Guide
7 min read

Loyalty program design: the decisions that decide whether it pays

Loyalty program design is the set of decisions made before anything is built. What earns. How fast. What it is worth. Where the tiers sit. What happens when a member complains, a point expires, or a reward costs more than planned.

These decisions look small on a whiteboard. Together they decide whether the program makes money or quietly loses it. Software can be rebuilt in weeks. A bad earn rate, promised to thousands of members, cannot be taken back without a fight.

What loyalty program design covers

Design covers five things, in this order: the customer the program is for, the earn rate, the rewards, the tiers, and the rules. Each one depends on the one before it. Pick rewards before you know the margin and you have promised money you have not counted. Place tiers before you know how customers actually spend and the top tier fills with the wrong people, or with nobody.

This guide walks through each decision the way a studio works through it with a brand: from real customer data, with the cost modelled before the promise is made.

Start with the customer you want more of

Every program rewards somebody. The design question is who. A program built for your most frequent customers looks different from one built to rescue customers who are drifting away, and both look different from one built to recruit new buyers.

Pull the last year of transactions and group customers by how often they buy and how much they spend. Most businesses find the same shape: a small top group that carries a large share of revenue, a wide middle that buys occasionally, and a long tail that bought once. The program's job is usually to move the middle up, while keeping the top group feeling recognised. Design for that movement. A program that mostly rewards the top group for what they already do is a discount with extra steps.

Set the earn rate from margin, not from rivals

The earn rate is the share of each sale you give back as reward value. A common working figure is around 5 percent of spend, and our types of loyalty apps guide uses the same rule of thumb. Treat it as a starting point for the maths, not as the answer.

The answer comes from your margins on the products members actually buy. If members mostly buy your highest-margin lines, a richer rate is affordable. If they mostly buy promoted, low-margin lines, the same rate can erase the profit on the sale. Model the rate at full redemption: assume every point earned gets spent. If the program only works when most points are never used, it does not work, and members will eventually say so.

Place tiers on real spending

Tiers fail in a predictable way: the thresholds are round numbers picked in a meeting, and a year later 2 percent of members hold the top tier and 90 percent sit in the bottom one, feeling nothing.

Place the thresholds on the spending data instead. Sort members by annual spend. Put the middle tier where your good-but-not-best customers already are, so the next tier up is visible and reachable. Put the top tier where only the genuine top few percent qualify. Three tiers are enough for almost every brand: entry, recognised, and top. The reward for climbing should be status and service as much as points: early access, a named contact, first refusal on limited stock. Those cost less than points and are harder for a rival to copy.

Choose rewards people can picture

A reward works when a member can picture it while they are still deciding where to buy. "500 points" pictures nothing. "A free treatment", "a case of the wine you reorder", or "the tool hire is on us" pictures something specific.

Two design rules follow. First, the first reward has to be reachable fast: around eight weeks for a regular customer is the working target, because a reward a year away motivates nobody. Second, the catalogue should mix one headline reward members talk about with smaller rewards they can reach often. The headline reward sells the program. The small rewards keep it alive between big purchases.

Price the liability before launch

Every point issued is a promise to give something later. Finance calls the total of those promises the points liability, and it lands on the books whether the program team planned for it or not.

Design keeps it under control with three levers, all set before launch. The earn rate sets how fast the liability grows. The redemption value sets what each point costs when it is used. Expiry rules set how long an unused promise lasts, and they have to be stated in the open, because members treat surprise expiry as theft. Model all three together, at the enrollment numbers you hope for, not just the ones you expect. A program that succeeds beyond plan should be a good problem, not an unbudgeted one. Our loyalty program cost guide breaks down how these build and running costs fit together.

The design mistakes that cost the most

Four mistakes account for most failed designs. The first is copying a famous program. A coffee chain's earn rate works at a coffee chain's margins and visit frequency. Pasted onto a furniture retailer or a pharmacy group, the same numbers either lose money or motivate nobody.

The second is rewarding the purchase and nothing else. Programs can also reward the behaviours around the purchase: booking ahead, buying the range rather than the single item, leaving a review, coming back inside a set window. Each rewarded behaviour should be one the business can name a value for. If nobody can say what a behaviour is worth, it does not belong in the earn table.

The third is hiding the good rewards. Members judge a program by the best reward they can picture and by how close it feels. A catalogue where everything desirable sits at a level nobody reaches teaches members that points are worthless, and they price every later promise accordingly.

The fourth is designing for the launch offer instead of the steady state. A generous launch rate trains members to expect that rate forever. When it tightens, as it must if it was never costed, the tightening reads as a cut. Launch at the rate the margin can hold for five years, and let launch excitement come from the rewards, not from a rate you plan to withdraw.

The one-sentence test

When the design is drafted, explain the whole program in one sentence to someone outside the business: who earns, for what, and what they get. If the sentence needs a second clause full of exceptions, the program is too complicated to run at a till, and too complicated for members to value.

Then test the same sentence on the people who will operate it. In one study cited in our types guide, only 6.1 percent of counter staff ever asked customers to join the program they were running. A design the team cannot say out loud does not get offered, and a program that is not offered does not enroll. Simplicity is not a style choice here. It is the enrollment mechanism.

Bottom line

Good loyalty program design is mostly arithmetic done early: earn rate from margin, tiers from real spending, rewards people can picture, and the liability priced before the first point exists. Get those four right and the build is straightforward. Get them wrong and no software fixes it afterwards, because the promises are already out in the world.

Talk to the studio

If you run a mid-size or large business and want a program designed around your margins and your customers before anything is built, the next step is a scoping call. Bring a year of sales data if you have it. The design conversation starts there.

Book a scoping call. Projects start at $9,500.

Chloe Reed

Written by

Chloe Reed

Part of the LoyaltyPass editorial team. All articles draw on primary sources: brand announcements, industry research, and academic literature. Statistics are attributed inline. About our editorial team

No demo wall, no jargon. Here's what brands ask us.