How to Run a Customer Interview Without Pitching
Founders interviewing potential buyers should ask what happened the last time the problem occurred, how it is handled today and who approved the last related purchase. Keep the product out of the conversation until those questions are answered, then ask for one small next step. Afterwards, compare what the buyer said with what they did. The main constraint is discipline: once the product is described, answers become feedback on a pitch rather than evidence.
The result this interview should produce
This guide is for founders who are about to interview potential buyers, usually businesses, before building or while a first version is still changing. It covers how to run one conversation from preparation to follow-up so that it produces evidence instead of reactions to a pitch. It assumes you already have a rough segment and a problem you suspect is real. If you still need to choose what to ask, start with the customer discovery questions that reveal buying intent.
A useful interview produces three things: a specific account of the last time the problem happened, a clear picture of how it is handled today, and a named path to whoever approves spending. It ends with one small next step that the buyer agrees to take. Compliments, feature requests and statements that the person would use a product are not results. They are reactions, and a pitch produces them easily.
Pitching is tempting because it feels productive. Once the product is described, the conversation changes: the buyer starts evaluating your idea and being polite about it, and you stop learning about their situation. The SBA guide to market research lists in-depth interviews among the direct research methods, alongside surveys, questionnaires and focus groups, and notes that direct research can be time consuming and expensive. An interview that costs that much effort is too valuable to spend on a pitch.
Before the call: prepare around a recent purchase
Preparation decides whether the interview stays on the buyer's situation. Before the call, write down the single decision the interview should inform, such as whether this segment is worth a paid test. Then prepare questions about past behavior, because past behavior is the closest thing to buying evidence an interview can produce.
- The last occurrence. When did the problem last happen, what did it cost, and what did the person do next?
- The last related purchase. What did the organization last buy to deal with this or a similar problem, who suggested it, who approved it and how long did it take?
- The artifact. Ask in advance whether they can show the spreadsheet, form, message thread or paper list they use today, with anything confidential removed.
- The rule for yourself. The product is not described until the research questions are done. Write that rule at the top of your notes.
Leave the slide deck and the demo closed. If the person asks for background when booking, describe the problem you are studying, not the solution you are building: you are researching how teams like theirs handle a specific task and want to learn from someone who does it every week.
During the call: work through the current workaround
Open with the last occurrence and let the person tell it in order. Ask what happened first, what they did next, who else got involved and how it ended. When the story turns general, ask for the most recent concrete instance instead of the usual pattern. The usual pattern is where people describe how things are supposed to work.
Then ask them to show, not describe, the current workaround. Watching someone open the spreadsheet or walk to the whiteboard reveals steps they would leave out of a summary: the copy and paste, the second check, the colleague who gets a phone call. For each step, ask what happens when it fails and when that last happened. Note what works well enough today, because a workaround that is cheap and familiar is the competitor your product has to beat.
Moments when founders start pitching, and what to do instead
Most pitching is not planned. It happens at predictable moments, and each one has a research response.
| Moment | The pitching reflex | The research response |
|---|---|---|
| The buyer asks what you are building | Describe the product and its features | Say you will share it at the end and ask to finish their story first |
| The buyer describes a pain your product solves | Explain how the product would fix it | Ask what it cost the last time and what they did about it |
| A silence after a question | Fill it with an explanation or a suggested answer | Wait, then ask what else happened |
| The buyer proposes a feature | Agree and add it to the roadmap | Ask what they do today when they need that |
| You want reassurance | Ask whether they would use or pay for it | Ask what they paid for the last time they tried to solve it |
Questions about the future, such as whether someone would use or buy something, invite a polite yes. Questions about the past are harder to answer generously, because the person either did something or did not.
Resolve the decision-maker and budget owner without selling
The person with the problem is often not the person who pays. Find out without turning the interview into a sales qualification call. Return to the last related purchase: who first brought it up, who compared the options, who signed and whether anyone outside the team had to approve it. These are questions about history, so they rarely feel like a sales tactic.
If the interviewee is a user rather than the budget owner, ask how a request like this usually reaches the person who decides, and whether that person would be willing to talk about the problem. Keep the request about research. An introduction is a useful completed action, but it shows access, not intent to buy. Record the role of each person you hear about: user, influencer, approver or budget owner.
Only after the research questions are done, and if it fits the conversation, ask permission to describe what you are working on in a few sentences. Treat the reaction as a separate, weaker kind of information and record it apart from the interview evidence.
After the call: compare what they said with what they did
Finish every interview by proposing one small next step that costs the buyer something real: sending the anonymized artifact, introducing the budget owner, joining a second session with a colleague or agreeing to a defined paid test. Then write up the interview the same day in a say-and-do evidence map. The map separates the interview finding from the buying behavior that followed it.
| Field | What to record |
|---|---|
| Last occurrence | Date, what happened and what it cost in time or money, as reported |
| Current workaround | Steps you saw demonstrated, and steps you were only told about |
| Last related purchase | What was bought, who suggested it, who approved it and how long it took |
| Stated interest | What the person said about solving the problem, kept in its own row |
| Agreed next step | The specific action and the date the buyer agreed to |
| Completed behavior | What actually happened by that date, checked rather than assumed |
The comparison happens in the last three rows. Stated interest followed by a completed next step is evidence. Stated interest followed by silence is a reaction. Keep both in the map. A run of enthusiastic interviews with no completed actions is a finding in itself, and the guide on validating a startup idea before building explains how to turn that pattern into a build, narrow or stop decision.
