The people who left know something you don't

There's a hierarchy of honesty in user research, and it runs backwards from how we usually collect it. Your happiest, most engaged users are the least reliable narrators of your product. They like you. They want you to succeed. They will soften every answer without meaning to. The people who cancelled, quietly closed the tab and moved on? They have no politeness left to preserve. They are the most honest participants you will ever get.

Most teams measure churn as a number and stop there. The number tells you that something went wrong; only the people can tell you what, and usually in the first sentence. “We never got it set up.” “My manager didn't see the point.” “It worked fine until it didn't.” That's not sentiment, that's a diagnosis, handed to you by someone with nothing left to gain from sparing your feelings.

At one product I took over, support success was measured on active customers, and leavers simply disappeared from view. Nobody had called one in over a year. In my first month I called twenty of them. Twenty conversations produced a clearer picture of the product's real failings than the previous four quarters of surveys combined, and the fix that came out of it was smaller and cheaper than anything on the roadmap.

The practice is unglamorous and easy to keep: make leaving easy to study. Put one honest question in the cancellation flow, keep a running list of everyone who answers, and call a few each month. Not a post-mortem when the numbers dip, a standing habit, like brushing your teeth.

Your roadmap should be informed by the people who chose you. It should be corrected by the people who didn't. Only one of those groups will tell you the truth on the first call.