Points, Tiers, or Cashback: Choosing the Right Loyalty Program Model

The short version: Choose a loyalty program model from your own order export, not from a competitor: points need customers who come back often enough to reach the first reward, tiers need revenue concentrated in a minority of customers, and store-credit cashback buys comprehension but nothing to build toward. Contribution per order caps the earn rate whichever model you pick.

A loyalty program can fail before launch, at the moment someone says "let's just do points." The model you choose is not a branding decision — it is an economic contract with your customers, and each of the three models makes a different promise.

Key takeaways

• Compute orders_to_reward = reward_value / (AOV × earn_rate), then check what share of your customers ever place that many orders, and one more to spend the reward.
• Tiers become a candidate when your top 10% of customers bring several times their 10% headcount share of 12-month net revenue.
• Very regular reorder gaps point to a subscription before points.
• A low 12-month repeat rate means fixing the second order before launching any program.
• Contribution per order caps the earn rate, whatever model you pick.

Points: flexible, but easy to get wrong

Points work when customers come back often enough to save up toward a reward, and whether yours do is a number in your order export, not a property of your category. The two levers that matter are earn rate — the effective discount, which you price from your own margin — and burn friction (how easy redemption feels). When the same basket arrives on a steady schedule, compare points with a subscription first; reorder variance is the test that separates the two.

Reward reachability is the order count at which a new member earns a first reward: orders_to_reward = reward_value / (AOV × earn_rate), rounded up. The reward can first be spent on the order after that one. Then read your own order-count distribution: what share of customers ever place that many orders, and how many place one more? The classic failure: a cheap earn rate puts the first reward on order 10, almost nobody places 10 orders, and the program is dead on arrival — the liability sits on your books while the motivation never materializes. There is no standard share to aim for, so pick one, write it down and recheck it after a quarter.

Tiers: status for high-variance spend

Tiers fit when a minority of customers bring a large share of revenue, and you can check that first: compute the share of 12-month net revenue your top 10% of customers bring. With spend spread evenly it would be 10%; the further above that yours sits, the more a tier has to work with. You are not paying for transactions; you are paying for identity, which only works if members care about keeping a status they could lose — test whether yours do before funding perks.

Set the thresholds from your own spend percentiles, not a fixed share of customers: see loyalty tier thresholds set with spend percentiles. Then judge a tier by the extra orders it produces, not by the spending gap that already existed — the question of when to add loyalty tiers.

Cashback: simple, honest, expensive

Cashback, for a small online shop, is best run as store credit: a percentage of each order returned as credit once the return window closes, spendable next time. At the same earn rate, store credit and points issue the same face value. Credit may redeem more because there is no reward threshold; budget each model at your own observed redemption rate, or at 100%, the conservative face-value case, until you have measured it. It buys comprehension and a balance waiting for the next order, but nothing builds toward a bigger prize, so a rival with a higher rate competes on the only thing you offer. For the prepaid version, see what Starbucks Rewards gets right.

How to choose a loyalty program model

A loyalty program model can be chosen from four numbers built from an order export (customer email, order date, order total) and your cost figures. Use enough prior history to identify true first orders and record an observation date; compute the 12-month repeat rate only for acquisition cohorts with a completed 12-month follow-up. Use a stated trailing 12-month window for revenue concentration. Repeat rate: the share of first-time customers who order again within 12 months. Concentration: the share of net revenue from your top 10% of customers. Reorder regularity: how much the gap between orders varies for customers with three or more orders; also record median first-to-second-order days among repeaters in those mature cohorts. Contribution per order: order value minus product cost, shipping, payment fees and returns.

Each number triggers a rule. A low repeat rate means fixing the second order comes before any program; see loyalty programs for infrequent purchases. High concentration makes tiers a candidate. Very regular reorder gaps point to a subscription before points. Contribution per order caps the earn rate whatever you pick. None has a universal cut-off: "low" and "high" are calls you make on your own export, so record the value you used.

The model is the skeleton. The next question — how generous to be — is where the real margin math starts. Begin with loyalty program examples that name the funding source behind each perk, so the model you pick arrives with a budget attached.

The same business under all three

An illustrative online shop (replace every input with your own): $55 average order, 30% contribution margin, so $16.50 per order, and a median of 60 days between orders. Suppose its export shows 35% of customers placing a second order and 15% a third.

At conservative full redemption, points at 2% of spend cost $1.10 an order, 6.7% of contribution. A $10 reward needs 10 / (55 × 0.02) = 9.1, so it is earned on order 10 — nine 60-day gaps, about 18 months — and first spent on order 11, in a shop where only 15% reach a third order. The cheap rate has not made the program cheap; it has made it pointless.

At 5% and full redemption, each order costs $2.75, 16.7% of contribution, and a $5 reward needs 5 / (55 × 0.05) = 1.8, so it is earned on order 2, which 35% of customers reach, and first spent on order 3, which 15% reach. The trade is now explicit: 16.7% of contribution on every order, including from customers who would have returned anyway, for a reward about a third of customers earn and 15% get to spend. Both choices are decisions; drifting into a distant reward is not.

Tiers on this shop hinge on the number the inputs leave out: concentration. If the top 10% of customers bring barely more than 10% of revenue, a top tier separates customers by a handful of orders a year, and status that easy to reach and undramatic to lose gives members little to protect. If they bring several times their headcount share, a tier has room to work, and the remaining test is whether it adds orders.

Store credit at 5% issues the same $2.75 face value per order, a $2.75 cost at full redemption, and buys something different: any repeat order can redeem it, and nothing builds toward a bigger prize. What it does buy is comprehension, which matters when customers compare your price with a cheaper option before every order.

At the same face-value earn rate, realized cost still depends on redemption; the models aim at different behaviors. The model question is not which one is generous — it is which mechanic your purchase pattern can support.

Frequently asked questions

Can I combine two models?

Yes, and tiers layered on points is a common combination: points do the transactional work, tiers do the identity work. What to avoid is three currencies a customer has to hold in their head at once. If you cannot state the whole program in one sentence at checkout, it is too complicated to change behavior, whatever the feature list says.

How do I actually set the earn rate?

Start from what you can fund, not from a competitor rate. Take contribution margin per order, decide what share of it you are willing to spend on repeat behavior, and work back to a percentage of spend. A 2% earn rate in a 25%-margin business commits 8% of the profit pool on every transaction, including those from customers who were returning anyway. Then check the other end: compute orders_to_reward at that rate and look up what share of your customers ever place that many orders, plus one more to spend the reward. If few do, the rate is too low to motivate anyone, whatever the model says.

What if my margins cannot fund any of these?

Then the answer is not a thinner version of the same program. Non-discount benefits — priority access, service levels, useful information, faster support — still incur incremental operating costs that must fit within contribution margin. Price those costs and test whether customers value the benefits enough to change purchasing. A thin-margin business running 1% cashback has bought the right to be outbid.