Customer Discovery Call: Questions, Structure, and a Template

A customer discovery call is a focused, one-on-one conversation whose only goal is to learn how a customer thinks, works, and struggles — before you pitch a solution or build anything. It is not a demo, not a sales call, and not a survey read aloud. The output is not a yes; it is understanding: the customer''s real problem, how they solve it today, what that costs them, and whether it is worth solving at all.

Get this call right and everything downstream — positioning, roadmap, pricing, messaging — rests on evidence instead of guesswork. Get it wrong, and you collect polite encouragement that feels like validation and predicts nothing.

This guide gives you the structure of a great discovery call, a copy-ready question script, the biggest mistakes to avoid, and how AI-native platforms like Koji let you run the equivalent of dozens of discovery calls in parallel.

What a discovery call is (and is not)

The cardinal rule, popularized by Rob Fitzpatrick''s The Mom Test, is simple: talk about the customer''s life, not your idea. People lie to be nice when you ask about your idea ("Sure, I''d use that!"). They tell the truth when you ask about their past behavior, specific problems, and what they have actually done about them.

So a discovery call avoids three traps:

• Pitching — the moment you describe your solution, you stop learning.
• Hypotheticals — "Would you pay for X?" produces fantasy answers. Ask what they pay for today.
• Compliments — "That sounds great" is data about politeness, not demand.

The structure of a strong discovery call

A 30-minute discovery call has five phases:

1. Warm-up (2–3 min). Build rapport, set expectations, get permission to record. "I''m trying to understand how people like you handle X — I''m not selling anything today."
2. Context (5 min). Their role, their goals, their world. You cannot interpret a problem without understanding the person who has it.
3. Problem exploration (10–12 min). The heart of the call. How do they handle the relevant task today? Where does it break? What have they tried?
4. Impact and motivation (5–7 min). What does the problem cost — time, money, stress? Have they looked for a solution or paid for one? This separates "nice to have" from "hair on fire."
5. Wrap-up (3 min). Ask what you missed, get permission to follow up, and request an introduction to others like them.

A discovery call question script

Use these as a backbone, and probe every answer with a follow-up — the second and third question are where the insight lives.

Context

• Tell me about your role — what are you actually responsible for?
• Walk me through a typical week as it relates to [domain].

Problem (past behavior, not hypotheticals)

• How do you handle [task] today? Walk me through the last time you did it.
• What''s the hardest or most annoying part of that?
• What have you tried to make it better? What happened?

Impact

• When [problem] happens, what does it cost you — in time, money, or hassle?
• How often does it come up?
• Have you ever paid for anything to solve this? What?

Motivation and priority

• On a scale of 1–10, how big a deal is this compared to everything else on your plate?
• If a perfect solution existed, what would it have to do for you to switch?

Wrap-up

• What should I have asked you that I didn''t?
• Who else do you know who deals with this that I should talk to?

Map your script to structured question types

A discovery call is mostly qualitative, but a few quantitative anchors make findings comparable across many calls. Koji supports six structured question types — open_ended, scale, single_choice, multiple_choice, ranking, and yes_no:

• open_ended for the problem and impact narrative, with AI follow-up probing.
• scale for the 1–10 severity question, so you can rank pains across all your calls.
• yes_no for "Have you paid to solve this before?" — a sharp demand signal.
• single_choice to categorize which workaround they use today.
• ranking when you want them to order competing pains.

See the structured questions guide for how each type rolls up into a report.

Common discovery call mistakes

• Leading the witness. "Don''t you hate when X happens?" plants the answer. Ask "What''s frustrating about your current process?" instead.
• Pitching too early. Save your solution for the very end, if at all — once you pitch, the learning stops.
• Accepting the first answer. "It''s fine" is not an answer; it''s the start of one. Probe.
• Talking more than the customer. Aim for an 80/20 split — they talk 80% of the time.
• No notes on exact words. Verbatim quotes are gold; paraphrases lose the signal.

How Koji scales discovery calls

The hard limit on manual discovery is time. Founders and PMs know they should talk to 20 customers; they manage five, because every call is a scheduling dance plus 30 minutes plus note-writing plus synthesis. Koji removes that ceiling. Its AI interviewer runs discovery conversations — by voice or text, in any language, with no moderator — and automatically asks follow-up questions when an answer is shallow, applying Mom Test-style probing on every single conversation.

Because interviews run in parallel, talking to 30 customers takes about as long as talking to one. Koji then analyzes every transcript automatically, clustering pains into themes, surfacing representative verbatim quotes, and aggregating your scale and yes/no anchors into charts — so you see which problem is "hair on fire" across your whole sample, not just the call you happen to remember best. It is the depth of a one-on-one discovery call with the breadth of a survey, delivered as a shareable report in hours instead of weeks.

How many discovery calls should you run?

The honest answer is "until you stop being surprised." Most founders and product teams reach a useful signal somewhere between 15 and 30 discovery calls — enough that the same pains, workarounds, and language start repeating across different people. The danger is stopping too early: after three or four enthusiastic calls it is tempting to declare validation, but small samples are dominated by whoever happened to say yes to a call, who are rarely representative of the market.

Reaching 20+ genuine discovery conversations manually is a real project — each is a scheduling negotiation, a 30-minute call, and 20 minutes of note-writing, easily a full day of work per five calls. That cost is exactly why so many teams cut discovery short and pay for it later in a mis-built product. Running discovery conversations asynchronously with AI removes the scheduling tax and lets you hit a representative sample without the calendar grind.

What to do with the recording

The single highest-leverage habit in discovery is capturing verbatim quotes, not paraphrases. "The exports take me an hour every Friday and I dread it" is far more useful than "user mentioned export pain" — it carries the emotion, the frequency, and the language you will reuse in your positioning. Tag each quote by theme, count how often each pain recurs, and let the patterns — not the most memorable single call — drive your decision. Koji does this synthesis automatically, transcribing every conversation, clustering pains into themes, and surfacing the quotes worth putting in front of your team.

Related Resources

• Structured Questions Guide — the six question types and how each aggregates
• The Mom Test at Scale — applying Mom Test principles with AI interviews
• Customer Discovery Interviews: The Complete Guide — the broader discovery methodology
• The 7-Day Customer Discovery Sprint — a fast discovery plan for founders
• Requirements Gathering Interviews — turning discovery into documented needs
• Customer Interview Questions: 60+ Examples — a deeper question bank