Playbooks

Loyalty programs that retain (not just discount)

Most loyalty programs just train guests to wait for discounts. How to design one that actually builds habit, data, and repeat visits.

By The Crubby TeamPublished on 15 May 20266 min read

Most restaurant loyalty programs are discount programs wearing a name tag. They teach your best guests to expect money off, and quietly erode the margin on the visits they'd have made anyway. A program worth running does something harder: it changes behavior.

The short version

  • The default loyalty program subsidizes visits guests would have made regardless, a pure cost, not an investment.
  • Habit, status, and surprise beat raw discount at driving incremental repeat visits.
  • The most durable payoff is often the first-party data, not the reward itself.
  • Simple, frequent, and frictionless usually outperforms clever and complicated.

The discount trap

Here is the uncomfortable math. If your loyalty program hands out, say, a free item after a set number of purchases, a large share of those rewards go to people who were already loyal. You're not buying a new visit, you're buying a markdown on an existing one. Industry discussion of loyalty economics consistently lands on the same warning: an undisciplined program can quietly shave several points off margin while looking, on the dashboard, like a roaring success.

It gets worse when discounts become the relationship. Train guests to wait for the deal and they'll do exactly that, deferring full-price visits, stacking offers, and treating your brand as a coupon source. The signal you want is incremental behavior: a visit, a larger basket, or a returning lapsed guest that would not have happened otherwise. Everything else is cost dressed up as engagement.

Pick one job

Before launching anything, write down the single behavior you want more of, visit frequency, off-peak traffic, larger checks, or lapsed-guest recovery. A program that optimizes for everything optimizes for nothing.

Three structures, three philosophies

Almost every program is a variation on one of three models. They are not equal, and they don't suit the same operator.

Points

The familiar earn-and-burn currency. Points are flexible and easy to understand, which is their strength and their weakness: flexible currency drifts toward being a discount by another name. Points work best when the earn rate is calibrated so rewards feel reachable but still reward genuine frequency, and when you resist the urge to bolt on a discount every time traffic dips.

Tiers

Status-based programs (spend or visit enough to unlock a level) lean on a different motivator: status and progress. People will work toward a threshold they can see. Tiers are powerful for higher-frequency or higher-ticket concepts, but they punish the casual guest and can feel exclusionary if the climb is too steep. The trick is making the first rung easy and the perks at each level feel like recognition, not just a bigger coupon.

The fastest-growing idea in the category: guests pay a monthly fee for ongoing perks, free delivery, a daily drink, members-only pricing. A paid program is self-selecting (only your frequent guests buy in), generates predictable revenue, and creates a powerful sunk-cost nudge to visit and 'use' the subscription. The catch is that the perks have to be genuinely valuable, or churn is brutal.

What the leaders actually did

The major chains widely cited as loyalty winners didn't win on the size of the discount. They won by wiring loyalty into the ordering experience, the app, the pre-order, the saved payment, so the reward and the habit are the same motion.

What actually drives repeat visits

Discount is the laziest of several levers, and usually the weakest per dollar spent. The ones that compound:

  • Habit. Frequency itself is the goal. Anything that makes the next visit the default, a saved order, a standing 'usual', a streak, beats a one-time markdown. Behavioral research broadly agrees that habit formation, not motivation, is what sustains repeat behavior.
  • Status. Recognition is cheap to give and disproportionately valued. A named tier, early access, or 'we remembered your order' often outperforms an equivalent dollar discount.
  • Surprise and delight. An unexpected reward (a free dessert, a birthday item) generates goodwill and word of mouth that a predictable, earned discount never will, precisely because the guest didn't transactionally 'buy' it.
  • Convenience. Removing friction, re-order in two taps, auto-applied perks, quietly raises frequency more than most promotions.

The quietly valuable part: data

Even a program that barely breaks even on rewards can pay for itself in first-party data. Knowing who your repeat guests are, what they order, how often, and when they lapse lets you do things a transaction record never could: win back a guest who's gone quiet, time an off-peak nudge, build a menu around what your regulars actually buy. In an era where third-party delivery platforms and ad networks increasingly sit between restaurants and their customers, a direct, owned relationship with your guests is strategically valuable on its own.

The reward gets the guest to enrol. The data is what you keep.

Design for frequency, and keep it simple

A few principles that separate programs that work from programs that just exist:

  1. 1.Match the cadence to your concept. A coffee shop can reward daily; a steakhouse cannot. Set the earn threshold against realistic visit frequency, not aspiration.
  2. 2.Make the first reward fast. Early momentum drives enrolment and the first repeat visit; a reward that takes ten purchases to reach loses most sign-ups before they get there.
  3. 3.Make redemption effortless. If staff have to look up a code or the guest has to remember a card, you've added friction exactly where you wanted to remove it.
  4. 4.Protect off-peak and incremental behavior. Bonus earning on slow days or for lapsed guests is far better spent than a blanket discount during your busiest service.
  5. 5.Measure incrementality, not enrolment. Sign-up counts flatter everyone. The real question is whether members visit more than they otherwise would.

Common failure modes

  • It's just a discount. No status, no data use, no habit mechanic, only money off. This is the most common and most expensive mistake.
  • Too complex. Multi-currency points, blackout rules, expiring tiers. Complexity kills participation; if a guest can't explain the program in one sentence, it's too clever.
  • Rewarding the wrong behavior. Paying for visits that would have happened anyway, or discounting your highest-margin, highest-demand items.
  • Set and forget. The data gets collected and never used. A program that doesn't drive a single win-back campaign is a database, not a strategy.
Points, tiers, or paid, which should I start with?
Match it to frequency and ticket. High-frequency, low-ticket concepts (coffee, fast casual) suit points or paid subscriptions. Higher-ticket or aspirational brands often do better with tiers and status. When in doubt, start simple, a clean points or visit-based program, and add structure only once you understand your data.
How do I know if my program is actually working?
Look past enrolment numbers to incrementality: do members visit or spend more than comparable non-members, and are lapsed guests coming back? If members' behavior looks identical to what it would have been without the program, you're funding a discount, not driving retention.
Aren't free rewards just a cost?
Only if they go to behavior you'd have gotten for free. A reward that pulls an incremental visit, a larger basket, or a lapsed guest back is an investment. A reward redeemed on a visit that was always happening is pure margin erosion, which is why targeting matters more than generosity.
Is the data really worth more than the rewards?
For many operators, yes over time. The reward cost is paid once per redemption; the guest data, frequency, preferences, lapse signals, keeps generating value across every future campaign and menu decision, and it's yours rather than a platform's.

The bottom line

A loyalty program is not a discount you give to people who already love you, it's a system for changing what guests do and learning who they are. Pick one behavior to move, choose the structure that fits your concept, lean on habit and status before price, and actually use the data you collect. Do that and the program earns its keep. Skip it, and you've simply found an expensive new way to mark down the visits you already had.

Keep reading