Most loyalty programs start the same way. Someone notices a competitor has one, a plugin gets installed, customers earn a point per pound or dollar, and a year later the team reports that membership has grown impressively. What nobody quite measures is whether any of those members now buy more often than they would have anyway. In a surprising number of businesses, the honest answer is that the program mostly hands discounts to people who were already coming back.
That is the core risk in loyalty design, and it reframes the whole choice between models. A loyalty program is only worth its cost if it changes behaviour, not merely rewards it. The wrong model does not simply underperform; it solves the wrong problem at full price. So the question is not "points, tiers or paid membership?" in the abstract. It is which model your customers' actual buying pattern can sustain, and how you will prove it is doing more than subsidising loyalty you already had.
The variable that decides it: purchase frequency
Before comparing models, answer one question honestly: how often does a typical customer buy from you? Purchase frequency is the most constraining variable in loyalty design, because most mechanics depend on momentum. Points, progress bars and streaks only motivate if customers transact often enough to see themselves getting closer to something before they forget they enrolled.
That gives a rough map. Weekly or more frequent buying can sustain almost any model, which is why coffee, groceries and convenience categories run successful points programs. Monthly buying suits points with a quickly reachable first reward, or tiers with long qualification windows. Quarterly or rarer buying rarely supports traditional points at all, because customers never accumulate enough to care; for furniture, mattresses or appliances, referral incentives, service perks and an excellent post-purchase experience usually beat any currency. Your own data answers this better than any benchmark, which is exactly what a cohort analysis shows you.
| Model | Works when | Watch out for |
|---|---|---|
| Points | Frequent purchases; simple, habitual categories | Rewards nobody reaches; pure discounting |
| Tiers | Spend varies widely; status carries weight | Top tier out of reach; harsh demotion |
| Paid membership | A frequently used benefit clearly beats the fee | Selection effect disguised as success |
| Hybrid (points + tiers) | Enough volume to run both layers well | Complexity customers stop understanding |
Points: simple, and easy to get wrong
A points program turns spending into a currency, and its appeal is that everyone understands it. It works best in categories with frequent, habitual purchases, and the best-known examples are enormous: Starbucks has reported that its rewards members drive more than 40% of its US company-operated revenue. The mechanic is powerful when purchases are frequent enough for progress to feel real.
The common failure is a first reward set so far away that customers never reach it. If someone needs eight purchases to earn anything and typically buys twice a year, the program is invisible to them. Set the first redemption within roughly two or three purchases, make the balance visible in every email and receipt, and consider varying the earn rate to steer behaviour you value, such as more points for app orders or for trying a new category, rather than paying the same rate on everything. That kind of targeted earning pairs naturally with good email segmentation.
Atria
AI ad platform that researches, creates, launches, and analyzes winning Meta and TikTok creative.
Best for: Performance marketers scaling Meta and TikTok ads
Tiers: selling status, not just savings
Tiered programs add a status ladder, and the psychology is genuinely different: the goal is not only to earn something but to become something. That is why tiers suit categories where customer spending varies widely and recognition matters, such as beauty, fashion, travel and premium retail. Industry research, including Antavo's loyalty studies, has reported noticeably higher returns for tiered programs than flat ones, though figures like these come largely from loyalty vendors and deserve some caution.
Design details make or break tiers. Keep the number of levels small, usually three, so progress is legible. Make the next tier feel reachable rather than aspirational in the abstract, and add small rewards between thresholds so customers do not stall halfway. Think carefully about demotion: dropping someone a tier is a genuine loss that can sour the relationship, so generous qualification windows and a grace period are worth their cost. The status logic is the same one behind tiered affiliate programs, applied to customers rather than partners. Higher tiers are also where non-monetary benefits shine: early access, better service and member-only products cost little and feel special.
Paid membership: powerful, and easy to misread
A paid membership charges an upfront fee for ongoing benefits, most commonly free or faster delivery, member pricing or exclusive access. When it fits, it is the most powerful model of the three, because the fee creates a psychological commitment: members want to get their money's worth, so they concentrate spending with you. It suits businesses where committed customers buy often enough that a clear, frequently used benefit visibly pays back the fee.
It also produces the most misleading success story in loyalty marketing. Paid members almost always spend far more than non-members, and that comparison is routinely presented as proof the program works. But people who choose to pay were already your most engaged customers, so much of that difference existed before they joined. This is a selection effect, not program lift. The honest test is whether members spend more than they did before joining, or more than similar customers who did not join, which is the same discipline we covered for subscription-style income in recurring revenue economics.
The test A loyalty program is only worth its cost if it changes behaviour, not merely rewards it. If membership grows while purchase frequency stays flat, you have built an expensive discount for customers you already had.
Do the margin maths before you launch
Every reward is paid out of gross profit, not revenue, which is why a reward rate that sounds modest can be expensive. It is worth doing the break-even calculation before choosing an earn rate.
A member spends $1,000 a year at a 40% gross margin → $400 gross profit.
You return 5% of spend as rewards → up to $50 cost, which is one eighth of that profit.
→ If every point is redeemed: the program must generate about $143 of extra annual spend to pay for itself (each extra dollar earns 40¢ margin but also 5¢ in rewards) → roughly 14% more spending.
→ If 30% of points go unredeemed (breakage): cost falls to about $35 → roughly $100 of extra spend needed → about 10% more.
The question to answer honestly: will members genuinely buy 10 to 15% more because of the program? If not, a lower earn rate plus non-monetary perks is the safer design. And do not build a program that relies on breakage — counting on customers forgetting their rewards undermines the loyalty you are paying for.
