What people say they do, and what they do

There is a constant gap between what people say they do and what they actually do, and most of the products I have watched fail up close got lost in it.

It helps to understand this is not about lying. When somebody tells you they would check their metrics every morning, they mean it: they are describing the person they would like to be. What they cannot describe is the person they become on a Tuesday at six, with fifteen things outstanding and no energy left to open another tool.

Why statements predict so badly

When you ask about future behaviour, a person has to simulate a situation that does not exist yet, without the tiredness, without the rush and without the other nine priorities competing on that day. It is an optimistic simulation by construction.

There is also a cost asymmetry that makes it worse. Saying yes costs nothing in a conversation; saying no requires justifying yourself. So the bias does not spread evenly: it piles up in the same direction every time, towards yes.

What to look at instead

What already happened has the advantage of having happened. A workaround built by hand in a spreadsheet, a folder of neatly sorted screenshots, an alarm set on a phone to remember something: all of it is evidence that somebody is already paying a cost for the problem.

Then there is behaviour with friction. Any action that costs something, leaving an email, reserving a place, paying up front, installing anything, is worth more than a hundred affirmative answers, because it carries a price that a statement does not.

And there is repeated behaviour. Of everything you can measure in a product, the only thing that convinces me is somebody coming back a second time without being reminded. Curiosity explains the first visit. Only usefulness explains the second.

Where this hurts most

The classic case is the product everybody likes and nobody uses. The conversations were enthusiastic, the survey numbers were good, people asked to be told when it launched. Then it launches and the usage curve flattens inside the first week.

What happened is not that people lied. It is that interest got measured and read as intent. Anything well told generates interest; intent only shows up when something successfully competes with what that person already does.

How I apply it

Before building I look for the cheapest test that demands a real action. A page describing the product with a single call to action does the job nearly every time: you do not need the product, you need to know whether anybody moves.

And once the product exists, I watch two things above all the rest: how many people come back on their own, and what the ones who stay actually do, which almost never matches what I expected them to do.

Questions people ask about this

So surveys are useless?

They are useful for measuring what already happened, not for predicting what will. Asking what did you do last time is useful; asking what would you do next time is not. The difference between those two questions is the difference between a fact and a wish.

How do I measure real intent with no product yet?

Put a small cost in front of it. An email address, a waiting list, a reservation, a refundable token payment. The amount does not matter, what matters is that a barrier exists and only people who genuinely want it cross.

Which metrics do I watch once the product works?

The one that tells me most is spontaneous return: how many people come back without being reminded. After that, what path the ones who stay take. Almost everything else describes your marketing team activity rather than the usefulness of the product.