Customer Development: Get Out of the Building

A technical founder has an idea. They love it. They sit at their desk for six months and build it. They launch. And nobody comes.

The product works perfectly. It just solves a problem that no one will pay to fix. This is the single most common way good builders fail, and it is almost completely avoidable.

There is a cure, and it has a slogan you should tattoo on your brain: get out of the building.

Why this matters

Your scarcest resource is not money. It is the months of your life you can only spend once.

When you build first and ask later, you bet all of those months on an untested guess. If the guess is wrong, you find out at the end, with nothing left to fix it.

Inside your building, your desk, your code editor, your own head, there are only guesses. The facts that decide whether your business lives or dies live outside, in the heads of real customers. Your job is to go get those facts cheaply, before they get expensive.

The good news: you do not have to invent any of this. Two famous thinkers already mapped it out, and their work is now the standard playbook in Silicon Valley.

• Steve Blank, a serial entrepreneur and Stanford and Berkeley teacher, created Customer Development in his book The Four Steps to the Epiphany (2005), along with the phrase “get out of the building.” His point: building a product is a known engineering process, but finding customers is a process of discovery that has to happen face to face.
• Eric Ries, a founder who studied under Blank, turned these ideas into a repeatable loop in The Lean Startup (2011). He gave us two terms worth knowing.

Build-Measure-Learn is a repeating cycle. You build the smallest thing that tests an idea, you measure how real customers react, and you learn whether your guess was right. Then you go around again.

Validated learning is progress measured not by “we wrote more code” but by “we learned something true about customers, proven by their behavior.” Writing 5,000 lines of code nobody wants is zero progress. Learning that nobody wants it, after one week of conversations, is real progress.

Think of building before talking to customers like cooking a huge dinner for 50 guests without asking if anyone is hungry, what they like, or whether they are even coming. The smart cook asks first, then cooks. Customer Development is asking first.

The four steps, in plain terms

Blank splits the early life of a company into four steps. As a solo founder you mostly live in step one for a long time, but it helps to see the whole map.

1. Customer Discovery. Find out if the problem you imagine is a problem real people actually have, badly enough to want a fix. This is where you spend almost all your early time.
2. Customer Validation. Prove people will actually pay, and find a repeatable way to sell. Can you get a few real customers using a sales process you could run again?
3. Customer Creation. Now that selling works, spend money to create demand and bring in more customers through marketing, ads, and growth.
4. Company Building. Switch from “search mode” to “run a company mode”: hire teams, build departments, scale up.

Here is the deep insight. Steps 1 and 2 are a search for a business model. Steps 3 and 4 are execution of one you have already found.

Most startups die because they jump straight to execution, hiring and spending and scaling, before they have actually found something that works. And the model expects you to loop back. You will go around discovery and validation several times before it clicks. Messing up is fine, as long as you plan to learn from it.

Two finish lines, in order

You will hear two phrases everywhere. They are simply two milestones, and the order matters.

Problem/solution fit is the first milestone. You have proof that a real, painful problem exists and that your proposed solution would plausibly fix it. You move from hoping people want this to having evidence they do, before building the full thing.

Product/market fit is the bigger, later milestone. Your actual, built product satisfies a real market so well that demand pulls you forward. People keep buying, keep using, and tell others. This is the famous moment a startup is “working.”

You cannot skip to the second. Problem/solution fit is the foundation product/market fit is built on, and you reach it cheaply, mostly through conversations.

  TALK to customers ──reach──> PROBLEM/SOLUTION FIT
     (cheap, fast)             "Yes, real pain. Fix makes sense."
                                          │
                                     then BUILD
                                          │
                                          v
  USE real product ──reach──> PRODUCT/MARKET FIT
    (the built thing)          "The market pulls; demand is real."

Turn your fuzzy idea into testable guesses

A hypothesis is just a fancy word for a specific guess you can check. Right now your idea is a big, fuzzy belief. The first move is to break it into small, sharp guesses you can test by talking to people.

Do not write: “Print shops need better software.” That is too vague to test. Break it down instead:

• Who exactly has the problem? “Owners of small print shops with 2 to 10 staff.”
• What painful thing happens today? “They re-key online orders into their machines by hand and make costly mistakes.”
• How much does it hurt? “They lose time and money on reprints every week.”
• What do they do about it now? “They use spreadsheets and sticky notes.”

A good hypothesis is a sentence you can mark true or false after 10 conversations. “Small print shop owners lose at least an hour a day re-typing online orders by hand” is a hypothesis. “My product is great” is not.

Why this happens before and during building, not after

Here is the trap: “I’ll just build it, then ask people.” By then you have already spent the months you can never get back.

Customer Development flips the order. You test the riskiest guesses with cheap conversations first. Then you keep talking to customers while you build, so every feature is steered by real feedback.

