Why founders skip validation
Not laziness. Validation is emotionally expensive. You have an idea you like, and the entire purpose of the exercise is to find out whether you are wrong. Building, by contrast, feels like progress from day one and does not require anyone to tell you no.
There is also a speed argument that sounds reasonable: AI tools make building so cheap that testing by building is faster than testing by asking. Sometimes true. But building answers 'can this exist', while validation answers 'should this exist and will anyone pay'. A working product with no demand is a more expensive way to learn the same thing.
The reconciliation is that validation and building overlap. A concierge version delivered manually, or a landing page with a real checkout, is both a test and the beginning of the product. What matters is that a real person makes a real commitment before you commit the budget.
How to run a customer interview that produces signal
- —Ask about the last time the problem occurred, in detail. Specific memories are reliable; general opinions are not.
- —Ask what they did instead. Every real problem already has a workaround, and the workaround tells you the real competitor.
- —Ask what that workaround costs them in time, money or risk. If they cannot quantify it, the pain is not sharp enough to sell against.
- —Ask what they have already tried and stopped using, and why. This is where you learn what will kill your product later.
- —Never describe your solution until the end, and when you do, watch for whether they ask when they can have it or whether they say it sounds interesting.
Demand tests that actually mean something
Ranked by signal strength: a paid pre-order is the strongest, followed by a signed letter of intent with a named budget, then a paid pilot at a discount, then sustained use of a manual concierge version, then a deposit or waitlist with a card on file. A plain email waitlist is near the bottom — it costs nothing and predicts almost nothing.
The landing page test is popular because it is fast, and it is useful with one condition: it must show real pricing and lead to a real attempt to pay. A page that collects emails measures curiosity. A page with a checkout measures intent.
Set the threshold before you run the test. 'Three of fifteen conversations produce a paid pilot commitment' is a decision rule. 'See how it goes' is a way of guaranteeing you continue regardless of the outcome.
What to do when validation says no
Most of the time the answer is not a clean no. It is that the problem is real but the segment is wrong, or the problem is real and the pain is not expensive enough to command a price, or the segment and problem are both right and your proposed solution is not the one they want.
Each of those has a different next move. Wrong segment means run the same test with a different group, which is cheap. Insufficient pain usually means moving up market where the same problem costs more. Wrong solution is the best outcome available short of a yes, because the customers have just told you what to build instead.
And sometimes it genuinely is a no. Stopping at that point, two weeks and a few thousand pounds in, is the single highest-return decision available to an early-stage founder. It is also the one nobody celebrates.