How to do customer interviews when you have zero users

You have no users. That is normal. It is not a reason to hide in the editor. It is the moment to talk to people who already live with the problem you think you are solving.

In 2026 you can build a product in a weekend. AI writes the boring parts. Building is not the hard part. Learning whether anyone has the pain is. That is why you validate the idea before you write code. An interview with zero users is not a survey about your mockups. It is a short conversation about their life last month: what broke, what they tried, and what it cost.

Indiecity is for one person or a team under twenty. You do not need a research team. You need a calendar link, ten honest notes, and questions that do not beg for compliments.

Why most “validation” chats are useless

You pitch the idea. They say it sounds great. You ask if they would use it. They say maybe. You leave happy. You learned nothing.

People protect your feelings. Friends and family do it on purpose. Strangers do it out of politeness. Rob Fitzpatrick wrote The Mom Test for that exact failure: everyone will lie to you if you make it easy. The book still holds in 2026. The job is not to get them to like the idea. The job is to make lying hard.

YC partners still teach the same trap. Eric Migicovsky’s Startup School talk names three errors: you talk about your idea, you ask about a future that has not happened, and you talk more than you listen. Fix those three and the call starts working.

The Mom Test in three rules

Fitzpatrick’s rules are short. You can tape them next to the keyboard.

1. Talk about their life, not your idea.
2. Ask about specifics in the past, not opinions about the future.
3. Listen more than you talk.

Do not open with the product. Do not ask “Would you use this?” Do not ask “How much would you pay?” Ask what they did the last time the mess showed up. Ask what they already pay in money, hours, or headcount. Ask what failed. Past behavior is the data. Future promises are guesses.

Compliments are not signal. “Cool idea” is not signal. “I would definitely buy that” is not signal until money moves. When they praise you, nod and drag the talk back to the last time the problem hurt.

Where to find people when you have zero users

You are not looking for “users.” You are looking for people who already have the problem. Jason Cohen wrote the field list for founders who have no product, no brand, and no list: go where the ideal buyer already is, frame the ask as advice, and expect a low reply rate on cold outreach.

Start warm. Ask your whole network for one intro to a specific job title with a specific pain. Most of your friends are not the buyer. Many of them know one person who is. Cohen’s line is useful: describe the buyer so tightly that a name pops into their head.

Then work the rooms and records where the pain is already written down:

• LinkedIn search by title and industry (Cohen used this for early WP Engine interviews)
• Niche Slack, Discord, Facebook Groups, and subreddits where buyers vent about the workflow
• One-star reviews and “anyone know a tool for X” threads
• People actively switching tools, building workarounds, or posting job reqs that describe the mess
• Industry meetups and conferences where your buyer goes to talk about their work
• Your own list: coworkers, past clients, people who complained in a chat last year

Do not confuse founder rooms with customer rooms. If your buyer is not a founder, Indie Hackers and Product Hunt still give you claps, not proof. Sit where the buyer already complains in their own words.

End every interview with a name. “Who else should I talk to?” Cohen’s version: if you average one intro per call, you stop running out of people. Ask like you mean it. You are trying to have a set number of conversations, and an intro is the most useful thing they can do for you right now.

Scripts that get a yes to a call

People hate surveys. They often like being asked for advice. Keep the note short. Name their world. Offer a tight slot. Put a calendar link in the message so they do not have to negotiate times.

Hey [name]. I am talking to [job title] who still deal with [pain in their words]. Not a pitch. I want 20 minutes on how they handle it today. Do you know one person I should write to? A name is enough.

Hey [name]. I saw [specific thing: their post, review, role, talk]. I am researching how [job title] handle [problem], not selling anything. Could I get 20 minutes this week for your advice on what actually breaks? Here is my calendar: [link]. Happy to work around you.

Hey [name]. Your note about [exact pain they wrote] matches what I keep hearing. I am not pitching a tool. I am trying to understand what people already tried. Open to a 15-minute call this week? [calendar link]

Write each note for one person. Inbox noise in 2026 is full of the same ChatGPT opener. A line that only fits them still gets opened.

