Startup College Startups 101

Module 3 of 9 · 18 minutes

Validate before you build

A customer discovery plan with your riskiest assumptions, interview questions, one simple test, a success threshold and a date to decide.

Why test before you build

The goal of this module is to help you find out, in weeks rather than months, whether the beliefs your idea depends on are true, before you spend serious time or money building it.

Most startups fail, and a common reason is building something people didn't want enough to pay for or change their habits for. Building feels like progress, so founders often start there. Yet a few weeks of conversations and small tests can tell you whether the problem is real, whether customers care enough to act and whether they'll pay, at a fraction of the cost.

Validating doesn't mean proving yourself right. It means trying honestly to find out where you're wrong while being wrong is still cheap. It can't tell you the idea will work, but it can stop you from betting years on a guess you could have checked in a month.

Before you read on, write down the one belief that would make you stop if it turned out to be false. Keep it nearby. By the end of this module, you'll have a plan to test it.

Find your riskiest assumption

Every idea rests on assumptions, and most stay invisible until you write them down. Write at least one for each of these areas:

  • Customer: this group exists and you can identify them.
  • Problem: they have this problem often, and it costs them something.
  • Solution: your approach works better for them than what they do today.
  • Payment: they, or someone, will pay enough to sustain the business.
  • Reach: you can find and win customers at a reasonable cost.
  • Building: you can make it with the skills and money you have.

Write each one as a statement that could turn out to be false. "Groomers like technology" is too vague to test. "Mobile dog groomers lose at least one booking a week to late cancellations" can be checked.

Then rate each assumption on two questions. How much does the idea depend on it: if it's false, does the idea fall apart or just need adjusting? And how much evidence do you have, from real behavior rather than your own hunch? Your riskiest assumption is the one the idea depends on most and has the least evidence behind it. Test that one first.

Beginners often test what's comfortable instead of what's risky. Building a prototype mostly tests whether you can build it, which is rarely the real question. Whether anyone will pay usually is.

A founder planning a booking app for mobile dog groomers lists 12 assumptions. She's confident she can build a booking calendar. She has no idea whether groomers will move from free messaging apps to a paid tool. If they won't, the idea falls apart, and she has no evidence either way, so that assumption goes to the top of her list.

Conversations that don't lead the witness

Asking "Would you use an app that does this?" feels like research, but the answers are close to worthless. People are kind, they don't want to discourage you, and saying yes to an imaginary future costs them nothing. Survey researchers have a name for this tendency to give the answer that seems nicest or most acceptable: social desirability bias.

The fix is to ask about what people have actually done, not what they think of your idea or might do someday. Past behavior is evidence; opinions and predictions mostly aren't. If you can, don't describe your idea at all until the end of the conversation.

Questions that tend to work:

  • "Tell me about the last time this happened. What did you do?"
  • "What did that cost you, in time or money?"
  • "What have you tried already? Why did you stop using it?"
  • "How often does this come up?"
  • "Who else gets involved when it happens?"
  • "What happens if you just leave it?"
  • "Who else should I talk to about this?"

Questions to avoid:

  • "Would you use…?" asks for a prediction.
  • "Do you like the idea?" invites politeness.
  • "How much would you pay?" produces a guess. Ask what they pay for today instead.
  • "Wouldn't it be easier if…?" is a leading question that hands them the answer.

Keep each conversation to 20 or 30 minutes and listen much more than you talk. When someone mentions a workaround, ask them to show it to you. Note their exact words, and write a short summary within an hour while it's fresh.

You ask a groomer about her last late cancellation. "Last Tuesday. Two of them, actually." You ask what she did. "Messaged my waiting list one by one. It took about an hour to fill one slot." In two minutes you've learned how often it happens, what it costs her and how she copes, without mentioning your idea once.

Who to talk to, and how to reach them

Talk to the specific customer you described, not to friends and family who want to support you. If someone you know genuinely fits the description, talk to them, but treat their answers as weaker evidence.

Aim for 10 conversations to start. Patterns often appear by then. If every answer points somewhere different, your customer description is probably too broad, so narrow it and keep going.

