Why loyalty programs fail — the death spiral no one talks about

Most café, restaurant and bubble-tea loyalty programs don't fail because the rewards are weak — they fail because two structural gaps interlock into a closed loop. The first gap is **blind points**: customers earn points but have no self-serve way to check their balance without asking a human, so they see no reason to keep playing. The second gap is **the cashier bottleneck**: enrolment and point-attachment depend entirely on the cashier remembering to ask 'do you want to earn points?', which cannot hold at a busy till. This piece argues both gaps can be patched without replacing the POS — and that once a shop is inside the loop, no reward on its own can pull it out. Before the fix, recognise the spiral through five familiar symptoms.

Before explaining the loop, recognise it through five familiar symptoms.

## The warning bells

If your shop has three of the five symptoms below, you're almost certainly inside the spiral: (1) very few members actually redeem — the voucher pool sits untouched; (2) new staff still can't pitch the program after two weeks on the floor; (3) the member list barely grows even though the shop serves hundreds of guests a day; (4) members never ask 'how many points do I have?'; (5) the owner cannot answer 'is the program working, and by how much?'. These aren't separate problems. They're all surface expressions of the same loop.

## The spiral — picture a circle

Imagine a six-node circle. Each node is the consequence of the one before it, and the final node feeds the first. There isn't really a 'starting point' — you can fall in at any node, but once you're in, the loop keeps itself running. The six nodes, detailed in the tactics list below:

Customers can't see their points → Customers decline to join → Cashiers get rejected repeatedly and stop asking → Member count stagnates → There isn't enough data to run campaigns → The owner concludes 'loyalty doesn't work here' → (back to the top: the program stops getting invested in, so members see even less)

This is not a checklist. It's a circle. Patching one node isn't enough, because the other five keep the wheel spinning.

## Two names for two gaps

Of the six nodes, the first two are the root causes — everything else is downstream. They deserve names.

**Blind points.** A customer earns points but has no way to check their balance without asking a human. On paper stamps, losing the card loses the points. On a POS module, the balance only shows on the cashier's screen — there's no customer account to look up. The customer lives in the dark, and darkness kills the intrinsic motivation to play.

**The cashier bottleneck.** Enrolment and point-attachment depend entirely on the cashier remembering to ask 'do you want to earn points?'. During the peak — 8am coffee, noon office lunch, 6pm milk tea — cashiers are counting change, printing receipts, calling the kitchen. One extra sentence is the first thing that drops. After it's been declined 3–5 times a shift, not asking becomes the default.

## This is not the staff's fault

Worth stating clearly: the cashier bottleneck is not a cashier problem. Widely accepted retail-loyalty guidance is that enrolment must happen at the register in the moment, and that every cashier must be trained to explain the program in fifteen seconds and to answer when a customer asks how many points they have. That is the standard, and it isn't controversial.

But look at what the standard assumes: it assumes a human will remember, every shift, every customer, while a queue is forming. And it assumes customers will ask another human to learn their own balance. Both assumptions are fragile. Public documentation from major POS vendors — the same pattern is visible across Toast, Square, and regional systems like KiotViet and iPOS — describes the loyalty module as a register-side enrolment flow: the customer hands a phone number to the cashier, who types it in. That description is accurate; it is not wrong. It is also an exact description of the bottleneck.

The spiral isn't an execution failure. It's the natural consequence of a design that loads the two heaviest tasks — enrolment and balance-lookup — onto the most fragile points: the cashier's memory during a rush, and the customer's willingness to ask.

## How to break the spiral — vendor-neutral first

There are two patches, one for each gap, and both are needed. Patching only one is not enough.

**Patch blind points.** Give every customer a way to check their balance any time, without asking anyone. Three common methods in order of friction: (a) a QR code on the table or menu — the customer scans and sees their balance in two seconds, no app install; (b) a QR code printed on the receipt — the customer takes the bill home and checks later, which also lets shops that never captured a phone number at the till still issue points; (c) a mini-app or channel inside a platform the customer already has (Zalo in Vietnam, WhatsApp elsewhere, LINE in Thailand) — one tap and the balance is visible, no new app. None of these require replacing the POS.

**Patch the cashier bottleneck.** Give customers a way to enrol themselves without the cashier having to ask. The strongest method today is post-payment receipt scanning — the customer photographs the bill, AI OCR reads the amount, timestamp, and items, and points are credited without any keystroke from staff. This moves the load off the cashier (peak hours) and onto the customer (relaxed, after the meal). It also works with any POS, including ones with no loyalty API.

When both gaps are patched, the spiral reverses: customers see their points → they see a reason to redeem → they enrol on their own → member count grows steadily without the cashier asking → there's enough data to run real campaigns → the program starts proving lift → the owner invests further.

## Where PEKO fits (briefly)

PEKO is one of the platforms that patches both gaps at once: AI OCR reads receipts on the customer's own phone (no POS swap — KiotViet, iPOS, Sapo, MISA, Toast, Square all keep running), and members see their balance and redeem inside the messaging channel they already use daily. The principle isn't PEKO-exclusive: any solution that combines the two layers can break the spiral. The mechanics of the OCR step are covered in /answers/how-ai-receipt-scanning-works-loyalty.

## Self-audit — 10 yes/no items

Score each item. Six or more 'yes' answers means you're inside the spiral and need to patch soon.

(1) Is there no way for customers to see their balance without asking staff? (2) Can customers self-enrol, or do they have to wait for the cashier to ask? (3) In the last 30 days, have fewer than 5% of regulars asked 'how many points do I have?'? (4) After 14 days on the floor, have new staff still not been trained to pitch the program? (5) Did member count grow less than 10% month-over-month while customer traffic held? (6) Is more than 70% of the voucher pool still unredeemed from the start of the quarter? (7) Can the owner not state the exact percentage of bills linked to a member? (8) During peak (60+ customers/hour), do cashiers skip the loyalty ask so entire weeks pass with almost no new enrolments from the till? (9) Have you not run a single win-back campaign to at-risk customers in the last 6 months? (10) Has the program been running longer than 6 months without a single measured revenue lift attributable to it?

6+ yes = inside the spiral. 4–5 yes = at risk. 0–3 yes = the program is working.