Software development is getting absurdly fast. Features that took months now take days. Whole systems get built in weeks. That is not a forecast. That’s Tuesday.
And it changes the conversation with the customer.
IT departments and digital agencies now say: “We’ll just ask the customer what they want, then build it.” It sounds sensible. It isn’t.
When building is cheap and fast, the temptation is enormous. Ask what the customer wants. Build it. Ship it. Repeat. You feel productive. The customer feels heard. Everyone is happy, until nobody uses it.
The problem is the same as always. Customers know what they experience. They rarely know what they need. Henry Ford said it over a hundred years ago: if he had asked people what they wanted, they would have said faster horses.
That trap hasn’t gone away. It has become more dangerous. Because now you can actually build the faster horses, in a week. With a polished interface and an onboarding flow on top.
Stop asking what they want
So what do you do instead? You stop asking “what do you want?” and start asking “what do you do today?”
Rob Fitzpatrick describes it in The Mom Test: good questions are about the customer’s life and behavior, not about your idea. The difference is subtle but decisive.
“Can you tell me how you usually solve this?” is a good question. “Do you think this is a good idea?” is a bad one. “What else have you tried?” is gold. “Would you buy a product that did X?” is worthless. “Walk me through the last time it happened” opens a door. “How much would you pay for X?” closes it again.
Good questions uncover what the customer actually does. Bad questions ask the customer to predict their own behavior, and humans are notoriously terrible at that. What you’re really asking is: “Hey, could you knock out a market analysis for me while you finish your coffee?” No. They can’t.
Discovery has not become less important
With the superpowers we’ve gained as builders, it’s tempting to skip discovery. Why spend two weeks understanding the problem when you can build a solution in three days?
Because the most expensive feature is not the one that takes long to build. It’s the one that should never have been built. And with AI in the toolbox, you can now build the wrong things at a pace that would have made even the most optimistic project manager from 2019 weep with envy.
Discovery has not become a nice-to-have. It has become the most important part of the whole process. Not because the world got slower. But because everything else got faster.