Structured programs treat interviews as sustained work across many accounts. The National Science Foundation's Innovation Corps Teams program describes customer discovery as talking to potential customers, partners and other industry stakeholders. Its Teams program solicitation expects teams to have completed at least 100 interviews with potential customers and partners by the end of the curriculum. A founder outside such a program can act on fewer interviews, but the say-and-do comparison only means something across several accounts, not one.
A hypothetical worked example
The following is a hypothetical example with invented figures. It illustrates the method and does not describe a real company, customer or measured result.
A founder is exploring room-status software for small independent hotels and interviews eight housekeeping supervisors. Each interview starts with the last shift when a room was missed or turned over late. At one hotel, the supervisor walks the founder down the guest corridor and shows the printed room list clipped to the housekeeping cart, with finished rooms crossed off in pen and late checkouts written in the margin. The founder asks what happens when the list and the front desk disagree, photographs the list with permission and does not describe the software.
Six of the eight supervisors describe a missed or late room in the past month, and five say they would definitely use an app for it. That is the interview finding. The say-and-do map for three of the accounts shows the buying behavior that followed:
| Account | What was said | What happened by the agreed date | Reading |
|---|---|---|---|
| Hotel A | Would use it; the general manager approved a new booking system last year | Sent a week of room lists and introduced the general manager | Buying path visible |
| Hotel B | Would definitely use it | Promised the room lists, then did not reply to two follow-ups | Interest without action |
| Hotel C | Unsure it was worth changing | Sent the room lists; software is chosen by the owner group, not the hotel | User is not the buyer |
Across all eight accounts, two of the five supervisors who said they would use the product completed the agreed next step, and only one connected the founder with a person who approves spending. The founder narrows the next test to hotels where the general manager approves software locally and offers Hotel A a short paid trial, with the success condition written down before it starts.
Acceptance criteria and the next action
An interview is complete when all of the following are true. If one is missing, note the gap rather than guessing.
- You have a dated, specific account of the last occurrence, in the buyer's words.
- You saw at least part of the current workaround, or you recorded that you only heard about it.
- You know who approved the last related purchase, or you know that nothing has been bought for this problem.
- The product was not described before the research questions ended, and any reaction to it is recorded separately.
- The buyer agreed to one specific next step with a date, and you checked whether it happened.
The next action follows from the map, not from how the conversations felt. If completed steps cluster in one type of account, narrow to it and design a small paid test there. If stated interest is high and completed steps are rare, change the segment or the problem before building. If the budget owner keeps turning out to be someone you have not met, interview that role next.
Common mistakes and limits
- Opening with the solution. A short product introduction at the start frames every later answer as feedback on it.
- Asking about the future. Would-you questions collect politeness. Ask what happened last time.
- Accepting the summary. A description of the usual process hides the workaround. Ask to see the most recent instance.
- Counting interest as a result. Only a completed next step goes in the behavior row.
- Interviewing only users. Without the budget owner's view, a strong problem can still lead to no purchase.
The method has limits. It suits products sold to identifiable buyers, especially businesses, where a purchase has a history and an approver. Consumer products bought on impulse rely more on observed purchase tests than on interviews. Interviews also cannot size a market; they explain the buying situation inside accounts you can reach. The run sheet and the say-and-do map are an editorial framework, not a measured result.
Related guides and next step
- Choosing the questions: Customer Discovery Questions That Reveal Buying Intent
- Turning interviews into a decision: Validate a Startup Idea Before Building
- Placing interviews in a wider plan: How to Build a Startup Go-to-Market Strategy by Stage
Daniel's go-to-market mentoring page describes his work with founders as starting with a specific customer problem validated directly with buyers rather than with surveys, before defining the positioning and the offer and winning the first customers personally. That account is self-reported on his site.
Bring your say-and-do map and interview notes to a free 30-minute strategy session to review what your customer discovery interviews actually show and which buyers to test next.
Frequently asked questions
How do you run a customer interview without pitching?
Ask about the last time the problem happened, have the buyer show the current workaround and ask who approved the last related purchase. Keep the product out of the conversation until those questions are done, then ask for one small, dated next step.
What evidence should a founder collect from customer discovery interviews?
A dated account of the last occurrence, the workaround you saw demonstrated, the approval path of the last related purchase, the next step the buyer agreed to and whether it actually happened. Keep stated interest separate from completed behavior.
What are common mistakes in customer discovery interviews?
Opening with the solution, asking whether someone would use or pay for a product, accepting a summary of the usual process instead of a recent instance, counting enthusiasm as a result and interviewing only users who do not control a budget.
What should you say when an interviewee asks what you are building?
Say that you will share it at the end and ask to finish their story first. If there is time after the research questions, describe it briefly and record the reaction separately from the interview evidence.