A retention loop is four moving parts, and the whole design question is which one is currently broken. Trigger, action, reward, return: if a person is not coming back, one of those four is failing, and each one fails for a different reason and needs a different fix. Most teams change the wrong one.
The four parts of the loop
Most broken loops break at the reward. The action happens, the person leaves, and the credit arrives in a notification the next morning. Nothing connects the two events, so the next time the trigger fires, the person has to remember why they cared.
The ones that break at the return are the most expensive. A person who opens, acts and gets paid but has no reason to come back will do it once and drift. The loop needs something tomorrow that is worth slightly more than today - a streak that is still alive, a tier that is within reach, a mission that has a deadline.
Choosing a trigger you can afford
Triggers are the part of the loop most likely to be over-used, because they are the cheapest to build and the most expensive to keep. A notification that produces a session is worth a great deal. A notification that produces an uninstall costs the user you already had.
The discipline that holds up is a simple test: a notification is worth sending only if it carries information the person cannot get by opening the app. A streak that expires tonight is information. A badge saying three offers are waiting is not - the person can see the offers when they open the app.
The cost of getting this wrong is not just churn. It is the withdrawal of push permission, which is granted generously at install and taken back quickly once notifications feel like noise. Permission you have spent is not available again, so the notification budget of a loyalty programme is a finite resource rather than a dial.
The action has to fit in the first thirty seconds
The action is the step people skip designing, because the app already has an offer wall and it seems obvious that the offer wall is the action. Offer walls are a catalogue, not an action, and a person choosing from a catalogue is making a decision every time.
A small repeatable action - one tap, one minute, something that can be finished without choosing - becomes a reflex. Refracts survive a bad week. Decisions do not. The bigger action in the app can be the install or the survey; the small one is what brings the person back tomorrow.
This is also where the notification and the loop meet. A person who opens on a streak reminder and can complete something in thirty seconds is far more likely to be back the next day than one who opens to a catalogue and has to decide.
The arithmetic a loop has to beat
A worked example rather than a benchmark, because published benchmarks for this are mostly invented. With 100,000 daily actives, 30% returning on day one, 12% of those on day seven and 4% still active on day thirty, you have 30,000 on day one, 3,600 on day seven and 1,200 on day thirty. Move those three numbers to 35%, 16% and 6% with the same audience and you have 35,000, 5,600 and 6,000 - 4,800 more people still there on day thirty.
The number to watch alongside it is payback period: how long until the revenue from a user covers what it cost to acquire and keep them. A day-30 figure that looks acceptable can hide a six-month payback, which is longer than many of these apps exist.
Loyalty or rewards, or both
A loyalty loop and a rewards loop are different shapes and they pull in different directions. The loyalty loop wants a small daily action. The rewards loop wants a bigger task, because the reward has to be worth interrupting for.
Running both is normal and works, as long as the two balances stay distinct. A person who returns for a streak and finds a survey worth a couple of dollars is better monetised than one who returns for neither - but the moment the offer wall becomes the main event, the loyalty programme is competing with itself for the same session.
The EY 2025 US Loyalty Market Study puts a useful frame on why the relationship has to be built rather than assumed: more than half of shoppers engage with their loyalty programmes weekly, yet a significant portion report feeling little to no emotional connection with them. Their relationship is transactional. The mechanics are what turn the same weekly engagement into something people defend.
Related reading
Run rewards in your app
Tell us the platform and roughly how many daily active users you have, and we will tell you whether rewarded games, surveys or both are the better fit.
Questions
What is a retention loop?
The cycle that brings a person back: something triggers them to open the app, they take a small action, they get a reward that lands while they are still there, and they have a reason to come back before the cycle closes. The four parts each fail differently, so diagnosing a retention problem means finding which part is broken rather than adding features.
How often should a loyalty app send notifications?
Only when the notification carries something the person cannot get by opening the app. A streak at risk, a mission nearly complete and a reward ready to redeem are reasons. A count of available offers is not. Push permission is granted generously at install and withdrawn quickly once notifications feel like noise, so a programme that spends it on generic reminders has spent the only asset it had.
How do I calculate LTV for a loyalty app?
Multiply the days a user stays by the value they generate per day, and separate in-app spend from rewarded revenue because they respond to different mechanics. The number to watch with it is payback period: how long until a user's revenue covers what it cost to acquire and keep them. A strong day-30 retention figure can still hide a payback period longer than the app's expected life.
Should a loyalty app include a rewards wall?
It can, and many do, but keep the two balances and the two purposes apart. The loyalty loop wants a small daily action; the rewards loop wants a bigger task. If the offer wall becomes the main event, the loyalty programme starts competing with itself for the same session, and the person stops coming back for the reason they came back.