What to ask on the call

Fifteen to twenty minutes. Phone or video. Notes open. You are there to get facts about their life, not to demo.

Migicovsky’s five questions still work as a spine:

1. What is the hardest part about [doing the thing]?
2. Tell me about the last time you hit that problem.
3. Why was it hard?
4. What, if anything, have you already done to solve it?
5. What do you not love about the solutions you tried?

Push for numbers when you can. How often does it happen? How much time or money does it burn? Who has budget to fix it? Migicovsky’s point on first customers is practical: frequency, cost, and budget tell you whether this person can ever buy.

When was the last time? Walk me through that day. What did you open first? What did you pay for? Who else got involved? What would break if you ignored it for a month?

If they have never tried to fix the problem, treat that as data. A pain nobody spends time or money on is often not a business. If they already duct-taped a spreadsheet, hired a person, or bought a tool they hate, you are in the right room.

Keep learning calls and sales calls apart

Mixing them ruins both. On a learning call you do not pitch. You do not defend the roadmap. You do not ask them to sign up. You thank them, write the notes the same day, and ask for one intro.

When the same story repeats and people already spend on a workaround, you switch modes. You put a price on one page and you ask. That is a different conversation. For the ask itself, use how to get your first 10 paying customers without ads. Interviews tell you the words. Payment tells you the truth. If you are stuck on what a yes looks like, read how to tell if people will pay.

This was useful. I am not selling you anything today. Who else hits the same problem at work? A name and an intro helps more than another feature idea. Can I follow up if a pattern shows up in these talks?

How you take notes so the week is not a blur

Same day, not “later.” Capture their words, not your cleaned-up summary. Keep a simple grid:

• Who (role, company type)
• Last time the problem hit (date-ish, trigger)
• What they tried
• What it cost (time, money, people)
• Exact phrases they used
• Whether they have budget or authority
• Who they said to talk to next

After five to ten calls, read only the phrase column. Those phrases replace the headline on your page. If every call is a different story, you do not have a segment yet. Narrow the buyer before you write more code.

A two-week plan with zero users

1. Day 1. Write the buyer in one sentence: job title, context, pain in their words. Write 30 names or places you will look.
2. Days 2 to 4. Send two personal notes a day. Track sent, replied, booked. Ask every network contact for one intro.
3. Days 5 to 10. Run the calls. Use the five questions. No pitch. Same-day notes. Ask for intros at the end.
4. Day 11. Read the notes for repeats only. Rewrite the problem statement with their phrases. Cut anything you invented.
5. Days 12 to 14. Book the next five calls, or if the pain is clear and they already spend, put a price on a page and start the sales ask.

If you cannot get a single conversation with the people you claim to serve, that is a result. The buyer is wrong, the pain is weak, or you are fishing in the wrong rooms. Fix that before you ship another version. For the pre-build checklist, stay with validate the SaaS idea before you write code.

What counts as a good interview

You leave with a story about a real week in their life. You know what they tried. You know what it cost. You have phrases you did not invent. You have a next name. That is a good interview.

A bad interview is a demo with applause. A bad interview is a feature wishlist. A bad interview is you talking for twelve of the fifteen minutes.

What to stop doing

• Pitching on a learning call and calling it research.
• Asking “Would you use this?” and treating yes as proof.
• Interviewing only friends who want you to feel supported.
• Hanging out in founder forums when founders are not the buyer.
• Building v2 because interviews feel awkward.
• Collecting compliments instead of past behavior.
• Skipping the intro ask at the end of the call.

Then you know what to build (or kill)

Zero users is not a reason to wait. It is the cleanest time to learn, because you have nothing to defend. Run the Mom Test. Find people where the pain is already public. Send human notes. Ask about last time, not next time.

When the same pain shows up and someone will pay to stop it, get the first ten customers. If you learn the idea is wrong, that is a cheap win. Write it down and move. If you ship something real and want the story on the record, send it in. You can join Indiecity while you do the week. The calls still have to be yours.

Book the first call today.