wally.
Get a free demoSign in
Home/Solutions/AI personalization

Last updated

AI personalization · Journeys

AI personalization for loyalty programs. Every customer on their own journey.

Wally is the first wallet loyalty platform with a built-in personalization agent. It reads the visits on your stamp or points card, plans the journeys, sends each offer at that customer’s usual hour in Apple and Google Wallet, and reports the lift every Monday. You set the guardrails. It does the work.

First in wallet loyalty
Apple & Google Wallet
Lift measured against a holdout
Personalization agentBean & Leaf · stamp card · 812 active cardsSample

I looked at Bean & Leaf over the last year. Only 47% of first-timers come back for a second visit, which is where you lose most people. 128 customers are drifting right now.What should I work on first?

Get first-timers to a second visitWin back the drifting onesEverything · you decide
Get first-timers to a second visit · a free drink

Here’s the plan, each with the reason it exists:

2nd visitSecond-visit treat1 free drink · sent at their usual hour · 10% holdout~90a month
OverdueDrifting away3 days past their own usual gap61qualify today
Near rewardOne stamp awaymessage only · right after the visit94qualify today
Activate the plan

Live. First offers go out Tuesday at each customer’s usual hour. 20 customers are in the holdout and will never know. I’ll report on Monday.

Bean & LeafnowWelcome back, Maya! Your second visit earns a free drink, on us, until Saturday.
Offer sent · 07:45
Maya’s usual hour, not yours
Lift · second visit
+14 pts
41% with the offer · 27% held out
+14 pts
second-visit lift vs customers held out
1 in 10
customers kept as the control, so lift is real
2
questions to answer before the first journey goes live
0
segments, workflows or dashboards for you to build

Sample programme figures — illustrative, not a customer result

The problem we solve

A Friday blast is not a loyalty strategy.

Most digital loyalty programs stop at the stamp. The card fills, the reward redeems, and every customer gets the same message on the same day, whether they were in yesterday or have not been back since June. Personalization at this level has only been possible for brands with a data team. The agent does that job for a single café.

The challenge
Everyone gets the same offer, so regulars get rewards they did not need and strangers get nothing that fits.
With Wally
Each customer is on their own journey. A first-timer gets a second-visit treat; a regular who is on track gets nothing at all; a drifting customer gets the win-back, at their own usual hour.
The challenge
You find out a customer has gone a month after they left, if you find out at all.
With Wally
The agent knows each customer’s own cadence. A weekly regular who misses two weeks is caught in days, not weeks, because “drifting” means past their usual gap, not a fixed 30 days.
The challenge
You cannot tell whether an offer worked or whether those customers would have come back anyway.
With Wally
One customer in ten who qualifies is held out and gets nothing. Lift is the return rate with the offer minus the return rate without it. If a journey does not beat its holdout, the agent pauses it and tells you.
The challenge
Marketing automation means building segments and workflows you will not have time to maintain.
With Wally
There is nothing to build. The agent proposes the journeys with the numbers behind each one; you say yes, change something, or edit a journey by hand. The plan is live the next morning.
How it works

Reads the card. Plans the journeys. Runs them. Reports back.

The agent works from the visits already recorded on your stamp card or points card. It needs no integration with your till and no customer data beyond what the pass itself records.

1Read

Signals from the card

Visits in the last year, usual gap between visits, usual day and hour, stamps to go, whether earlier offers brought them back, and a churn risk score per customer. Refreshed every night.

2Plan

Journeys with a reason

You pick a goal and how generous it may be. The agent proposes the journeys, each with the number behind it: how many customers qualify today, the expected lift, the cost in rewards.

3Run

Inside your guardrails

One offer per customer a week, at most four a month, sent between 07:00 and 21:00, nothing to customers who are on track, one in ten held out. Delivered as a pass update on the lock screen.

4Report

One message every Monday

Who came back, with and without the offer. What the agent changed on its own, each one undoable. The two things it would like your yes on. Nothing else to check.

Journeys the agent runs

The moments that decide whether a customer stays.

Every journey is a trigger, an offer and a reason, run per customer. These are the ones the agent proposes most often for a café or salon card; you can edit any of them by hand or add your own.

2nd visit

Second-visit treat

TriggerThe day after a first visitOffer1 free reward, expires in 3 daysSentAt their usual hour, else 07:00
Why: fewer than half of first-timers return. The fourth visit is where habits form, so the second is the one worth buying.
Overdue

Drifting away

Trigger3 days past their own usual gapOffer2 stamps, or a free drink for lapsed customersSentAt their usual hour
Why: their cadence, not a calendar. A weekly regular is caught after ten days; a monthly one is left alone for five weeks. See the win-back playbook.
Near reward

One stamp away

TriggerRight after the ninth stampOfferMessage onlySentImmediately
Why: a nudge costs nothing, and completing the card lifts the fifth-visit rate, after which most customers stay.
At risk

Morning regulars

TriggerChurn risk above 60 for a weekday-morning customerOffer20% off until 11:00SentAt their usual hour, 20% holdout
Why: morning regulars decide where to go before they leave home. A mid-morning message arrives after they have already chosen somewhere else.
Habit

Fourth visit

TriggerFourth visit recordedOffer1 free drink, expires in 7 daysSentSame day
Why: the agent proposes this when your 4→5 transition sits below the benchmark for cards like yours. The reward buys the fifth visit; after that, most stay.
Points

Win-back 300

TriggerMore than 21 days since the last visit on a points cardOffer300 pointsSentAt 12:00, 10% holdout
Why: on a points card the balance is the hook. A visible top-up on the pass is a reason to come and spend it.
Under the hood

An agent you can trust with your customers.

