Winning Back Users Who Went Quiet

A Quiet Month Makes a Stranger

Memory of software decays fast. After a few weeks away, a returning user is close to a new install with one important difference: they already decided once that this was worth trying, and something made them stop. That gives you more to work with than a new user gives you, and less time to use it.

So the return session has a job: re-establish what the app was for, in one screen, and make the next action obvious. It is not primarily a selling moment. Sell on the way back in and you answer a question the person has not asked yet.

A returning user is not a new user and not a current one. They are the only group that has both forgotten your product and chosen it before.

Two Ways to Notice Someone Drifted

There are only two families of signal here, and they differ in when they fire.

The calendar signal is the distance between now and the last time the person opened the app. It is simple to compute, needs no history beyond one timestamp, and is impossible to misread. It also fires late by construction: you learn someone drifted only once they have already drifted, and only if they come back.

The behavioural signal is the shape of use before the gap — sessions getting shorter, a core feature abandoned, a weekly rhythm becoming a monthly one. It can fire while the person is still around, which is the only time intervening is cheap. It costs you event instrumentation and a definition of “at risk” that you have to maintain.

Most teams should eventually run both: the behavioural one to prevent the gap, the calendar one to handle the returns you did not prevent. If you can only build one this quarter, the rest of this post is about picking which.

Where a Calendar Gap Is the Right Signal

• Use is seasonal or occasional by design. Tax software, travel planners, ski apps. A long gap is the normal shape of the product, so the gap itself is the only honest marker of "back again".
• You have no behavioural history worth trusting. Young products, or products that only just started sending events. A date is a signal you already have; a model of declining engagement is one you do not.
• The action you want to take is the same for everyone. If the plan is "remind them what this app was for", fine-grained segmentation earns nothing.

There is a second, quieter argument for starting here: the calendar version can be built and judged in a week. One number, one screen, one measurement. If it does nothing for your product, you learned that cheaply and the behavioural work is still ahead of you, not wasted.

Where the Gap Is Just Noise

• Daily-use apps. For a habit tracker, a week away is not drift — it is a holiday. Treating it as drift greets your most loyal users with a win-back screen.
• When you can see the slowdown coming. Sessions getting shorter, one feature abandoned, streak broken. All of that happens before the gap opens, which is when intervening is still cheap.
• Shared devices and long trips. A date cannot tell "lost interest" from "was on a boat". The copy has to survive being wrong about that.

The test is one question: in your product, does a gap of this length mean something has gone wrong? If the answer is “sometimes”, the threshold is too tight. Widen it until the answer is “usually”, and if no width makes it usually, a behavioural at-risk signal is the tool you actually wanted.

What the Return Moment Should Carry

Three things, in this order of importance: what they had, what changed, what to do next. “Your six saved routes are still here” is worth more than any offer, because it answers the question the person is actually holding. If something genuinely improved while they were away, that is the one sentence of news worth spending.

Measure it as a habit question, not a conversion one. Did the returning session turn into a second session that week? Conversion on a win-back screen can look excellent while the person disappears again a day later, and you will have bought one payment at the cost of the relationship. Watch uninstalls in the days after, too: a pushy return screen can finish what the gap started.

One boundary to be clear about, because it decides whether this approach fits you at all. This is an in-app mechanism. Your app measures the gap — it reads its own last session timestamp and writes the result as a user attribute — and Amply uses that to route the session into a different flow. Amply sends nothing: no push notification, no email, nothing on a lock screen. If the person never opens the app again, nothing here reaches them, and reaching them is a different tool's job.

Once the signal is chosen, the reactivation campaign, step by step covers the attribute your app writes, the conditions, and how the flow stops matching.