How to Talk to Customers Without Lying to Yourself
The Mom Test for weekend builders: how to run 10 customer interviews that produce real evidence, which questions to ban, and how to tell a signal from politeness.
Quick Answer
Never ask anyone whether your idea is good. Ask about their life instead — specifically, the last time they hit the problem, what they did about it, and what it cost them. Rob Fitzpatrick's The Mom Test is built on one rule: ask questions even your mother couldn't lie to you about, which means questions about the past, not opinions about the future.
Ten to twenty interviews is the working range for customer discovery before you build. Most weekend builders need about ten, because patterns show up faster than people expect when the questions are right.
This matters because the most common cause of startup death is building something nobody needed. CB Insights' post-mortem analysis puts "no market need" at 42% of failures — the largest single reason, ahead of running out of cash.
Who This Is For
You have an idea you like and no evidence anyone else does.
You've talked to people about it and everyone said it sounded great, and you correctly suspect that means nothing.
You're about to spend a weekend building and want to spend it on the right thing. If you don't have a candidate yet, pull one from the startup ideas library first, then run this process against it.
You find cold outreach uncomfortable and want a script that isn't a pitch.
Why Everyone Lies to You
Nobody is being malicious. Three ordinary human reflexes corrupt almost every founder conversation:
Politeness. Telling someone their idea is bad is socially expensive. Telling them it's interesting costs nothing.
Bad prediction. People are genuinely poor at forecasting their own future behavior. "Would you use this?" asks for a prediction, and the honest answer is that they don't know.
Your framing. The moment you describe your solution, you've told them the answer you want. Everything after that is contaminated.
The consequence is a founder with 30 enthusiastic conversations, zero customers, and no idea what went wrong. Enthusiasm is not data. Behavior is.
The Three Rules
1. Talk About Their Life, Not Your Idea
You are not there to describe what you're building. Ideally, you never mention it.
The interview is an investigation into how they currently handle a problem. Your idea is a hypothesis you're testing quietly in your own head while they talk.
2. Ask About Specifics in the Past, Not Generics in the Future
"Would you pay for X?" is worthless. "What did you pay for the last thing you tried?" is evidence.
The past is verifiable. The future is imagination — usually flattering imagination.
3. Talk Less
If you're talking more than a third of the time, you're pitching. Ask, then be quiet long enough to be uncomfortable. The best material almost always arrives after the pause you were tempted to fill.
Questions That Work
Open with context, then go specific:
"Walk me through the last time this happened." The single best question. It forces a real memory instead of a general impression. If they can't recall a specific instance, the problem isn't frequent enough to build on.
"What did you do about it?" Reveals the actual current solution — which is your real competitor. Usually it's a spreadsheet, an intern, or nothing.
"How long did that take? What did it cost you?" Converts vague annoyance into a number. Numbers are what let you price later.
"What have you already tried?" If they've tried three tools and abandoned them, you have a hot problem and a tough retention challenge. If they've tried nothing, the pain may be theoretical.
"How often does this come up?" Frequency determines whether you're a habit or a novelty.
"What happens if you just... don't fix it?" The best disqualifier in the set. If the honest answer is "nothing much," you've found a vitamin, not a painkiller.
"Who else deals with this?" Ends every interview. It's how ten interviews become twenty without any more cold outreach.
Questions to Ban
Cross these out of your notes before you start:
- "Would you use this?" — asks for a prediction, gets you politeness
- "Do you think this is a good idea?" — asks for approval, gets you approval
- "How much would you pay for this?" — hypothetical money is not money
- "Would you pay $29 a month for this?" — a leading question with a number attached
- "Don't you hate it when...?" — you've supplied both the problem and the answer
- Anything that begins with a description of your product
If a question can be answered "yes" by someone who wants the conversation to end pleasantly, it's the wrong question.
Reading the Answers Honestly
Not all positive responses are equal. Rank them:
| Signal | What it means | Weight |
|---|---|---|
| "Can I use it now?" | Real | Highest |
| Paid you, or paid a competitor | Real | Highest |
| Gave a specific recent example unprompted | Real | High |
| Introduced you to a colleague | Real | High |
| Currently pays for a workaround | Real | High |
| "That sounds useful" | Politeness | None |
| "I'd definitely use that" | Prediction | None |
| "Great idea, keep me posted" | Polite exit | None |
Commitment is the dividing line. Real interest costs the person something — time, a referral, a calendar slot, money. Anything free is noise.
The rule of thumb: if nobody gives up anything, you have learned nothing. A "keep me posted" list of 40 people typically converts to almost zero paying customers.
How to Actually Find Ten People
This is the part that stops most builders, and it's easier than it looks.
Go where they already gather. The subreddit, Facebook group, Slack community, Discord, or trade association for your niche. Don't pitch — post a genuine question about how people handle the workflow.
Ask for 15 minutes, not "a chat." A bounded ask gets answered. "Can I ask you about how you handle invoicing? 15 minutes, I'm not selling anything" works because it's true.
Use the referral question. Every interview ends with "who else deals with this?" Two names per conversation and you never cold-outreach again after the fourth.
Talk to people who recently bought something adjacent. Recent buyers are in a spending mindset and remember their evaluation clearly.
Ten conversations at 15 minutes is two and a half hours of talking. That is less time than you'd spend on the landing page.
What to Do With What You Learn
After each conversation, write three lines while it's fresh:
- The specific problem instance they described
- What they currently do instead
- Any commitment they made, or none
After ten, sort the problems by frequency and emotional intensity. The intersection — happens often and makes them visibly annoyed — is what you build first.
Then look at what they currently do instead. Your MVP has to beat that, not beat your imagined competitor. Beating a spreadsheet is a different job from beating a funded startup, and usually an easier one.
If no problem cleared both bars, that's a successful outcome. You spent an afternoon instead of a weekend, and you can go pick a different idea from the startup ideas library without having built anything.
What If It Goes Wrong
Everyone is polite and vague. You're describing your idea too early. Don't mention it at all for the first ten minutes.
They confirm the problem but won't commit. The pain is real but not urgent, or you're talking to someone without budget authority. Try a more senior contact in the same organization.
You can't find ten people. That's a reachability failure and it's a finding, not a blocker. If you can't locate ten of them for a free conversation, marketing to them later will be far harder.
They ask for features you didn't plan. Note them, don't promise them. Feature requests during discovery are usually descriptions of the current workaround, not the product they'd buy.
FAQ
How many customer interviews do I need before building?
Five to twenty is the usual discovery range; ten is a good target for a weekend MVP. Stop when new conversations stop producing new information — that saturation point usually arrives between ten and fifteen.
What is The Mom Test?
A customer-interview framework from Rob Fitzpatrick built on asking questions so grounded in past behavior that even someone who loves you couldn't mislead you with the answer. It bans opinions about your idea in favor of facts about their life.
Should I tell them what I'm building?
Not until the end, if at all. Once you describe the solution, you've anchored the conversation and every subsequent answer is about your idea rather than their problem.
Do these interviews replace a landing page test?
No — they pair. Interviews tell you whether the problem is real and how it's phrased. A priced landing page tells you whether people act. Run interviews first, then build the page using their exact words.
What counts as real validation?
Money, a signed commitment, a referral, or a scheduled follow-up they chased you for. Verbal enthusiasm counts for nothing.
TL;DR
Ask about their life, not your idea. Ask about specific past events, not future intentions. Talk less than a third of the time. Ban every question that invites politeness. Weigh commitment — money, time, referrals — and discard compliments entirely. Ten interviews, three notes each, sorted by frequency and intensity, tells you what to build this weekend. Need a candidate to test first? Pick one from the startup ideas library.