Founders experience week-one churn as a mystery: installs look healthy, the product works, and yet the retention curve falls off a cliff before day seven. Churn benchmarks across categories confirm the shape is normal, which is exactly why it gets ignored.

The mystery dissolves when you stop looking at the curve and start looking at the moments inside it. Week-one churn is not one event. It is a handful of specific, findable moments where a new user decides your product is not worth a second session. This post maps those moments so you know exactly where to look in your own data.

Key takeaways

• The largest share of week-one churn is decided before the first session ends. The first minutes carry more weight than the entire rest of the week.

• After the first session, churn concentrates in three places: the return trigger, the second-session experience, and the paywall's position relative to value.

• Instrument the first week as a funnel, find the steepest cliff, and fix that single step. Then measure the same cohort curve again.

What happens in the first five minutes?

The largest share of week-one churn is decided before the first session ends. The first minutes carry more weight than the entire rest of the week.

A new user arrives with a specific hope and a very small budget of patience. Every screen between them and the first real moment of value spends that budget: the signup wall, the permissions prompt, the empty state, the setup form.

In almost every app we have worked on at On the User, the single biggest retention lever sits in the first three to five minutes of the experience. Not in month two. Not in the feature set. In the opening minutes, where users either reach something valuable or quietly leave.

Watch session replays of brand-new users and count how many reach your product's core value in their first session. If the answer is under half, your week-one churn is not a retention problem in the classic sense. It is an arrival problem: most users never got to the part of your product worth coming back to.

Where exactly do users drop between day one and day seven?

After the first session, churn concentrates in three places: the return trigger, the second-session experience, and the paywall's position relative to value.

The first place is the return trigger. A user who had a decent first session still needs a reason to come back. Products with strong day-two retention almost always have something pulling the user back: unfinished progress, a notification tied to real value, an outcome that matures over time. If your product goes silent after session one, day-two retention becomes a coin flip.

The second is the second session itself. First sessions get all the design attention, and second sessions inherit whatever is left. The user returns, the guided experience is gone, and they face the raw product for the first time. If that moment feels like starting over, they do not book a third visit.

The third is the paywall. Not its price, its position. A paywall placed before the user has experienced real value converts a few and evicts the rest. The users who churn at the paywall do not report it. They just disappear, and the dashboard calls it week-one churn.

How do you fix week-one churn without guessing?

Instrument the first week as a funnel, find the steepest cliff, and fix that single step. Then measure the same cohort curve again.

Set up a funnel that covers the real first week: install, signup, first core action, second session, paywall view, day-seven return. Any analytics tool can do this; we use PostHog because session replays sit next to the funnel, and the replays are where the answers live.

If your product is AI-native, expect the curve to be steeper still, for reasons we cover in why AI products churn harder.

Find the steepest drop, watch five replays at that step, and fix only what you saw. Ship it, then compare the next cohort's curve to the last one. That is the whole method: one cliff at a time, measured honestly.

We wrote a free guide that walks through the four leaks that drive numbers like this, including the exact checks to run on your own product this week: get the free AI Leakproof Audit.