Talking to customers is never a one-time phase you finish and tick off. The founders who win make it a permanent weekly habit, even after launch. It runs forever, beside the code.

Common misconceptions

“I’ll just send out a survey.” Surveys and Customer Development are not the same thing, and surveys are weaker for an early idea.

Market research survey	Customer Development conversation
Many people, shallow, fixed questions	Few people (10 to 15), deep, follow-up questions
Asks for opinions and “would you?”	Asks about real past behavior
You cannot dig into a surprising answer	You ask “why?” and discover the unexpected
People say what sounds nice	You hunt for facts that cannot be faked

This connects to the single most useful book on customer conversations, The Mom Test by Rob Fitzpatrick. The title comes from a simple truth: even your mom will lie to you about your idea, to protect your feelings. So you must ask questions even your mom could not lie about. Fitzpatrick gives three rules:

1. Talk about their life, not your idea. The moment you pitch, they start being polite.
2. Ask about specifics in the past, not opinions about the future. “Would you use this?” is worthless. “When did you last hit this problem, and what did you do?” is gold.
3. Talk less, listen more.

The whole thing rests on one line: opinions are worthless; behavior is everything. “I would totally buy that” is a polite opinion. “I paid $200 last month for a tool that didn’t work” is a hard fact. Chase the facts.

“Talking to users doesn’t scale, so it’s a waste of time.” Early on, unscalable is exactly right. Paul Graham of Y Combinator wrote a famous 2013 essay, Do Things That Don’t Scale, arguing you should recruit your first users by hand, one at a time, and delight them in ways you could never repeat for a million people. The Airbnb founders flew to New York and went door to door signing up hosts, even taking professional photos of their apartments themselves. Totally unscalable. Exactly right for the start.

A real example: testing one assumption with 12 people

Say you are building software for small print shops. Your riskiest guess is: “Shop owners lose real money re-typing online orders by hand.” Here is how you test it without writing a line of code.

You find 12 print shop owners by calling, emailing, visiting, and asking for introductions. You meet each for 20 minutes. You do not pitch. You ask Mom-Test questions.

You: “Walk me through what happens when an order comes in from your website.”

Owner: “Oh, it’s a nightmare. I print the email, then type it all into our machine software again.”

You: “When was the last time that went wrong?”

Owner: “Last Tuesday. I fat-fingered a quantity, printed 500 instead of 50. Ate the cost myself.”

You: “Wow. How often does something like that happen?”

Owner: “Couple times a month, honestly.”

You: “What have you tried to fix it?”

Owner: “I made a checklist. Doesn’t really help.”

You: “Have you ever paid for anything to solve this?”

Owner: “I looked at a tool once but it was built for big factories. Too expensive, too complex.”

Notice what you got: a real story, a recent date, a dollar cost, a current workaround, and proof they already tried and failed to solve it. That is validated learning.

After 12 of these, you will see a pattern. Maybe 9 of 12 describe the same painful re-typing. Now your guess is validated. As a bonus, you learned the exact words real owners use, “nightmare,” “fat-fingered,” “built for big factories,” which become your marketing copy later.

How to use this

1. Write down your riskiest guess as one testable sentence. Pick the belief that, if wrong, kills the whole idea. Phrase it so you could mark it true or false.
2. List who has the problem, specifically. Not “businesses.” Name the exact person: their size, role, and situation.
3. Find 10 to 15 of those people. Call, email, visit, ask friends for introductions. Aim for short 20-minute chats.
4. Do not pitch. Ask about their life and their recent past. “Walk me through the last time this happened.” Then ask “why?” and stay quiet.
5. Hunt for behavior, not opinions. What did they actually do, pay, or try? Recent stories with dates and dollars are the prize.
6. After each chat, write three things: the most surprising thing you heard, any real story with a date or amount, and whether your guess now looks more true or more false.
7. Look for the pattern, then loop. If the pain is real and repeated, refine your guess and go again. If it is not, you just saved yourself six months. Keep the habit weekly, forever.

        ┌──────────── THE LOOP (repeat forever) ───────────┐
        │                                                   │
   WRITE a guess  ──>  GET OUT, talk to 10–15 people  ──>  LEARN
   (hypothesis)        (Mom-Test questions)               (true/false?)
        ^                                                   │
        └────────── refine the guess, go again  <───────────┘

Conclusion

If you remember one thing, make it this: the facts that decide your fate are not at your desk. They are in customers’ heads, and the only way to reach them is to get out of the building and ask.

Y Combinator boils the early founder’s whole job down to two tasks: write code and talk to users. For most technical founders, the uncomfortable half of that sentence is the more important one, and it is the half you can start today, before writing another line.

But getting people to talk is only the first hurdle. The harder, quieter skill is getting them to commit, to hand over their time, their email, or their money. That is where a real conversation turns into a real customer, and it is where we go next.