Loyalty programs: when they pay for themselves and when they're a margin drain

The pitch vs. the P&L reality

The internal pitch for a loyalty program almost always sounds the same: customers say they want it, competitors already have one, the app costs a modest monthly fee, and points feel free to give away because they're not cash out the door today. Enrollment numbers climb, engagement looks healthy, and the program gets filed as a retention win.

The P&L reality is different, and it shows up on a delay. Points issued today are a liability that comes due later, when customers redeem — and by the time redemption volume is large enough to show up as a real cost, the program has been running long enough that pulling it back is a customer-experience problem, not just a budget line. This is the same delayed-recognition trap that shows up in how returns and discounts quietly compress margin: a cost that doesn't hit the P&L at the moment of the decision that created it is a cost most teams systematically underweight.

None of this means loyalty programs are bad. It means the decision to run one — and how to structure it — should be made with the same rigor as any other margin-affecting program, not waved through because the enrollment metric looks good and the sign-up flow was easy to build.

What a points program actually costs

The direct cost of a points program is the redemption liability: the value of points customers earn and eventually redeem, valued at what redemption actually costs you — your contribution margin or product cost on the redeemed item — not the full retail price the customer perceives they're getting. A program offering 5% back in points, fully redeemed, is a 5% margin cost on every dollar of qualifying revenue, functionally similar to running a permanent 5% off program, except it's accounted for as a marketing or loyalty expense rather than shown as a discount line, which is exactly why it tends to get less scrutiny than an equivalent-sized discount campaign would.

Three factors determine the real number, and all three need to come from your own program's data rather than an industry rule of thumb: the accrual rate (points issued per dollar spent), the redemption rate (what share of issued points actually get redeemed rather than expiring or going unused — commonly called breakage), and the redemption value (what the redeemed points actually cost you at your margin, not at face value). A program with a high accrual rate but low redemption rate can look expensive on paper and cost much less in practice; the reverse is also true, and guessing wrong in either direction either overstates a program's viability or kills one that was actually working.

The points weren't free when they were issued. They were a liability with a due date, and the due date is exactly when most teams first start asking whether the program is worth it.

Incremental behavior change vs. subsidizing what was already coming

The question that actually determines whether a loyalty program is working isn't "do enrolled customers have a higher repeat purchase rate than non-enrolled customers." That comparison is almost always misleading, because your most loyal, highest-frequency customers are also the ones most likely to notice and enroll in a loyalty program in the first place. A higher repeat rate among enrolled members mostly reflects who chose to enroll, not what the program caused them to do differently.

The real question is incrementality: of the purchases happening inside the program, how many wouldn't have happened — or would have happened later, or at a smaller basket size — without the points incentive. This is the identical logic from repeat purchase rate as a lever: a program that pays out to customers who were already going to buy again is a cost with no behavior change attached, while a program that pulls forward a purchase, prevents a lapse, or lifts basket size to hit a points threshold is creating real, attributable value.

In practice, most loyalty programs are a mix of both, and the mix is what determines whether the program is worth its cost. A program that's 80% subsidizing existing behavior and 20% creating incremental behavior is spending five dollars in redemption liability for every dollar of real lift — a number that would never survive scrutiny if it were labeled a customer acquisition cost instead of a "loyalty investment."

When loyalty programs earn their cost

Loyalty programs tend to pay for themselves under a specific, identifiable set of conditions:

• High natural repeat-purchase categories — consumables, beauty, pet, food — where the product itself creates a reason to come back, and the points program's job is to influence timing and basket size on visits that were largely going to happen anyway, which is a smaller lift to prove but a real one.
• Sufficient margin to absorb the redemption cost without the program eating into the contribution margin you need to fund acquisition and operations — a program layered onto an already-thin-margin business compounds the exact P&L pressure it's supposed to relieve.
• A points structure with real behavioral thresholds — tiers or redemption minimums that require a customer to actually change their behavior (hit a spend level, make a second purchase within a window) to unlock value, rather than paying out proportionally on every dollar regardless of what the customer would have done anyway.

When they're a quiet margin drain

The mirror conditions predict a program that costs more than it returns:

• Low-repeat, considered-purchase categories — furniture, mattresses, high-ticket electronics — where the natural repurchase cycle is so long that a points program is unlikely to be the deciding factor in bringing a customer back, and mostly just discounts the eventual purchase that was coming regardless.
• Thin-margin businesses where a proportional points payout consumes a share of contribution margin the business can't structurally afford, regardless of how much genuine loyalty it builds.
• A price-sensitive, discount-seeking customer base acquired largely through promotions and paid discounting — this segment tends to treat points as one more discount mechanism to optimize around rather than a reason for brand preference, which caps the program's ability to create the loyalty it's named for.
• Programs run without ever measuring incrementality — not a category or margin problem, but an organizational one: if nobody is checking whether enrolled-customer behavior actually changed versus a comparable non-enrolled group, the program persists on vibes and enrollment counts indefinitely, whether or not it's earning its cost.

Alternatives that don't scale cost with revenue

A structural weakness of percentage-back points programs is that the cost scales directly with revenue — the more successful the business is, the more the program costs, in lockstep, with no ceiling. A few alternative structures decouple loyalty from that direct revenue tax:

• Tiered VIP status — early access to new products or sales, unlocked by spend or purchase frequency, with a fixed operational cost rather than one that scales per dollar of revenue.
• Non-monetary perks — free shipping thresholds, exclusive SKUs, a dedicated support line — that create real switching cost and status without a redeemable-dollar liability sitting on the books.
• Milestone rewards instead of continuous accrual — a meaningful reward at a second or third purchase, when the retention curve is steepest and the marginal dollar of incentive does the most work, rather than a small reward on every single transaction indefinitely.

None of these are automatically better — they trade the points program's simplicity and broad appeal for a structure that's more capital-efficient but sometimes less immediately compelling to a shopper used to seeing "earn points" at checkout. The right choice depends on the same category and margin conditions above.

How to test before you commit

The reliable way to know whether a loyalty structure will earn its cost in your specific business is to pilot it against a control group before a full rollout — offer it to one segment or cohort, hold a comparable segment out, and measure the actual difference in repeat rate, time-to-next-purchase, and basket size between the two groups over a defined window. This directly answers the incrementality question instead of inferring it from enrolled-vs-unenrolled comparisons that are contaminated by self-selection.

This also connects the loyalty decision to the same customer lifetime value discipline that should govern any retention investment: a loyalty program is, functionally, a spend against future LTV, and it deserves the same scrutiny as an acquisition channel — what does it cost, what behavior does it actually change, and does the math clear at your real contribution margin, not at the optimistic version of the pitch.

Deciding with the numbers, not the pitch

A loyalty program is neither a guaranteed win nor a guaranteed drain — it's a P&L decision that happens to be dressed up as a customer-experience one. The businesses that get value from one are the businesses that ran the incrementality and margin math before launch and keep checking it afterward, not the ones that launched because customers said they'd like it.