You don't need a startup network to find 10 people:

  • Go where they already gather: trade associations, industry forums, online groups, markets, trade fairs and professional meetups.
  • Walk in: for local businesses, visit at a quiet time of day and ask for 10 minutes.
  • Follow introductions: ask everyone you know for one person who fits, and end every interview by asking for another.
  • Use public listings: business directories, professional registers and member lists of industry bodies.
  • Send short, honest messages, like this one:

"Hi, I'm researching how independent dog groomers deal with last-minute cancellations. I'm not selling anything. Could I ask you a few questions for 20 minutes this week? I'll send you a summary of what I learn from other groomers."

Respect people's time and privacy. Keep to the time you asked for, send the summary you promised and ask before recording a call. Recording without consent is illegal in some places, and asking first is good manners everywhere.

Evidence, compliments and simple tests

Not everything positive you hear is evidence. The useful line to draw is cost: real evidence costs the other person something.

  • Commitments: they introduce you to their boss or colleagues, or agree on a start date for a trial.
  • Money: they pay a deposit, place a pre-order or pay for a pilot.
  • Time: they book a second meeting or try a rough version with their real work.

Compliments cost nothing: "Love it", "I'd definitely use that", "Let me know when it launches". They're pleasant to hear and predict very little. At the end of a good conversation, ask for a small next step that costs something, and treat a no as useful information rather than failure.

Three simple tests can turn conversations into harder evidence:

  • A landing page. One page that describes your offer and asks for an action: join a waiting list, book a call or pre-order. Send it to people who fit your customer description and count how many act. An email address is weak evidence; a booked call or a payment is much stronger. Say clearly that the product isn't built yet.
  • A concierge version. Deliver the result by hand to a few customers before you automate anything. The grooming founder could fill last-minute gaps for five groomers herself, by phone and messages, for a small monthly fee. If they won't pay for a person doing it, they're unlikely to pay for software that does.
  • A pre-order or paid pilot. Ask for money, or a deposit, before you build. It's the strongest signal and the hardest to ask for. Be clear about delivery dates and refunds, and check the rules on taking advance payments where you operate.

Before you run any test, decide what would change your mind. Write down the assumption, the test, a success threshold with a number, and what you'll do if you miss it:

"If fewer than 4 of the 20 groomers who see the offer pay a deposit, I'll stop pursuing a monthly subscription and test a per-booking fee instead."

Setting the threshold in advance matters because, once you've seen the results, almost any outcome can be read as encouraging. That tendency is confirmation bias, and a number written down beforehand is the simplest protection against it.

Finally, set a date to decide, usually two to six weeks away. On that date, choose one of three paths: continue, change one assumption and test again, or stop. Stopping is a legitimate result. Learning in a month that an idea won't work can save you years.

Do the work

Fill in the customer discovery plan builder at the end of this module, working through it in order:

  1. Customer: one sentence, specific enough to find 10 of them this week.
  2. Riskiest assumptions: three statements that could be false, riskiest first.
  3. Interview questions: at least five, each about past behavior. Delete any question that mentions your product.
  4. Where to find 10 people: named places, groups or people, not just "social media".
  5. What would prove you wrong: what you'd need to hear or see to drop your riskiest assumption.
  6. The test: a landing page, concierge version or pre-order, and exactly what people will be asked to do.
  7. Success threshold: a number, set now.
  8. Decision date: a real date, in your calendar.

You're done when you could start tomorrow morning: you know who you'll contact first, your questions contain no pitch, your test has a number attached and the decision date is in your calendar. When that date comes, update your pitch memo from the previous module with what you learned.

Your document for this module

Customer discovery plan. A short plan for testing your riskiest assumptions with real customers before you build. It's for you and anyone helping with interviews, and it ends with a date to decide.

Open the customer discovery plan builder (+100 XP)

Check yourself

1. A founder planning a booking app for mobile dog groomers lists these assumptions. Which should she test first?
2. Which interview question is most likely to give you reliable information?
3. After you describe your idea, which response is the strongest evidence of real demand?
4. You want to know whether busy parents will pay for weekly meal plans before you build any software. What's a good first test?
5. Why should you set a success threshold before you run a test?