Customer discovery: how to talk to users before you build more

By MentionLeads · July 15, 2026 · 7 min read

In short: Customer discovery means talking to specific people who have recently experienced the problem you want to solve, then asking about their behavior rather than pitching your idea. A solo founder can find these people in focused communities, interview them with Mom-Test-style questions, and organize repeated problems, triggers, alternatives, and language into positioning. Do this before building more whenever you cannot clearly explain who has the problem, what they currently do about it, and why changing is urgent.

Customer discovery is not a round of calls designed to collect compliments or feature requests. Its purpose is to reduce uncertainty: whether the problem happens often, whether it carries a meaningful cost, and whether someone has already tried to solve it. The useful output is not “people liked the idea.” It is evidence about real behavior that changes what you build, whom you target, or how you describe the product.

Who should you talk to first?

Start with a narrow behavioral definition, not a broad demographic persona. “SaaS founders” is too vague; “solo B2B SaaS founders who manually search Reddit for sales conversations every week” gives you a group with shared circumstances and observable behavior. If you cannot describe the behavior that qualifies someone, interviews will blend several markets and produce contradictory advice.

Write a one-sentence recruitment filter containing the person, situation, and recent action. For example: “I want to speak with agency owners who lost or nearly lost a client because reporting took too long during the past three months.” Recency matters because people remember concrete events better than general opinions.

• Prioritize people who experienced the problem recently.
• Include people who built a workaround, paid for an alternative, or abandoned a solution.
• Speak with non-buyers as well as enthusiastic early adopters.
• Exclude friends who mainly want to encourage you.

Five detailed conversations with one coherent segment can teach more than dozens of calls across unrelated audiences. The goal is not statistical proof. It is finding patterns strong enough to form a narrower hypothesis that you can test through sales, a landing page, or a small product change.

Where can solo founders find people to interview?

Go where people already discuss the workflow or frustration without being prompted. Relevant places include specialized subreddits, Slack or Discord groups, Hacker News threads, X conversations, industry forums, product review pages, and the audiences of competing tools. Search for problem language such as “how do you,” “alternative to,” “frustrated with,” “takes too long,” and “what are you using for.” This guide to finding your target audience online provides a broader discovery process.

Communities are most useful when you observe before recruiting. Read several weeks of discussions and note repeated jobs, complaints, attempted fixes, and vocabulary. On Reddit, market research starts with existing conversations, not a survey dropped into a community where you have no context.

Recruit with a short, specific request: “I’m researching how independent recruiters track candidates who go quiet. I’m not selling anything. If this happened to you recently, could I ask about the last time?” Public replies are often better than unsolicited direct messages because members can judge your intent. If direct outreach is appropriate, reference the exact comment that made the person relevant and offer an easy way to decline.

What customer discovery questions reveal real demand?

The Mom Test, written by Rob Fitzpatrick, popularized a durable rule: ask about someone’s life and past behavior instead of requesting opinions about your idea. A useful interview follows one recent incident from trigger to outcome. Keep asking for specifics until you can reconstruct what happened without filling gaps with assumptions.

• “Tell me about the last time this happened.”
• “What triggered you to deal with it?”
• “What did you try first, and why?”
• “How are you handling it today?”
• “What is difficult or expensive about that approach?”
• “Who else was involved in choosing a solution?”
• “What happened when you did nothing?”
• “Have you paid for anything or spent time building a workaround?”

Ask follow-up questions using the interviewee’s words. If someone says reporting is “painful,” ask what painful means: hours of manual work, delayed decisions, client complaints, or embarrassment during meetings. Those consequences matter more than the adjective. End by asking who else has dealt with the same situation; warm introductions often produce better interviews than another generic recruitment post.

How do you avoid compliments and false positives?

Do not explain the product before learning how the person behaves. Questions such as “Would you use this?”, “Do you think this is useful?”, and “How much would you pay?” invite speculation and politeness. A positive answer creates little evidence because agreeing costs nothing.

Replace hypothetical questions with requests for proof. Instead of asking whether someone would pay for automated reporting, ask what they currently pay, how the budget was approved, and what made them search for a solution. Instead of asking which features they want, ask them to show you their current workflow and identify where it breaks.

Treat commitments as stronger signals than compliments. A person who shares data, introduces a colleague, schedules a second call, tests a manual service, or agrees to pay has accepted some cost. That still does not guarantee a market, but it is more useful than “I’d definitely use this.”

How do you turn interviews into positioning?

After every conversation, capture facts under the same headings: segment, triggering event, current alternative, cost of the problem, desired outcome, objections, buying process, and exact phrases. Do not summarize too quickly. “Lead generation is hard” loses the useful language in “I miss buying conversations because I cannot monitor ten communities all day.”

Then compare interviews and look for clusters rather than majority votes. A viable cluster might share the same trigger, workaround, and consequence even if interviewees request different features. Create a simple evidence table with one row per interview and one column per theme; mark only what the person actually said or demonstrated.

• For: the narrow segment with the repeated behavior.
• When: the trigger that makes the problem urgent.
• They need: the outcome, expressed in the customer’s language.
• Unlike: the current workaround or main alternative.
• We help by: the mechanism that creates the outcome.

A first positioning statement might read: “For solo SaaS founders who miss relevant community conversations, this product identifies high-intent discussions and prepares contextual replies, unlike manually checking feeds and keyword alerts.” This is a testable claim, not permanent brand copy. Revise it when sales conversations reveal a more urgent segment or a different reason people switch. For a deeper framework, see startup positioning.

When have you done enough customer discovery?

Stop interviewing temporarily when new conversations within the same segment mostly repeat known triggers, alternatives, objections, and language. The next step should then test behavior: ask for a pilot, charge for a manual service, publish focused positioning, or build the smallest workflow that resolves the repeated problem.

Return to discovery when evidence conflicts. Warning signs include low activation, prospects misunderstanding the product, long sales cycles without a clear objection, or users requesting unrelated capabilities. Discovery is not a phase completed once; it is the fastest tool for diagnosing why observed behavior differs from your assumptions.

Frequently asked questions

How many customer discovery interviews should a solo founder do?

There is no universal number. Start with a narrow segment and continue until interviews repeat the same behaviors, triggers, alternatives, and objections; then run a behavioral test. If answers remain inconsistent, narrow the segment before simply adding more calls.

Should I show a prototype during a customer discovery interview?

Usually, learn about the person’s recent behavior before showing anything. A prototype can be useful later for testing comprehension or workflow, but it can also anchor the conversation around your solution. Separate problem discovery from prototype feedback so you know which evidence came from lived experience and which came from reacting to your design.

Should customer discovery participants be paid?

Payment can be appropriate when you need specialized professionals, structured research time, or people outside your network. Keep the incentive modest and make clear that honest criticism is the job. For sales-oriented discovery, willingness to continue without payment can itself reveal urgency, although it should not be treated as definitive proof.

Start here

• Write one recruitment sentence that names a specific person, recent situation, and qualifying behavior.
• Find ten relevant community discussions and record the exact problem language, workarounds, and consequences people mention.
• Book three conversations, ask about the last real incident, and update one positioning statement using only repeated evidence.

If manually monitoring Reddit, X, and Hacker News is the bottleneck, MentionLeads can surface high-intent conversations so you can review them and respond personally.