** "Where did you get stuck or have to stop and figure something out?" → "Was there a point you almost gave up?" → "What were you hoping to accomplish, and did you get there?"
- **What good output looks like:** A named drop-off step. Strong transcripts let you map friction to a specific screen or action, not a generic "onboarding could be smoother."

**Script 8 — The aha-moment interview**

- **Goal:** Identify the moment a user first felt the product was worth it, so you can engineer activation around it.
- **Sample opener:** "Was there a moment where this finally 'clicked' for you — where you thought, okay, I get why this is useful? Tell me about it."
- **Key follow-up prompts:** "What were you doing right before that moment?" → "How long did it take you to get there from signup?" → "What almost stopped you from reaching it?"
- **What good output looks like:** A consistent activation event you can move earlier in the flow. Operationalize the findings with a [client onboarding template](/templates/client-onboarding) and the broader [continuous discovery habits framework](/blog/continuous-discovery-habits-in-2026-operationalizing-teresa-torres-s-framework-with-ai-conversations).

## Pricing interview scripts: willingness to pay

Pricing interview scripts measure willingness to pay and the perceived value behind it, rather than asking the useless question "What would you pay?" Direct price questions invite anchoring and people-pleasing. The Van Westendorp Price Sensitivity Meter sidesteps this with four indirect questions, and an AI interviewer can deliver them conversationally while probing the reasoning.

**Script 9 — The Van Westendorp pricing interview**

- **Goal:** Establish an acceptable price range grounded in perceived value.
- **Sample opener:** "I want to understand how you think about the value here. At what price would this start to feel so expensive you wouldn't consider it?"
- **Key follow-up prompts:** "At what price would it feel like a great deal — almost too cheap to be good?" → "At what price does it start feeling expensive but still worth it?" → "What is it about the product that justifies that number for you?"
- **What good output looks like:** Four price points per respondent, plus the value driver behind each. The qualitative "why" is what makes AI Van Westendorp better than the survey version.

**Script 10 — The value-metric interview**

- **Goal:** Discover which unit of value customers want to be billed on (seats, usage, outcomes).
- **Sample opener:** "If you were designing our pricing, what would feel fair to pay *for* — per person, per project, per result? Why that?"
- **Key follow-up prompts:** "When does our current pricing feel unfair to you?" → "What do your competitors' or alternatives' pricing models get wrong?" → "What would make you happily pay more?"
- **What good output looks like:** A value metric that scales with the customer's success. Pair this with a [pricing research interview template](/templates/pricing-research-interview) to keep the structure consistent across segments.

## Concept and feature testing scripts: validate before you build

Concept testing scripts validate a new idea or feature against comprehension, relevance, and intent before engineering invests. Surveys tell you *what* people think; running 5 to 10 qualitative interviews alongside every concept test tells you *why* — and the why is what saves you from building the wrong thing.

**Script 11 — The concept reaction interview**

- **Goal:** Test whether a new concept is understood, relevant, and wanted.
- **Sample opener:** "I'm going to describe an idea we're considering, and I want your honest gut reaction. [Describe concept in one sentence.] In your own words, what do you think this does?"
- **Key follow-up prompts:** "Who is this for, and is that you?" → "What problem would this solve in your world — if any?" → "What's confusing or missing?" → "What would you expect to pay for it?"
- **What good output looks like:** Accurate playback of the concept plus an unprompted use case. If customers can't restate the idea or can't name a problem it solves, the concept isn't ready. See the [AI concept testing approach to validating ideas in hours](/blog/ai-concept-testing-2026-validate-ideas-in-hours-not-weeks).

**Script 12 — The feature prioritization interview**

- **Goal:** Rank competing roadmap items by real pain, not the loudest request.
- **Sample opener:** "I'm going to mention a few things we're thinking about building. For each one, tell me whether it would actually change your day — and be brutally honest if it wouldn't."
- **Key follow-up prompts:** "What are you doing today to work around the lack of this?" → "If you could only have one of these in the next quarter, which and why?" → "What's missing from this list that matters more?"
- **What good output looks like:** A pain-ranked list grounded in current workarounds. This is the antidote to the [feature-voting board that quietly makes roadmaps worse](/blog/feature-voting-boards-are-quietly-making-your-roadmap-worse). Start from a [feature prioritization interview template](/templates/feature-prioritization-interview) and the [AI customer research approach to ranking the roadmap](/blog/feature-prioritization-framework-using-ai-customer-research-to-rank-the-roadmap).

