"Why did you cancel?" is the wrong question

Every churn survey asks some version of the same question. Why did you cancel? What made you leave? How could we have done better?

And every churn survey gets back the same shrug. Too expensive. Not using it enough. Found something else.

Those answers feel like data. You can put them in a pie chart. But you can't do anything with them, because none of them tell you what actually happened.

People are bad at explaining themselves

Not dishonest. Bad at it, the way everyone is.

When you ask someone why they did something weeks ago, they don't retrieve a stored reason. They construct one, on the spot, to sound sensible to a stranger. "Too expensive" is the most socially safe answer available. It ends the conversation without anyone feeling judged.

The reason they actually left is usually more specific and more boring. They hit a bug on a Tuesday and never came back. Their one power user went on leave. The thing they needed lived three clicks deeper than they expected and they assumed it didn't exist.

None of that arrives when you ask why.

Ask what they did instead

The fix is to stop asking for explanations and start asking for history. People are unreliable narrators of their motives and reliable narrators of their behaviour.

Compare these two:

Why did you cancel?

What did you do the last time you needed to run a survey, instead of using us?

The second one is almost impossible to answer vaguely. They either name the thing they used, or they tell you they stopped needing it. Both answers are useful. "I just used a Google Form and eyeballed it" is a product roadmap. "Too expensive" is not.

This is the core move from The Mom Test: ask about specifics in the past, not generalities in the future. It works on churn surveys better than almost anywhere else, because a cancellation is a real event with a real story attached.

Four questions that actually surface the reason

1. What were you trying to get done when you first signed up? Anchors the cancellation to the original job, which is often not the one you think you're solving.
2. What did you do the last time you needed that, instead of using us? The real competitor is frequently a spreadsheet, an intern, or nothing at all.
3. What was happening right before you decided to cancel? Gets you the trigger. There's almost always a specific week.
4. What would have had to be true for you to stay? The only forward-looking question worth keeping, and it only works once the first three have put them back in the actual situation.

Notice that none of them contain the word "why".

Read the words, not the categories

The other thing that quietly ruins churn research is bucketing. You collect two hundred open-ended answers, sort them into five tidy categories, and count.

The categories are where the information goes to die. "Too expensive" and "I never got it working, so it wasn't worth the money" land in the same bucket and mean completely different things. One is a pricing problem. The other is an onboarding problem wearing a pricing costume.

Read the sentences. The specific ones repeat more than you'd expect, and the repetition is the finding.

Timing matters more than sample size

Send it within a day of the cancellation. Memory for the trigger decays fast, and after a week you're back to getting constructed reasons.

You need fewer responses than you think. Patterns in open-ended answers usually stabilise somewhere around fifteen to twenty. Below that, treat what you're seeing as a hypothesis rather than a conclusion. Above it, you're mostly confirming what you already read.

Twenty people telling you what they actually did beats two hundred picking from your dropdown.