Monday report

One message a week, in plain language.

No dashboard to decode. The agent tells you who came back, what it changed on its own, and the two things it would like your yes on. Each decision shows the numbers behind it, and anything it did alone can be undone with one tap.

  • Offers sent, return rate with and without the offer, lift in points
  • Changes the agent made alone, each with its result so far
  • Proposals that wait for you: new journeys, bigger rewards, different holdouts
Monday report · week 3Three journeys live. Here is what happened.
What I did on my own
Moved “Drifting away” from 10:00 to each customer’s usual hour
Return rate went from 24% to 31% on the 62 sent since.
Undo
Paused “One stamp away” at weekends
0 of 41 weekend nudges were followed by a visit; weekdays run at 55%.
Undo
Needs your yes
Add a “Fourth visit” journey · 1 free drink
Your 4→5 rate is 73% against 81% for cards like yours. Expected: about 5 more regulars a month, up to 48 drinks.
YesNot now
Guardrails and autonomy

You decide what it may do alone.

The guardrails are always on: one offer per customer a week, at most four a month, sends only in your window, nothing to customers who are on track, opted-out customers never included. On top of that, six permissions you switch on or off. Anything not switched on waits for your yes.

  • A monthly cap on rewards given, so generosity has a budget
  • Rewrites keep the same facts and never make a new promise
  • Take over any journey by hand at any time
What the agent may do alone
Everything else waits for you in the Monday report.
Move send times
Inside your 07:00–21:00 window
Pause what isn’t working
After 30 sends with no lift, or 0 returns
Rewrite messages
In the card’s voice, same facts, never a new promise
Change rewards
What a customer is given, or its expiry
Add journeys
New triggers on this card
Change holdouts
How many customers are kept as the control
Explainable

Every offer keeps its reason.

Open any customer and you see the signals the decision was made from, the guardrails it passed, which arm of the holdout they fell in, and why it went out when it did. When a customer asks at the counter why they got an offer, your team can answer.

  • Signals, guardrails and timing stored with each decision
  • Competing journeys that did not fire, and why
  • Per-customer churn risk on a scale your staff can read
Win-back offer sent · Drifting awayMaya H. · 7 of 10 stamps · today 07:45
“Maya, it’s been a while. Two stamps are waiting on your card — see you this week?”
Trigger · days past usual gap > 3. Maya crossed it today (4).
Signals · 7 visits in a year · usual gap 9 days · last visit 13 days ago · usually Tuesday 07:00–08:00 · churn risk 62.
Guardrails · slot free (last offer 19 days ago) · 2 of 4 this month · personalization on · not “on track”.
Arm · treated (holdout 10%). Send · her usual hour, inside the window.
Competing · One stamp away did not fire (3 to go).
Customer control

The customer shapes their own journey.

Two taps on their pass page, no typing: when they usually drop by, and what would make a treat feel like theirs. The agent stops guessing. They can switch personalized offers off and keep their stamps, see exactly what the card uses, and download or delete their data.

  • Only visits with this card, stamps and rewards, and whether an offer worked
  • Never location between visits, never data from other apps or shops
  • Built for consent-first markets: opt-in at sign-up, opt-out any time
9:41●●●
Your offers, your call

Hi Maya. This card can send you a treat at the right moment, based only on your visits here.

Personalised offers
On · around your usual visit time
    What we use
  • When you visit with this card
  • Stamps and rewards on this card
  • Whether an offer brought you back
  • What we never use
  • Your location between visits
  • Anything from other apps or shops
Measurement

Lift you can defend, not open rates.

Personalization only matters if the customers who got the offer came back more than the customers who did not. Every journey is measured that way from the first send.

With the offer412 customers · came back within 14 days41%
Held out46 customers · same 14 days, no offer27%
+14 pts lift on the second-visit treat, Bean & Leaf, 30 days
Sample programme · illustrative
A holdout on every journey
One in ten by default, adjustable. Message-only nudges can run without one.
Pauses what is not working
After 30 sends with no lift, or zero returns, the journey stops and you hear about it.
Industry reference points
Your second-visit and habit rates alongside benchmarks for cards like yours.
Reward spend with a cap
Free drinks and stamps given this month, against the monthly cap you set.
Compared

Blast campaigns, automation suites, or an agent.

Three ways to get a customer back. Only one of them does the thinking and the measuring for you, and lands on the pass the customer already carries.