## How to run these scripts at scale with AI

Running these scripts at scale means handing the script to an AI interviewer that delivers the opener, fires the follow-ups, and probes vague answers across hundreds of conversations at once. The bottleneck in qualitative research was never the questions — it was the researcher's calendar. The [Nielsen Norman Group's guidance](https://www.nngroup.com/articles/why-you-only-need-to-test-with-5-users/) that roughly five users surface most usability issues was a constraint born of cost; when interviews cost almost nothing to field, you can talk to 50 or 500 and confirm the patterns hold.

The workflow is the same regardless of script: paste the goal and opener into the interviewer, set the follow-up logic, and let it run. [Perspective AI's interviewer agent](/agents/interviewer) handles the probing automatically, then [turns hours of transcripts into decisions](/blog/ai-interview-analysis-turning-hours-of-transcripts-into-decisions) with quote extraction and pattern synthesis. For a complete operating procedure, the [playbook for running AI-moderated customer interviews](/blog/how-to-run-ai-moderated-customer-interviews-2026-playbook) walks through setup, distribution, and analysis end to end.

## Frequently Asked Questions

### What is an AI customer interview example?

An AI customer interview example is a copy-ready script or opening prompt that instructs an AI interviewer how to conduct a specific type of customer conversation. It typically includes a goal, a sample opener anchored in the customer's real experience, follow-up prompts the AI fires based on the answer, and a description of what a useful transcript looks like. The 12 examples in this guide cover churn, product-market fit, win-loss, onboarding, pricing, and concept testing.

### How do I write a good AI interview script?

Write a good AI interview script by anchoring the opener in a specific past event, not a hypothetical. Ask "Walk me through the last time you did X" instead of "Would you use X," because past behavior is factual and hypotheticals are not. Then give the AI explicit follow-up logic — at minimum, instruct it to probe any vague answer once with "What specifically?" — and end with a single forward-looking question. Keep each script focused on one decision.

### How is an AI interview different from a survey?

An AI interview adapts to each answer in real time, while a survey collects fixed responses and stops. When a customer gives a vague or surprising answer, an AI interviewer follows up to capture the "why," whereas a form records the shallow answer and moves on. This is why AI interviews surface root causes — like the real reason behind a churn or a lost deal — that survey dropdowns miss entirely.

### How many AI customer interviews should I run?

Run as many AI customer interviews as it takes to see patterns repeat, which is usually more than the five users traditional budgets allowed. Because AI interviews field in parallel at near-zero marginal cost, teams commonly run 50 to several hundred per study to confirm a finding is a real pattern and not noise. For a product-market-fit read using the 40% test, aim for a statistically meaningful sample of active users.

### Can AI interviews replace human-moderated research?

AI interviews replace human moderation for most structured, scalable qualitative research — discovery, churn, onboarding, pricing, and concept testing — where consistency and volume matter more than rapport with one high-stakes participant. Human moderators still add value for sensitive executive interviews or highly exploratory studies. For the bulk of recurring research, AI-moderated interviews remove the scheduling and synthesis bottleneck that kept teams from doing research consistently.

## Conclusion

These 12 AI customer interview examples share one DNA strand: they ask about real, specific behavior and they always probe the vague answer. Whether you're running a churn exit conversation, a Sean Ellis product-market-fit deep-dive, a closed-lost win-loss interview, or a Van Westendorp pricing test, the script's job is the same — give the AI a clear goal, an opener grounded in the customer's actual experience, and the follow-up logic to dig one level deeper than a form ever could. That's the whole difference between an interview that informs a decision and one that just fills a spreadsheet.

The fastest way to put these AI customer interview examples to work is to run them as live conversations instead of static questions. [Perspective AI](/research/new) fields any of these scripts as an AI-moderated interview that follows up automatically across hundreds of customers at once — so you spend your time reading insights, not chasing calendars. Pick the script that matches your next decision, drop in the opener, and start listening.