Measure lift, not membership
Enrolment is the vanity metric of loyalty marketing. A growing member count says nothing about whether the program is changing anything, and comparing members against non-members mostly measures who chose to join. What you want to know is the incremental effect.
The cleanest method is a holdout: keep a small group of eligible customers out of the program for a period and compare their purchase frequency and retention against members. Where that is impractical, compare cohorts acquired just before and just after launch, or compare members' own behaviour before and after they joined. Track incremental purchase frequency, retention and margin per customer, not points issued. This needs your loyalty data to live alongside your other customer data, which is one of the practical arguments for consolidating around one customer record. The same measurement discipline applies across the business, as we argued in attribution and any honest funnel audit.
Start simple, then add layers
The strongest programs often end up as hybrids, commonly points as the everyday mechanic with tiers on top for aspiration, and sometimes a paid layer for the most committed customers. But few start there. Launching all three at once produces a program customers cannot explain to a friend, which is usually the first sign they will not engage with it.
A better path is to launch the one model your purchase frequency supports, prove it moves behaviour with a holdout or cohort comparison, then add a layer only when you can show the first one works. Communicate the program where customers already are, through SMS flows, email and your newsletter, and treat the member data it generates as a genuine asset: consented, first-party purchase history is exactly what a solid first-party data strategy is built on. Members who have lapsed are a separate job, better handled by re-engagement automation than by piling more points on accounts nobody is opening.
The short version
A loyalty program is only worth its cost if it changes customer behaviour rather than rewarding purchases that would have happened anyway. Choose the model by purchase frequency first: weekly buying sustains almost anything, monthly buying suits points with a reachable first reward or tiers with long qualification windows, and rare purchases rarely support points at all. Points work for frequent, habitual categories if the first reward sits within two or three purchases. Tiers sell status and suit categories where spending varies and recognition matters, provided the next level feels reachable and demotion is handled gently. Paid memberships are the most powerful model when a frequently used benefit clearly beats the fee, but member spending is inflated by a selection effect, so test against members' own past behaviour or similar non-members. Do the margin maths first, since a 5% reward on a 40% margin consumes an eighth of gross profit and needs roughly 10 to 15% more spending to break even. Measure lift with holdouts or cohorts rather than counting members, and start with one model before adding layers.
Want a loyalty program that changes behaviour, not just margins?
We help brands design and measure retention programs that pay for themselves.
Explore Email Marketing Automation →Frequently asked questions
What is the difference between points, tiered and paid loyalty programs?
They reward different behaviour through different psychology. A points program converts spending into a currency customers can redeem for discounts or products, and its logic is simple accumulation: buy more, get more. A tiered program adds status, moving customers up levels as they spend and unlocking better benefits at each step, so the motivation is becoming something rather than just earning something. A paid membership charges an upfront fee in exchange for ongoing benefits such as free shipping, member pricing or exclusive access, and it works by making the customer feel they need to use the membership to get their money's worth. In practice the strongest programs often combine models, most commonly using points as the everyday earning mechanic with tiers layered on top for aspiration. The right choice depends less on which model sounds best and more on how often your customers naturally buy, how much margin you have to fund rewards, and which behaviour you actually need to change.
Which loyalty program model works best for my business?
Start with purchase frequency, because it is the most constraining variable in loyalty design. Points and gamified mechanics need enough transactions to build momentum before customers forget they are enrolled, so they suit businesses where people buy weekly or monthly, such as coffee, groceries, beauty staples or pet food. Tiered programs suit categories where spending varies widely between customers and status carries weight, such as fashion, travel and premium retail, and they tolerate lower frequency if qualification windows are long enough. Paid memberships suit businesses where committed customers buy often enough that a clear, frequently used benefit, most commonly free or faster delivery, visibly pays back the fee. When purchases are rare, such as furniture, mattresses or appliances, traditional points rarely work because customers never accumulate enough to care; referral rewards, service perks and post-purchase experience usually do more. Margin matters too, since every reward is funded from gross profit, so low-margin businesses need a program that drives genuinely incremental purchases rather than simply discounting existing ones.
Do loyalty programs actually increase sales?
They can, but many measure the wrong thing and overstate the effect. Consumer surveys consistently show programs influence spending: in Deloitte's 2026 survey of more than 5,500 US loyalty members, 72% said a program makes them more likely to spend with a preferred brand. The trap is comparing members with non-members and calling the difference the program's impact. Members are not a random sample; the people who join, and especially the people who pay for a membership, were already more engaged, so much of their higher spending would have happened anyway. That is a selection effect, not program lift. To measure honestly, compare similar customers who did and did not experience the program, ideally with a holdout group that is eligible but not enrolled, or compare cohorts of customers before and after launch. Track incremental purchase frequency, retention and margin rather than enrolment numbers, because a program whose membership grows while transactions stay flat is simply giving discounts to customers you already had.
How much should a loyalty program give back?
Set the reward rate from your margin and the lift you can realistically achieve, not from what competitors advertise. Every reward is paid out of gross profit, so a reward that looks modest as a percentage of revenue can be a large share of profit. For example, returning 5% of spend on products with a 40% gross margin consumes one eighth of your gross profit on every member purchase. To break even, the program has to generate enough additional spending to replace that cost, which in that example means members need to spend roughly 10 to 15% more than they otherwise would, depending on how many points go unredeemed. Unredeemed points, known as breakage, reduce the real cost, but designing a program that relies on customers forgetting their rewards undermines the loyalty you are trying to build. Most programs work better with a modest earn rate, a first reward that is reachable within two or three purchases, and non-monetary benefits such as early access, better service or member-only products that cost little and feel valuable.
