Every loyalty program we have ever looked at has the same shape. Lots of customers join. Fewer than half come back for a second visit. The ones who reach five visits mostly stay. And in between, the business sends the same Friday message to everyone, whether they were in yesterday or have not been back since June. Today we are launching something to fix that: AI personalization for loyalty programs, built into Wally. A personalization agent reads the visits on your stamp or points card, plans a journey for each customer, runs it inside the guardrails you set, and reports the lift every Monday. As far as we know, it is the first wallet loyalty platform with an agent like this built in. Here is what it does, what it does not do, and how to switch it on. The full feature page is at AI personalization and journeys.
What AI Personalization for a Loyalty Program Actually Means
Personalization has been a word brands with data teams use. For a café with two branches it has meant a birthday message and not much else. The agent changes who gets to do it. It works from the one thing every Wally card already records: a timestamped visit each time a stamp is added or points are earned. From that it knows, per customer, how often they usually come, which day and hour, how many stamps they have to go, whether an earlier offer brought them back, and how far past their own usual gap they are right now.
That last point matters more than it sounds. Most win-back campaigns fire at a fixed number of days, usually thirty. A weekly regular who misses two weeks is already drifting, and a monthly customer who has been away for three weeks is perfectly on track. The agent uses each customer's own cadence, so the weekly regular is caught in days and the monthly customer is left alone.
The Journeys It Runs
A journey is a trigger, an offer and a reason, run one customer at a time. When you set the agent up it proposes a handful of them for your card, each with the numbers behind it. The ones it reaches for first on a café or salon card:
Second-visit treat. The day after a first visit, a free reward that expires in three days, sent at the customer's usual hour. Why: fewer than half of first-timers return, and the second visit is the one worth buying.
Drifting away. Three days past that customer's own usual gap, two stamps or a free drink for customers who have been gone a long time. Why: their cadence, not a calendar.
One stamp away. Right after the ninth stamp, a message only. Why: it costs nothing, and completing the card lifts the fifth-visit rate, after which most customers stay.
Every offer is delivered as a pass update on the lock screen, in Apple Wallet or Google Wallet, so there is no app for the customer to install and no inbox to compete in.
You Answer Two Questions. That Is the Setup.
The agent opens with what it found on your card: active cards, your second-visit rate against cards like yours, how many customers are drifting today, when your regulars come. Then it asks what to work on first, and how generous it may be per customer: your card's reward, a stamp or two, or messages only. It writes the plan with the reason for each journey, shows you the first message as a customer will see it, and goes live the next morning when you say yes. There are no segments to define and no workflow canvas to build. If you prefer, any journey can be edited by hand.
Guardrails First, Autonomy Second
We built the agent for businesses that do not want a robot giving away free drinks. Some rules are always on: one offer per customer a week, at most four a month, sends only inside your window, nothing to customers who are on track, and customers who opted out are never included. There is a monthly cap on rewards given.
On top of that, six permissions you switch on or off. By default the agent may move send times inside your window, pause a journey that is not bringing people back, and rewrite a message in your card's voice with the same facts. Changing what a customer is given, changing an expiry, adding a journey or changing the holdout always waits for your yes. Anything it does alone can be undone with one tap.
Lift You Can Defend
This is the part we are proudest of. Every journey keeps a holdout: by default one in ten customers who qualify is kept as a control and gets nothing. Lift is the return rate of the customers who got the offer minus the return rate of the holdout, measured over the same fourteen days. In the sample programme on the feature page, the second-visit treat brought back 41% of customers against 27% without it, a 14-point lift. If a journey has not beaten its holdout after thirty sends, the agent pauses it and tells you. Open and click rates never appear anywhere, because they do not tell you whether anyone came back. If you want the wider numbers on why repeat visits matter, see our customer retention statistics.
Every Offer Keeps Its Reason
Open any customer and you see the signals the decision was made from, the guardrails it passed, which side of the holdout they fell on, and why the offer went out when it did. When someone asks at the counter why they got a free drink, your team can answer. The customer gets control too: a preference page on their pass where they can switch personalized offers off and keep their stamps, see exactly what the card uses (visits with this card, stamps and rewards, whether an offer worked) and what it never uses (their location between visits, anything from other apps or shops), and download or delete their data.
Who It Is For
Anyone whose business lives on repeat visits: cafés and restaurants with morning regulars and a second-visit gap, salons and barbershops where a six-week cadence is easy to drift from, and gyms and studios where members stop coming long before they cancel. If you already run a digital stamp card or points card on Wally, the agent works from the visits you already have.
The Takeaway
A loyalty card tells you who your customers are. The agent decides what to do about each of them, does it, and shows you whether it worked. You set the guardrails. It does the work. See the full feature page at wally.ae/solutions/ai-personalization, or book a free demo and we will show you what the agent would propose for your own card.