Interview your churned users, not just the ones who stayed

Think about who you normally do research with. Active users, engaged users, power users. People who, by definition, find the product worth their time. Every insight you collect passes through that filter, and the filter only lets through people the product already works for.

The people it stopped working for are a different data set entirely, and almost nobody collects it. Which is odd, because a churned user has three properties a current user never has: they made a real decision with real consequences, they remember exactly what triggered it, and they gain nothing from softening the answer.

The dropdown is where the honesty dies

Most cancellation flows do technically ask why. They ask with a dropdown: too expensive, missing features, switched to a competitor, other. Every answer was written by the team, which means the survey can only ever confirm what the team already suspected. The one thing you needed - the reason you did not think of - is filed under "other".

Groove, the support tool now called Helply, ran the clean experiment on this. They replaced their exit survey's structured options with a single open-ended question, and the response rate went from 1.3% to 10.2%. Same users, same moment, roughly eight times the answers. Then they changed the wording from "Why did you cancel?" to "What made you cancel?" and responses nearly doubled again, to 19%.

That second jump is the interesting one. "Why did you cancel" quietly asks the user to justify themselves. "What made you cancel" asks them to point at a thing. People do not want to defend a decision, but they are happy to describe a moment.

What to ask instead

The survey is only the doorway. The real value is in the five or six churned users you get on a call, and what you ask them once you are there.

The instinct is to ask "what feature would have kept you?" Resist it. Users are bad at designing solutions and excellent at describing what actually happened - the same reason "what do you want?" leads straight into the Product Death Cycle, and churn does not change that. A churned user inventing a retention feature on the spot is producing exactly the kind of made-up answer you would get from anyone else.

So ask about the past, concretely:

• "Walk me through the day you decided to cancel."
• "What were you trying to get done right before?"
• "What did you switch to, or start doing instead?"

The first question finds the trigger. The second finds the job the product was failing at. The third is quietly the most valuable, because "nothing, we just stopped" and "we moved to a spreadsheet" and "we bought a competitor" are three completely different diagnoses that a dropdown would have merged into one row.

This is the same discipline as any good interview - ask about the last time, never the next time - and if interviewing is new territory for your team, talking to five customers this week covers the mechanics. Churned users just happen to be the segment where the discipline pays best, because there is no relationship left to manage.

Five stories make a roadmap, one makes an anecdote

One vivid cancellation story is dangerous. It arrives with a face and a quote, it gets retold in the next planning meeting, and suddenly the roadmap has a feature on it because one person left angry.

The unit of insight is not the story. It is the pattern. Five people independently describing the same broken moment is a finding you can build on. Five people describing five different moments is a signal of something else - possibly that you are serving an audience you should not be, which is a who problem rather than a what problem.

This is also the fastest route to understanding a soft retention number. A curve tells you that users leave and when. It never tells you why. If your retention sits under the benchmark for your segment, a week of churned-user interviews will explain more of the gap than a quarter of dashboard archaeology.

How to actually reach them

Response rates from churned users are real but not free. Three things move them:

Ask fast. Within days of cancellation, while the moment is fresh. A churned user from six months ago has already rewritten the story into something tidier than what happened.

Make it human and small. A short personal note from a founder or PM, asking for 15 minutes, no survey link. The pitch is honest: you left, that is fine, help us understand what broke.

Pay if you need to. A gift card is a fair trade for the single most honest interview in your research mix.

Then treat what you hear as input to discovery, not as a to-do list. The cancellations are clues to an opportunity, and the opportunity still needs the normal validation work before it earns engineering time. Running that loop end to end - from churned-user signal to tested solution - is exactly what we practice in a Product Discovery Workshop.

So this week, pull the list of the last ten cancellations and send ten short notes. The users who stayed can tell you what to protect. Only the ones who left can tell you what is broken enough to leave over.