Why your second-month users leave

A user is going to cancel next month, but they don't know it yet. They're still paying. They're still on the mailing list. They're still in your database, listed as active. The decision was made three weeks ago, when they tried to do something in your product, couldn't figure out how, and quietly closed the tab. You didn't see them try.

The cancel button is a receipt

The choice to cancel is not a decision. It's a recognition. By the time the user clicks the button, they've already left in every way that matters — they've stopped opening the dashboard, stopped finishing the flows they used to finish, stopped having the product be the thing they reach for. The button is an administrative acknowledgement that the relationship ended weeks ago.

This is why exit surveys tell you almost nothing. By the time someone is willing to fill one out, they're no longer trying to figure your product out. They're tying up the last loose end before they leave. What they wrote in the comment box is a polite version of the conversation they had with themselves a month earlier — and that conversation, you missed.

Watch the silence

The signals that matter are quiet ones. A weekly login that stops being weekly. A flow the user used to finish, half-completed and abandoned. The trial that is on day twelve with no setup done. The team that invited three teammates in week one and zero since.

You don't need a dashboard with forty retention metrics. You need three or four that fire alerts when they move the wrong way, watched by someone whose job is to act on them. Catch the silence before it hardens.

Outreach is product work

When you see drift, don't send a "we miss you" email. That email tells the user, in writing, that they've already left. You're confirming the diagnosis.

Send a person, or send a question. "Hey, noticed you tried setting up a webhook twice last week — got stuck somewhere?" gets answers. "We've been thinking about you" gets unsubscribes.

At-risk outreach usually lives in customer success — a separate team, separate playbook, separate from product work. That's backwards. The drift signal is product data. The response is a product question. The fix is almost always a product change. Outreach without willingness to ship something is theatre.

Cohorts beat aggregates

Monthly churn rate is a lagging indicator and a smoothed one. It tells you a percentage. It doesn't tell you whether the percentage is getting better.

The number that does tell you that is cohort retention by week. Does the cohort that signed up in March still look healthier at week 8 than the cohort from December did at week 8? If yes, you are improving the product. If no, your aggregate churn is staying flat because the new customers are buying time for the old ones.

The dashboard lead is usually monthly churn — which moves slowly, can be flat for the wrong reason, and tells you little about what to do. The right leading number is the cohort curve.

Doing any of this is mostly a question of which data is in front of you. Have three or four drift signals visible to someone who can act on them — a login cadence dropping, a setup flow abandoned, a key action skipped. Put cohort retention on the dashboard instead of aggregate monthly churn. The rest of the changes follow from those two moves.

The point of any of this is not to lower churn by a point. It's to find the second-month users who are about to leave, while there's still time to do something they haven't already given up on.

The churn number is just the count of users you stopped paying attention to first.