How to run customer interviews for a side project

From the SoleOS answers series — written about our own product space; grounded in published definitions and documented behavior, never invented numbers.

Run customer interviews for a side project by talking to 15-20 people in your target niche about what they already do — not what they'd hypothetically do with a product you haven't built. Ask about the last specific time they hit the problem, what they tried, and what it cost them in time or money, and keep your own idea out of the conversation entirely. If you can't find 20 people who've lived the problem recently, or none of them are already spending real time or money limping around it, you don't have a validated problem yet — you have a guess.

The Mom Test: ask what even your mom couldn't lie about

The core failure mode in customer interviews isn't lying — it's politeness. Ask "would you use an app that helps you track X?" and almost everyone says yes, including your mom, because she loves you and wants to be encouraging, not because she'd ever open the app. This is the premise behind what Rob Fitzpatrick calls the Mom Test: ask questions specific and past-tense enough that even someone trying to be nice to you can't give you a false positive.

Three rules make this practical:

• Talk about their life, not your idea. Don't describe what you're building; ask how they currently handle the problem.
• Ask about the past and specifics, not the future and generalities. "What did you do last time" beats "what would you do."
• Aim for facts, not opinions or compliments. "How much did that cost you" beats "do you think this is a good idea."

If you strip every hypothetical and every mention of your product out of your interview script, what's left is usually the good version.

Who to interview, and how to find about 20

You want roughly 20 conversations inside one reachable niche — enough that patterns repeat, few enough to schedule in a few weeks around a day job. Fewer than 10 and you're guessing from noise; more than 30 and you're stalling on building.

Sources that work better than cold outreach:

• Communities where the niche already gathers — a subreddit, a Slack or Discord, a niche forum, a professional association's members list.
• Public evidence of the problem — people complaining on X/Twitter, leaving one- and two-star reviews on a competitor's app, asking about it in a forum thread.
• Warm intros through your existing network, filtered hard for actual fit — a friend who happens to run the kind of business you're targeting counts; a friend who's just being supportive doesn't.
• Existing users of adjacent tools, if you already run other small products — the same instinct you'd use to spot a related idea worth validating as a SaaS idea applies to finding people to talk to about it.

Friends and family are fine as a starting point to practice your script, but don't let them be a meaningful share of your 20 — they're structurally biased toward telling you what you want to hear.

Questions that surface truth, and the ones that produce false positives

Good questions ask about a specific, recent, real event:

• "Walk me through the last time you ran into this."
• "What have you tried to solve it?"
• "What did that cost you — in money, in time, in hassle?"
• "What's frustrating about how you handle it today?"
• "When's the last time you went looking for a better way to do this?"

Leading questions produce enthusiastic-sounding false positives:

• "Would you use a tool that did X?"
• "Would you pay $20/month for this?"
• "Do you think this would be useful?"

The difference isn't tone, it's tense and subject. Past and specific gives you data. Future and hypothetical gives you a compliment.

Listen for evidence of real pain, not enthusiasm

The strongest signal in an interview isn't how excited someone sounds — it's what they're already doing about the problem. Are they paying for a workaround, even a bad one? Have they hired someone, built a spreadsheet, or cobbled three tools together to patch the gap? Have they spent real hours on it recently?

If a problem is real and costly, people have already tried to solve it badly. If nobody you talk to has spent money or meaningful time on it, be skeptical that they'll spend money on your solution either — no matter how enthusiastically they respond in the room. Track this explicitly rather than trusting your gut afterward: for each interview, write down whether they described an existing spend of time or money, and on what. That log is the closest thing to a number you can trust at this stage — don't invent a "% who said yes" score, since a show of hands after a pitch measures politeness, not demand.

Keep your idea out of your mouth

Don't describe what you're building until the interview is essentially over. If someone asks directly, deflect gently: "Still figuring that out — tell me more about how you handle it today first." The moment you reveal your idea, every subsequent answer becomes a referendum on being nice to you rather than a report of their actual behavior, and you lose the rest of the interview.

This is the hardest discipline for a solo founder to hold, because you're excited and the person in front of you is often being warm and encouraging. Hold it anyway. You can pitch after you've learned something; you can't un-pitch and get honest answers back.

Turning interviews into your first customers

Interviews that surface specific, costly, recent pain are frequently your first customers. At the close of a genuinely good conversation, it's fair to ask if they'd be open to trying an early version when something exists, or to a short follow-up call once you have a prototype. Don't pitch mid-interview — ask this only at the very end, and only if the pain was real.

Keep a simple running list: name, contact, the specific pain they described in their own words, and what they're currently spending on it. That list becomes your first outreach when you have something to show, and it's the same muscle you'll use later when you're working out how to land your first 100 in MRR — the earliest paying customers for most side projects are people who were interviewed, not people who found a landing page cold.

Common mistakes

• Pitching instead of listening. The interview turns into a sales call the moment you start selling, and you stop learning anything.
• Only talking to friends and family. They're structurally motivated to be encouraging rather than honest, and rarely represent your actual buyer.
• Hearing what you want to hear. Confirmation bias is strongest exactly when you're excited about an idea — write down answers verbatim, don't paraphrase them into agreement.
• Stopping after three or four conversations. Patterns don't reliably emerge until you're well into the double digits; early interviews often feel more conclusive than they are.
• Asking about the future instead of the past. Every hypothetical question is an invitation for a polite, unreliable answer.
• Not writing down specifics. "They seemed interested" is not a note. "Spends 3 hours/week on a spreadsheet, tried and abandoned two tools last year" is.

This guide is published by SoleOS, a portfolio dashboard for solo founders running multiple products — see the fuller library of founder playbooks for more of how we think about validation and early traction. You don't need SoleOS to run customer interviews; a notebook, a spreadsheet, and 20 conversations are genuinely enough for one side project. It starts earning its keep once you're doing this across several projects at once and need one place to see which idea actually turned into paying interviews and which one is still just enthusiasm.

Frequently asked questions

How many customer interviews do I need before I start building?

Around 20 in a niche you can actually reach is a reasonable target for a side project. Fewer than 10 and any pattern you think you see is probably noise; much past 30 and you're likely stalling rather than learning something new. What matters more than the exact count is whether the same specific pain, and the same evidence of existing spend, keeps showing up across different people.

Should I record interviews or just take notes?

Either works, but write down exact phrases in the moment rather than summarizing afterward from memory — memory tends to smooth answers into whatever you wanted to hear. If you record (with permission), still jot the specific costs, workarounds, and quotes as you go so you're not relying on transcribing everything later.

What if everyone I talk to says they'd pay for it?

Be suspicious of that exact sentence, especially if it came in response to you describing the product. It's a hypothetical, future-tense answer, which is precisely the kind the Mom Test warns against. Go back to what they're doing right now — what they've already tried, already paid for, already spent hours on — before you trust any statement about future willingness to pay.

Can I run this process part-time around a day job?

Yes — most solo founders do. Twenty 20-30 minute conversations spread over three or four weeks of evenings and lunch breaks is a normal pace. The discipline that matters isn't speed, it's consistency in the questions you ask and keeping your own idea out of the room until the interview is over.