Most startup ideas do not fail because they were built badly. They fail because nobody needed them enough. The cheapest way to find that out is to talk to people before you build anything, and you do not need to write a line of code to do it.
This guide is for people who have an idea and no technical background. It covers who to talk to, how to find ten of them, the twelve questions worth asking, how to avoid steering the answers, and how to turn a pile of notes into a decision.
The short answer. Pick one group of people. Talk to about ten of them for twenty minutes each. Ask about what they did last time the problem appeared, not what they would do in future. Count only actions, money and effort as evidence. Stop when one sentence describes the problem and a few people have committed to something.
Who to talk to (and who not to)
Choose one narrow group, written the way you wrote the “who” line in our launch guide. “Solo yoga teachers who take bookings by text message” is a group. “Small businesses” is not.
Good interview subjects have had the problem recently. Someone who might have it one day will give you opinions; someone who had it last month will give you a story.
Avoid these people, at least at first:
- Friends and family. They want to be kind. Their encouragement feels good and tells you almost nothing.
- Experts and investors. They give opinions about markets. You need facts about one person’s week.
- People who are not in the group. A useful conversation with the wrong person is the most misleading kind.
How to find ten people
You do not need an audience. You need ten names and a short message.
- Start with people one step away. Ask friends, “Do you know someone who runs a small studio / manages a rental / teaches?” Introductions get more replies than cold messages.
- Go where the group already talks. Communities, forums, local groups and social media channels for that profession. Read first; join the conversation before you ask for anything.
- Send a short, honest message. Say who you are, that you are not selling anything, what you are curious about, and that it takes twenty minutes. Offer to share what you learn.
- Ask for the next name. End every conversation with “Who else deals with this?” Two or three introductions per call is a good sign.
Expect many people not to answer. That is normal and says nothing about your idea.
The ground rules of a good interview
Your job is to listen. A few rules keep you honest:
- Ask about the past, not the future. “Tell me about the last time” produces facts. “Would you use…?” produces politeness.
- Do not pitch. If you describe your idea early, people start being nice to it instead of telling you about their life.
- Stay with specifics. When someone says “it’s a hassle”, ask, “What happened last time it was a hassle?”
- Be comfortable with silence. After an answer, wait three seconds. People often add the most useful part.
- Aim to talk about a quarter of the time. If you are explaining, you are not learning.
The 12 questions
Use them in order, but follow the person. You will not need every one every time.
Warm-up
- “Tell me a bit about what you do and how a typical week goes.” You learn their world in their words, and it relaxes both of you.
Past behavior
- “When did you last run into [the problem]? Walk me through what happened.” A recent, specific story is the best evidence you can get. If they cannot recall one, the problem may be rare.
- “What did you do about it?” This shows whether it mattered enough to act.
Current workaround
- “What do you use today to handle this?” Spreadsheets, text messages, a paper notebook, a tool they dislike. Your real competitor is usually the workaround, and it is often free.
- “What do you like and dislike about that?” The dislikes are your opening. The likes are what you must not break.
Cost of the problem
- “What does this cost you, in time, money or stress?” Let them answer in their own units. Ask how often it happens so you can see the size.
- “What happens if you do nothing about it?” If the honest answer is “not much”, you have learned something valuable.
Who decides
- “Who else is involved in choosing or paying for something like this?” The person with the pain is not always the person who pays. Find out before you build for the wrong one.
What they tried
- “Have you tried anything else to fix it? What happened?” People who already searched, paid or built a workaround are far more serious than people who never looked.
- “Why did you stop using it, or why didn’t you try?” Reasons to quit or not start are the hidden requirements of your product.
Willingness to commit
- “If this problem went away, what would be different for you?” This describes the value in their words, which later becomes your message. Do not ask for a price in dollars yet.
- “I’m going to build a small first version. Would you try it with me, and could I check back in two weeks?” Then ask for something real: an introduction, a deposit, a signed waitlist entry, a booked trial slot. How they respond to a concrete ask tells you more than any compliment. When money comes up, our pricing guide helps you set a first number.
