Customer discovery questions for startup founders

The dangerous version of startup validation takes place in a coffee shop with someone you already know. You explain what you are building. They say it sounds useful. You go home and spend six weeks building the feature they just approved.

Nothing in that conversation was malicious. Your friend wants you to win. You wanted encouragement. The question invited a prediction about a product that does not exist yet, and people are very generous with predictions that cost them nothing.

Customer discovery is how you get out of that loop. It is not a pitch disguised as research. It is a short conversation about a person's recent work, their current workaround, and what that workaround is costing them right now.

Here are the customer discovery questions that produce useful answers, how to run the calls, and the signals that tell you an idea is worth another week of your time.

Customer discovery questions are supposed to make your idea uncomfortable

You are not trying to prove that your solution is clever. You are trying to find out whether a specific problem has enough weight behind it that somebody will change behaviour or spend money to make it go away.

That means the best questions are about the past. A useful answer contains a date, a person, a tool, a workaround, or a number. "Last Thursday I spent two hours copying these invoices into a spreadsheet" is useful. "Yeah, that could be annoying" is not.

The rule is simple: ask for a story, not a verdict.

This is the same discipline that makes a failed launch post-mortem useful. You look at behaviour first, then make a diagnosis. A thoughtful sentence from a stranger is a clue. A calendar invite, a spreadsheet they made, or money already spent on a workaround is evidence.

Who should you interview first?

Start with people who have the problem, not people who love startups. The two groups overlap less than founders assume.

Write one sentence that defines the person in terms of their work:

Independent recruiters who schedule more than ten candidate interviews a week and still build their shortlists in spreadsheets.

That is better than "recruiters" because it gives you a way to screen. Someone who hires twice a year will give you a different, much weaker signal than someone who does this every afternoon.

Your first ten people can come from former colleagues, customers of a past project, niche Slack groups, local communities, or people who publicly describe the problem. Do not treat warm introductions as validation. Treat them as an easier route to a real conversation.

If you have a pre-launch waitlist, it is a useful source too. Email the people who signed up with a plain note: you are trying to understand how they currently do the job, and you would value twenty minutes of their time. The reply rate is not the result. The calls are.

The 12 customer discovery questions to ask

You will not need all twelve in every call. Pick the questions that follow naturally from the person's story, then stay quiet long enough for the useful details to arrive.

Start with the current process

1. Walk me through the last time you had to do [job].
2. What happened just before you started?
3. Which tools, people, or documents were involved?
4. Where did the process slow down or go wrong?

The first question does most of the work. If they cannot remember an instance, the problem may not happen often enough to anchor a product. If they pull up a spreadsheet while explaining it, ask what each tab is for.

Find the cost behind the irritation

1. How often does this happen in a normal week?
2. What does it cost when it goes badly: time, money, missed work, or stress?
3. Who notices the problem first, and who has to fix it?
4. Is there a deadline or consequence that makes it urgent?

Founders often hear "that is annoying" and translate it into a subscription business. Do not. Plenty of annoying things stay unsolved because the cost is low. You are listening for a problem that blocks revenue, creates risk, consumes a repeated chunk of time, or makes someone look bad at work.

Learn from alternatives before you mention yours

1. What have you tried to make this easier?
2. Why did that approach not stick?
3. What are you paying for now, including the manual work?
4. If you could change one part of this process tomorrow, which part would it be?

Competitors are not a bad sign. A paid tool, a contractor, a spreadsheet template, or a clumsy internal process means the job exists. The interesting question is why the current answer is still unsatisfying.

Only near the end should you describe your direction, and keep it short. "I am exploring a way to reduce the shortlist step you described. Would it be useful if I showed you something when it exists?" Their answer matters less than whether they make a concrete commitment, such as agreeing to a pilot or introducing you to a teammate.

A 20-minute customer discovery interview script

You do not need a formal research project. You need enough structure that every conversation is comparable.

Minute 0 to 2: set the contract. Say that you are researching a workflow and are not there to sell. Ask permission to take notes. Do not open with your product deck.

Minute 2 to 14: follow the last real example. Use the first eight questions above. When they use a vague word such as "usually" or "sometimes," ask for the last specific instance.

Minute 14 to 18: map alternatives and cost. Ask what they have tried, who approves a purchase, and what a better outcome would change for them.

Minute 18 to 20: make one small ask. Can you follow up with a prototype? Can they show you the spreadsheet? Do they know one other person with the same problem? A real next step tells you more than a compliment.

After the call, write down the answer before you explain it away. Use the same five fields each time: person, job, current workaround, cost, and commitment. This keeps the founder brain from remembering the flattering calls more clearly than the awkward ones.

What counts as validation before you build?

Validation is not a score you earn at the end of ten calls. It is a pattern that becomes hard to ignore.

Strong signals look like this:

• They already pay for a workaround, even if it is a person or a messy service.
• They describe the failure in concrete detail before you prompt them.
• They ask when they can use a solution or offer to introduce the actual buyer.
• Several people use the same words to describe the job and the consequence.
• Someone agrees to be a beta tester with a defined job to do, not just an account to click around in.

Weak signals are not evidence, even when they feel nice:

• "I would totally use that."
• "My cousin needs this."
• A waitlist email address with no reply to a conversation request.
• Likes on a build-in-public post.

None of this means you need permission from twenty people to ship. It means you should earn the right to build the narrowest version of the job you keep hearing about, rather than the broad platform you imagined alone at your desk.

Turn interviews into a landing-page sentence

The output of customer discovery is not a stack of notes. It is language you can reuse.

If three people say they spend Friday afternoon chasing approvals in Slack, your landing page should not promise "streamlined collaboration." It should say what the product changes: "Collect approvals without chasing people in Slack every Friday." That sentence is more likely to work because a customer gave it to you.

Use the same language in your first outreach messages, your product description, and your onboarding. If a prospect does not recognise themselves in the sentence, go back to the notes before you add another feature.

A one-week customer discovery plan

If you are stuck in research forever, put a hard edge around it.

Six conversations will not establish a market. They will, however, tell you whether you are hearing a repeated problem or trying to force one into existence. That is enough to make your next week less expensive.

Frequently asked questions

How many customer discovery interviews should a startup founder do?

There is no magic number. Start with ten to fifteen people in one narrow segment. Keep going while new calls are changing your view of the job. When you can predict the workaround, cost, and language before a call begins, you have enough to decide what to test next.

Should I show a prototype in a customer discovery interview?

Not at the beginning. A prototype is useful after you understand the current process, because it gives people something concrete to react to. Show it too early and the interview turns into a design review of your assumptions.

Can I interview friends and family?

Only if they genuinely do the job you are studying and will describe their own behaviour honestly. They can help you practise the questions. Do not count their enthusiasm as market evidence.

What if people describe different problems?

That is a useful result. Narrow the segment, or choose the most repeated and costly job. A product built for every version of a problem usually becomes a product nobody can describe.

The honest summary

Customer discovery is not the glamorous part of building. It is asking someone to show you the ugly spreadsheet, listening while they explain why the old tool failed, and accepting that the feature you wanted to build may not matter.

Do six conversations this week. Ask about the last time, the current workaround, and the cost. Count commitments, not compliments. Then build the smallest thing that removes the most repeated pain you heard.

When that thing is live, submit it on makers.page for a free, indexed product page. A launch gets more useful when the message came from the customers you spoke to first.

Related reading:

• How to find beta testers for your SaaS
• First 100 users without a marketing budget
• Your launch flopped. Here's what to do next