Blast campaignsMarketing automation suitesWally AI personalization
Who gets the offerEveryone on the listSegments you define and maintainEach customer, based on their own visits; on-track customers get nothing
When it is sentThe day you send itA schedule you buildAt that customer’s usual hour, inside your window
Who writes the strategyYouYou, on a workflow canvasThe agent proposes it with the numbers; you approve
How you know it workedOpen and click ratesConversion goals you set upLift against a holdout on every journey, reported weekly
Where it landsInbox or SMSInbox, SMS, app pushLock screen via the Apple or Google Wallet pass; no app, no inbox
Data neededA listA CRM and integrationsOnly what the pass records; no till integration required
Why first

The pass is the data

Every stamp is a timestamped visit on the customer’s own card. Wally has the visit history and the delivery channel in one place, which is what makes per-customer journeys possible without a data team.

Why now

The channel is already open

A pass update reaches the lock screen of every customer who added the card. The agent decides who, when and what; wallet push carries it.

Why it holds

Measured from day one

Because the holdout is built into every journey, the agent learns from its own results and reports them. Retention numbers stop being a guess: see the retention statistics that make the case.

FAQ

AI personalization, answered.

AI personalization means each customer gets offers chosen, timed and worded for them, rather than the same campaign going to everyone. In Wally the personalization agent reads the visits recorded on a customer's wallet pass, works out their usual gap between visits and their usual hour, and runs a journey for them: a second-visit treat the day after a first visit, a win-back when they drift past their own cadence, a nudge when they are one stamp from a reward. The offer lands on the lock screen through Apple Wallet or Google Wallet, with no app to install.

No. The agent does the analysis and the planning. You answer two questions in a short conversation: what to work on first, and how generous it may be per customer. It proposes the journeys with the reason for each, you approve, and they go live. There are no segments to define, no workflow canvases to build, and no reports to decode. Every Monday you get one message: who came back, with and without the offer, and what the agent changed.

Only what the card itself records: when the customer visited with that card, how often, stamps and rewards on the card, and whether an earlier offer brought them back. It never uses the customer's location between visits, and nothing from other apps or shops. Customers see exactly this list on their preference page, can switch personalized offers off while keeping their stamps, and can download or delete their data.

Only if you let it. 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 a reward's expiry, adding a journey or changing the holdout always waits for your yes in the Monday report. Each permission is a toggle you control, and any change the agent makes alone can be undone.

Every journey keeps a holdout: by default one customer in ten who qualifies is kept as a control and gets nothing. Lift is the difference between the return rate of the customers who received the offer and the return rate of the holdout, measured over the same 14 days. If a journey has not beaten its holdout after enough sends, the agent pauses it and tells you. There is no vanity reporting, because the customers who would have come back anyway are counted on both sides.

A journey is a trigger, an offer and a reason, run per customer. The trigger is an event on the card: a first visit, a lapse past the customer's usual gap, a ninth stamp, a birthday. The offer is a reward, stamps, or a message only. The agent sends it at that customer's usual hour inside the window you set, and it is delivered as a pass update that appears on the lock screen. Each journey also has its own guardrails: one offer per customer a week, nothing to customers who are coming anyway, and the holdout.

Yes. The journeys run on the same Wally pass whichever wallet the customer added it to. Apple Wallet shows the update on the lock screen and on the pass; Google Wallet raises a notification and updates the pass. The customer never installs anything, and your staff keep stamping with Wally Scan or an NFC reader as before.

Marketing automation suites give you a canvas and leave the strategy, the segments, the copy and the measurement to you, and they usually reach customers by email or SMS. Wally's agent brings the strategy: it reads your card, proposes the plan with the numbers behind it, runs it inside your guardrails, measures lift against a holdout and reports in plain language. Delivery is the wallet pass the customer already carries, so there is no inbox to compete in and nothing to unsubscribe from except the journey itself.

From the blog

Latest guides

Travel Agency Loyalty Program: Be on the Client's Phone When the Next Trip Is Decided
GuidesOct 11, 20265 min read

Travel Agency Loyalty Program: Be on the Client's Phone When the Next Trip Is Decided

A travel agency's best client books AED 20,000–50,000 a year — summer flights home, Umrah trips, visas, winter holidays — and still disappears to a booking portal between trips, because the portal emails in January and the agency doesn't. A travel agency loyalty program in Apple Wallet and Google Wallet puts points, a visible Gold tier and free lock-screen fare alerts on the client's phone — joined in one QR scan with the e-ticket, no app.

Read article
Dental Clinic Loyalty Program: Keep the Patient You Only See Twice a Year
GuidesOct 10, 20265 min read

Dental Clinic Loyalty Program: Keep the Patient You Only See Twice a Year

A dental patient is decades of six-month recalls, whitening and family bookings — yet clinics lose them one silent gap at a time: the recall call from an unknown number, the stalled treatment plan, the January insurance switch. A dental clinic loyalty program in Apple Wallet and Google Wallet puts the next check-up date on the face of the pass, sends free lock-screen recalls, turns packages into visible session cards — and fills Q4 with one push about insurance benefits expiring December 31.

Read article

See the agent plan your card’s first three journeys.

A 20-minute call: we look at your visit pattern, show what the agent would propose for your stamp or points card, and what it costs across your locations.