Push notification re-engagement: what actually works

Push notifications used to be one of the most reliable re-engagement tools in mobile. They're not anymore. Notification permission rates have fallen, opt-outs have climbed, and the share of pushes that actually produce a session has dropped in almost every category of game.

The instinct in response is usually to send more of them — more pushes, more variety, more segmentation. That's the wrong reaction. The studios still getting real value from push aren't sending more notifications. They're sending fewer, with a much higher bar for what counts as worth sending at all.

What re-engagement push is actually competing with

A push notification arrives on a player's phone alongside messages from people they know, work notifications, news alerts, and updates from other apps they care about. It has to win attention against all of them. For most players, the game's push doesn't win — it's dismissed without thought, batched into a notification group, or causes the player to swipe the notifications away and never look at any of them.

Worse, a push that doesn't win attention but isn't dismissed individually trains the player to ignore future pushes from your app. The next notification, even if it's genuinely interesting, lands on someone who has already decided your notifications aren't worth opening. Cumulative effect matters more than per-push effect.

A push that loses attention isn't neutral. It actively degrades the value of every push that follows it, and at some volume, it pushes the player to disable notifications entirely. Once that happens, your re-engagement channel is gone — there's no graceful recovery.

The three pushes that earn attention

In practice, almost all push notifications that actually produce a session belong to one of three categories.

Time-sensitive game state. A push that tells the player their energy is full, their resource queue is ready, their daily reward is available, or their building has finished — anything the player was waiting for — works because it's information they wanted. The push is doing them a favour, not asking for something. This category is reliable across player segments because it serves the player's existing intent.

Specific upcoming events the player has signalled interest in. A push that tells a player a tournament starts in an hour, when that player has played the previous tournaments, works because it's relevant to a behaviour they've already shown. The same push to a player who has never engaged with tournaments is irrelevant noise. The targeting is what makes it work.

Genuine personal relevance. A push that's about something the specific player did — their friend overtook them on a leaderboard, their guild needs them for an event in progress, their unfinished daily streak ends in two hours — works because it's about them. Generic versions of these notifications addressed to all players don't carry the same weight.

What these three have in common is that the player is the protagonist of the message. The push exists to serve the player's current situation, not to insert the game into a moment the player wasn't thinking about it.

What kills push effectiveness over time

The patterns that degrade push performance are predictable, and most studios fall into at least one of them.

Sending too often is the most common. Each notification has a small but real chance of producing an opt-out. At low volume, that risk is paid for by the engagement the notifications generate. At high volume, the cumulative opt-out rate exceeds the marginal engagement from each push, and the channel shrinks over time.

Sending generic messages to broad segments is the second pattern. A push that reads as marketing copy — "Don't miss our weekend sale!" — performs worse than a push that reads as information — "Your store has a new offer based on your last battle." Generic pushes train the player to associate notifications with marketing, which they're already inclined to ignore.

Repeating the same message dilutes both versions. A player who got the same "come back to claim your reward" push three days running stops reading it on the fourth. The notification is recognised as repetition before it's read. New wording on the same underlying offer has diminishing returns fast.

Timing: the part most studios get wrong

Push timing tends to be designed around the studio's schedule, not the player's life. A push fired at 9 AM in your office time zone reaches different players in different states — some are asleep, some are at work, some are at school, some are commuting. A meaningful share of the population is in a state where opening the game isn't an option even if the push interests them.

Player-time-zone delivery is the baseline. Beyond that, the higher-value optimisation is to send each player's push at a time they've shown they're available to play. The window when a specific player typically opens the game is observable in your session data. Aligning push delivery to that window meaningfully outperforms sending all pushes at a chosen global hour.

This doesn't mean every push needs to be personally timed. But the major recurring pushes — the ones every player gets — benefit from individual-time-zone and individual-availability targeting more than from copy optimisation. A great message at the wrong hour underperforms a good message at the right one.

Measuring push without flattering yourself

The metric most studios report on push is the open rate — the percentage of recipients who tapped the notification. This number is misleading on its own.

What it omits is what happened to the players who didn't open. A push with a 5% open rate that drove that 5% to a meaningful session is a different product than a push with a 5% open rate that also caused 0.5% of recipients to disable notifications. The first push is net positive. The second push is net negative for any timeframe past the current campaign.

The full measurement of push is open rate, plus session quality among openers, minus the opt-out rate, minus the soft attrition rate — the decline in open rates of future pushes from the same audience. Studios that track only the first number convince themselves their push strategy is working while the channel itself is shrinking underneath them.

See how we measure push effectiveness over time →

The cleanest framing of push notification design is that it's a privilege the player extends to you. Each push is a small withdrawal from that account. Pushes that serve the player credit the account; pushes that interrupt them debit it. Studios that send push as if the channel is a free resource burn through it. Studios that treat the player's attention as scarce extend the life of the channel and the trust that makes it work.

Talk to us about re-engagement that doesn't burn the channel →