Understanding the Value and ROI of Customer Loyalty Programs

Loyalty program ROI is the incremental value a program creates, the revenue and margin it actually causes, measured against its full cost over a period long enough for that value to show. It is not the total revenue of everyone who happens to be a member.

Sooner or later, someone asks whether a loyalty program is worth it. It is the right question and a hard one to answer well, because loyalty value arrives through indirect channels, such as retention, deeper spending, data and advocacy, and separating the program's effect from what would have happened anyway takes deliberate measurement. The answer matters: McKinsey reported in 2021 that "around two-thirds of established loyalty programs fail to deliver value, with many actually eroding value."

This guide covers where loyalty value comes from, what a program really costs, how to calculate its return honestly with a worked example, the value that resists measurement and the mistakes that make programs look better or worse than they are. A note on the evidence: some widely quoted loyalty statistics are hard to trace to a source, so every statistic here is dated and attributed, the worked example is labeled illustrative, and the guide says where a number shows an association rather than a cause.

Loyalty programs create value through retention, increased spend, engagement, first-party data and advocacy, and only some of these are easy to measure.

Retention is the foundation. Research by Frederick Reichheld of Bain & Company, summarized in Harvard Business Review in 2014, found that "increasing customer retention rates by 5% increases profits by 25% to 95%." It is older research on retention in general, not on loyalty programs, so treat it as direction rather than a forecast. The mechanism is what matters for a program: retained customers do not have to be reacquired and tend to buy more over time.

Spend and engagement sit on top. McKinsey found in 2021 that "a typical active loyalty-program member spends 10 percent more than someone who is enrolled but not active," and redeemers spend 25 percent more than enrolled but inactive members. Gallup reported in 2014 that a fully engaged customer represents an average 23% premium over the average customer, measuring engagement as how emotionally connected customers feel to a business rather than as program activity. Both are comparisons between groups of customers, so they show where value concentrates, not how much a program created; that is the measurement problem this guide returns to. First-party data and advocacy are real sources of value even when they resist a clean number.

A full cost includes design and build, the technology platform, operations and staff, marketing and the rewards themselves, not the rewards alone.

Program costs are easy to underestimate because the most visible one, the rewards, is only part of the picture. A full accounting includes the cost of designing and building the program, the platform that runs it, the operations and staff that manage it, the marketing that promotes it and the rewards members earn. In the worked example below, rewards are $300,000 of an $850,000 first-year cost, so a rewards-only view would miss almost two-thirds of the investment.

Rewards need particular care. Points that members have earned but not yet redeemed are typically carried as a liability until they are redeemed or expire, and how much of that balance is expected never to be redeemed (breakage) affects the accounting. Confirm the treatment with your auditors. A high unredeemed share can look like a saving, but it often signals members who are not engaging, one of the warning signs covered in the guide to why loyalty programs fail. Budget reward costs at the rate members are expected to redeem, and treat a growing unredeemed balance as a prompt to look at the program, not as a profit. A program erodes margin when the rewards it pays cost more than the incremental margin it creates, for example when members are rewarded for purchases they would have made anyway, or when earn rates rise faster than the extra spend they produce.

ROI is the incremental margin a program creates minus its full cost, divided by that cost, and the hard part is the word incremental.

The trap is selection bias. A brand's best customers are the ones most likely to join a loyalty program, so members will look more valuable than non-members whether or not the program changed anything. Counting all member revenue as program-driven can overstate ROI many times over. The stronger method compares members with a holdout: otherwise similar customers, or regions, not offered the program, or offered it later in a staggered rollout. A matched group of non-members is the next best option, though customers who chose not to join may still differ from those who did. Measure the gap over a full year or more as well, because bonus offers can pull forward purchases members would have made later, which looks like lift over a short window.

A holdout has prerequisites. The brand needs spend data for the holdout as well as for members: store or regional sales for a regional holdout, or identified transactions from accounts, ecommerce or payment data for a customer holdout. Without that data there is nothing to compare. The holdout has to be large enough and run long enough for the gap to stand out from normal variation, and customers in it should not be reached by the program through other channels. Finance should sign off on the design before launch, because the held-out customers are deliberately kept out of the program for the test period. Where a true holdout is not practical, as in a B2B program with a small number of large accounts, compare enrolled accounts with similar accounts that enroll later, and present the result as weaker evidence.

Enrolled-versus-not comparisons are still useful as a signal. A leading Canadian regional distributor running its program on BENGAGED™, Brandmovers' B2B loyalty and incentives platform, reported that "Sales among enrolled customers grew by an average of 25%, while non-enrolled customers saw only a 5% average increase" (disclosed by Brandmovers; case study). Enrollment was not randomly assigned, so part of that gap may reflect which accounts joined rather than what the program did. It is a reason to run a holdout, not a substitute for one.

Whatever the platform, the design decision comes first: decide who is held out, for how long and how the comparison will be read before launch. The guide to the loyalty ROI calculation framework covers the formulas in more depth.

A worked example shows why incremental margin, full cost and time horizon change the answer.

The numbers below are illustrative, not drawn from a client, and keep membership flat at 100,000 for simplicity. A retailer launches a program in some regions and holds out comparable regions for two years, comparing the customers who join with holdout customers who matched them on spend and visit frequency in the year before launch.

Three things stand out. First, the naive view would credit the program with all $52 million of member spend in year 1 and report an ROI in the thousands of percent; the incremental view credits it with $2 million. Second, year 1 shows a small loss because the build cost lands up front, which is why a program judged on its first year alone can look weak. The example assumes the spend gap widens in year 2; if it stays at $20, year 2 ROI falls to 0% and the two-year result to a small loss, so the year 2 assumption deserves as much scrutiny as the year 1 cost. Across both years combined, $2.4 million of incremental margin against $1.65 million of cost gives an undiscounted ROI of about 45%; a finance team that discounts year 2 margin to present value will see a somewhat lower figure. Third, the result depends on the holdout: if the holdout regions differ from the program regions in ways the matching missed, the incremental figure is wrong in either direction. The guide to building a CFO-ready business case shows how to present a calculation like this to finance.

The value of a loyalty program is real, but it is not automatic, and it is not what a count of member revenue suggests. Loyalty can pay through retention, spend, engagement and data, and it costs more than its rewards. The reliable way to know whether a program is worth it is to measure the margin it incrementally creates against its full cost, over a horizon long enough for the value to show, while being clear about the strategic value that resists measurement. Done that way, most of the question of whether a program is worth funding becomes a matter of evidence, and the rest is a judgment stated openly.