Avoiding leading questions
A leading question hands people the answer you want. Compare:
| Leading (avoid) | Better |
|---|---|
| “Wouldn’t it be great if you could book online?” | “How do people book with you today?” |
| “Would you pay $10 a month for this?” | “What have you spent on this problem so far?” |
| “Do you find scheduling frustrating?” | “Tell me about the last time scheduling went wrong.” |
| “Would you use an app that does X?” | “What did you do the last time you needed X?” |
| “Is price an issue?” | “How did you decide what to pay for what you use now?” |
The dollar figures above are invented illustrations of a style of question, not data about any market. The pattern to remember: a question that can be answered “yes” in one word is probably leading.
How to record and synthesize
Ask permission before recording, and if people prefer not, take notes by hand. Right after each call, spend ten minutes filling one row of a table. One page, one row per person:
| Person | Recent story (one line) | Current workaround | Cost they described | Who decides | Committed to | Their words worth keeping |
|---|---|---|---|---|---|---|
| P1 | ||||||
| P2 |
After five or six rows, patterns appear. Highlight a cell when the same workaround, the same cost or the same phrase shows up a second time. Do not average; look for repetition. One vivid story is a lead; the same story from several people is a finding.
If you want help organizing transcripts, you can paste notes into an AI chat and ask it to fill the table. Check its answers against your notes, and keep the quotes exact.
What counts as evidence, and what is politeness
Sort everything you heard into two piles.
Evidence (count it):
- A specific recent story of the problem
- Time or money they already spend on it
- A workaround they built themselves
- They looked for a solution and tried something
- They introduced you to someone else
- They agreed to a deposit, a trial or a follow-up date
Politeness (do not count it):
- “That’s a great idea.”
- “I’d definitely use that.”
- “You should build it.”
- “Let me know when it’s ready.”
- Interest that disappears when you ask for any commitment
A useful test: if you stopped replying to messages, would they notice? If a person has the problem badly, they will chase you. If they only liked the idea, they will not.
When to stop talking and build
Return to the proof statement from the launch guide: “Ten teachers use a booking page for two weeks and at least five say they would pay for it.” Interviews should leave you able to write a stronger version.
You can start building when you can finish this sentence with real names behind it: “[Group] currently handles [problem] by [workaround], which costs them [cost], and [N] of them have agreed to [commitment].”
Stop interviewing and build the smallest version when:
- Several people told you the same recent story, in similar words.
- At least some committed to something beyond kind words.
- Your idea has narrowed, not grown: one group, one problem, one first flow.
Go back and change the idea or the group when:
- People like it but nobody has done anything about the problem so far.
- Everyone’s story is different and no pattern appears after about ten conversations.
- The person with the pain is not allowed or able to pay, and nobody else wants to.
Changing direction after ten conversations is a success. It cost you a few hours, not a few months. The 30-day plan from idea to paying customers places these conversations in the first week of a calendar.
Where equipe fits
Everything above works without any tool. Interviews are a human job: nobody should talk to your customers for you.
Where equipe can help is the step after. It is an AI startup team for people who cannot code: a PM turns your table and proof sentence into a plan and asks you questions first, engineers build the smallest version, QA tries it as a customer would, and anything that goes live, costs money or reaches customers waits for your approval. What an AI startup team does describes the roles in detail.
To be honest about fit: if you are still unsure whether the problem exists, you do not need a team yet, and a spreadsheet or a free chat tool is enough. You can start free with $5 of credit and no card (see pricing), but do the interviews first.
Where to go next
- Write your one-line group and your proof sentence today.
- Send ten short messages this week and book as many calls as you can.
- Fill the table after each call, and look for repetition after five.
- Once you build, measure it. Three numbers a small product should track shows what to count, and build, measure, learn with an AI team shows how to act on it.
- For the full path, read how to launch a real product without code.