How to Talk to Users Before You Build (Without Getting Polite Lies)

Almost every founder who says they validated their idea by talking to people actually collected encouragement. The problem is not that people lie, it is that "would you use this?" is a question about an imaginary future, and being nice about someone's imaginary future costs nothing. Ten conversations about what someone did last month will teach you more than a hundred about what they might do next year. This is how to run those conversations when you have no product yet.

Table of contents

• Why "would you use this?" is worthless
• Where to find ten people
• How to ask for the call
• The questions that work
• Running the call
• Reading the answers honestly
• Where surveys fit
• Frequently asked questions

Why "would you use this?" is worthless

Ask ten people whether they would use your product and most will say yes. They are not deceiving you. They are answering a hypothetical politely, with no cost attached, usually while wanting the conversation to stay pleasant. The answer is real as social behaviour and worthless as evidence.

The single most useful correction, and the core idea of Rob Fitzpatrick's book The Mom Test, is to stop asking about your idea and start asking about their life. Their past behaviour already happened, so it cannot be inflated to spare your feelings.

Notice that none of the useful questions mention what you are building. That is deliberate. The moment you describe your product, the person switches from telling you about their work to reacting to your idea, and you cannot switch them back in the same call.

Where to find ten people

This is the part founders treat as impossible and it is usually the easiest, because you are asking for twenty minutes rather than money.

• Your waitlist, first. These people already raised a hand. If you asked a question on the signup form, you can even pick the ones whose answers suggest they have the problem rather than the curiosity.
• The people who did not sign up. Harder to reach and more informative, because they are the majority. If you can find even three, their reasons are the ones you cannot guess.
• Communities where the work happens. Not founder communities. The subreddit, Discord, or Slack for the job itself. Answer questions there for a fortnight first, then ask.
• Two degrees out from you. Ask people you know for one introduction each to someone who does the job. This scales badly and works extremely well.

Ten is a reasonable target. You will usually notice the same two or three themes repeating by the sixth or seventh conversation, and when new calls stop surprising you, you have enough.

How to ask for the call

Short, specific, no pitch, and an easy way to say no. The most common mistake is describing your product in the request, which turns it into a sales message and halves your reply rate.

Subject: Quick question about how you handle invoicing

Hi Sam,

You mentioned on the freelancers subreddit that chasing
late invoices eats your Fridays. I am researching how
people actually handle that, and I am not selling
anything.

Would you be up for 20 minutes in the next week or two?
Happy to work around your schedule, and equally happy
if this is a no.

Rouzbeh

"I am not selling anything" has to be true. If you end the call with a pitch, you have spent the goodwill of everyone that person might have introduced you to.

The questions that work

Six questions is plenty for twenty minutes. Ask them roughly in this order and let the conversation wander, because the useful material is almost always in the wandering.

1. Walk me through the last time you did this. What happened, step by step?
2. What did you use? Include the spreadsheet, the notebook, and the colleague you asked.
3. What was annoying about it? Push past the first answer, which is usually the polite one.
4. Have you tried to fix it before? What happened to that attempt?
5. Roughly how often does this come up, and how long does it take?
6. Who else deals with this? Can you introduce me?

Question four is the one that separates a real problem from a mild irritation. If someone already tried to solve it, badly, with a spreadsheet at eleven at night, that is a person with a problem. If they have never thought about it until you asked, you are looking at a preference.

Question six is how ten conversations turn into twenty without any more cold outreach.

Running the call

• Twenty minutes, and end on time. People give you a second call if you respected the first one.
• Talk less than a third of the time. If you are explaining, you are not learning. Silence after an answer is the most productive tool you have, because people fill it with the detail they were editing out.
• Ask for the story, not the summary. "Last Tuesday" beats "usually". Summaries are already interpreted; stories still contain the facts.
• Record it if they agree, and write notes either way. Write them within the hour. What you remember the next morning is the part that confirmed what you already believed.
• Save the pitch for the very end, if at all. If you must, ask after you have closed the notebook: "I am building something in this area, can I show you in a few weeks?" That is a request, not a demo.

Reading the answers honestly

This is where most of the value is lost, because it is where your own hope gets a vote. Sort what you heard into evidence and noise.

The right-hand column is not worthless socially, but none of it should change your roadmap. A useful discipline: after each call, write one sentence answering "what did I learn that I could not have guessed?" If you cannot answer it, you talked too much or asked about your idea.

Watch for the pattern where everyone confirms the problem and nobody has done anything about it. That combination usually means it is real but not painful enough to pay for, which is the most expensive category of idea to build.

Where surveys fit

Interviews and surveys answer different questions and neither replaces the other. Interviews tell you why, in language you would never have invented. Surveys tell you how common it is, once you know what to ask.

The order that works: talk to five or ten people first, then use their words to write the survey. A survey written before any conversations asks the questions you already thought of, in your phrasing, and confirms your existing assumptions with impressive precision. Once the survey is running on your signup form, the answers also tell you which new signups are worth a call, which is how the two loops feed each other.

More on that in how to run a product survey, and on why the answers matter more than the count in why answers beat emails.

Frequently asked questions

How many conversations is enough?

Stop when new calls stop surprising you, which is usually somewhere between eight and fifteen. If the tenth person tells you something genuinely new, keep going.

Should I pay people for their time?

Rarely necessary at twenty minutes, and it changes the dynamic: paid participants try to be helpful, which is exactly the bias you are trying to avoid. Offer something if you are asking for an hour or for a specialist's time.

What if I already built the thing?

Same questions, harder discipline. You will want to show it, and showing it converts the call into a demo. Do the interview first and the demo in a separate conversation if you need both.

Can I do this over email or chat instead?

You can, and you will get shorter, more considered, less useful answers. The value is in the follow-up question you did not plan, which needs a live conversation. Async is a fine fallback when the alternative is nothing.

Everyone I speak to is positive. Is that good?

It usually means you are describing your idea rather than asking about their work. Uniform enthusiasm is a symptom of the wrong questions, not a sign of a great idea. Rerun a call sticking strictly to the past-tense questions and see whether the picture holds.

The whole discipline reduces to one habit: ask about what already happened, and shut up long enough to hear the answer. Ten of those conversations will change what you build more than any number of signups, because a signup tells you someone was curious and a story tells you what they actually do on a Tuesday. Do them before you write the code, when changing your mind is still free.

Lighthouse keeps your waitlist and its survey answers in one place, so you can see who is worth a twenty minute call and reach them without exporting anything. From an indie dev, for indie devs and makers.