The Numbers Look Great — Until You Ask Who Joined

The case for loyalty programs looks airtight on the surface. According to Capital One Shopping's research compilation, 92.7% of brand loyalty programs report positive ROI, with only 3.8% showing negative returns in 2024. A 2024 report cited by Tremendous adds that programs generate 4.8x more revenue than they cost, and members spend 12–18% more than unenrolled customers.

Those are the numbers that end up in board decks. They're also the numbers your finance team should be questioning — and increasingly, they are.

EY's analysis of loyalty program ROI names the problem directly: a company's best and most loyal customers are the ones most likely to sign up for its loyalty program in the first place. That means the classic member-vs-non-member comparison "rightfully incurs scrutiny among statistically minded reviewers of loyalty program performance, such as finance teams."

This is selection bias, and it quietly inflates nearly every headline statistic in the loyalty industry. If your biggest spenders enroll at the highest rates, then of course members outspend non-members — they did before you gave them a points card.

Consider what the standard dashboard actually shows you:

• Members spend 12–18% more — but you don't know what they spent before joining
• 76% of consumers say they spend more with brands whose programs they join — self-reported, not measured
• ROI figures of 89.6%–92.7% come from surveys of marketers who already invested in the programs
• Redeeming members show 2x–5x higher lifetime value in vendor case studies — comparisons that even the source data can't establish as causal

None of this means your program is failing. It means you can't tell whether it's succeeding. The measurement most businesses run answers the wrong question: not "did the program create this spend?" but "where did our best customers end up?"

EY's recommended fix is methodological, not technological. Pre- and post-enrollment comparisons, difference-in-difference analysis, or A/B testing for companies that can't track pre-signup purchases all produce a more robust and defensible analysis than enrollment snapshots. The goal is isolating incremental lift — spend that exists because of the program, not alongside it.

This is the same discipline that applies to any growth investment. At Worqd, the starting point for any engagement is finding where growth is actually stuck before touching anything — because crediting the wrong lever with a revenue lift leads to scaling the wrong thing. Loyalty programs are no different. A 4.8x return that mostly reflects pre-existing customer behavior isn't a return; it's a relabeling exercise.

The uncomfortable truth is that many programs may be doing real work — driving frequency, preventing churn, generating first-party data EY calls one of their primary benefits. But until you separate the customers your program changed from the customers it merely collected, every impressive number on your dashboard carries an asterisk.

Where the Evidence Is Actually Strong: Paid Programs and Real Engagement

Not all loyalty evidence is shaky. When you strip away the correlational member-vs-non-member comparisons, a few findings hold up — and they all point to the same conclusion: the money is in engagement, not enrollment.

The strongest case comes from paid membership programs. According to McKinsey research cited by Tremendous, shoppers are 60% more likely to spend more with a brand after subscribing to a paid loyalty program — compared to just 30% for free programs. Because these figures capture behavior after joining, they suggest the program itself is changing behavior, not just attracting already-loyal customers.

Amazon Prime shows the model at full scale. Prime members spend roughly twice as much as non-members, and McKinsey's analysis puts their lifetime spend at four times that of nonmembers — with about 75% of U.S. households now holding a membership. When customers pay to belong, they have a built-in reason to consolidate their spending with you.

The second defensible finding is the engagement gap. Customers who actually redeem points show a customer lifetime value 6.3x higher than non-members, and reward redeemers spend 3.1x more than non-redeemers, per Capital One Shopping's research. Yet most programs never unlock that value:

• Only 56.2% of members actively use their memberships
• 65% of consumers actively engage with fewer than half the programs they belong to
• The average consumer belongs to roughly 21 programs but is active in only about 11

In other words, the typical loyalty program is a database of dormant names. The revenue lives in the minority of members who redeem — which means the highest-leverage move isn't growing enrollment, it's activating the members you already have.

This is where measurement discipline matters. EY recommends pre/post comparisons or difference-in-difference analysis rather than raw member-vs-non-member gaps, so you can see whether engagement actually caused the lift. Track redemption rates, purchase frequency, and CLV — not just signup counts.

The same principle applies across your funnel. At Worqd, we see it constantly in lead follow-up: a CRM full of untouched contacts behaves exactly like a loyalty program full of non-redeemers. The value was never in collecting the names — it's in the fast, consistent follow-up that turns them into booked conversations. Whether it's points or prospects, activation beats acquisition.

So where is the evidence strong? Paid programs that create real commitment, and engaged members who actually redeem. Everything else — the enrollment numbers, the inflated ROI claims — is decoration on top of those two drivers.

When Loyalty Programs Backfire — and What Failures Teach You

Not every loyalty program earns its keep. Some of the biggest brands in the world have launched ambitious programs, watched the numbers, and quietly walked them back — and those retreats teach you more about program economics than a dozen success stories.

Target Circle eliminated its 1% cashback feature in April 2024, with the company acknowledging that profit margin-wise, the loyalty reward program wasn't working. That is a remarkable admission from one of the largest retailers in the United States, and it highlights the core failure mode: a flat reward on every purchase pays out on spend that would have happened anyway.

This is where EY's selection-bias warning becomes a margin problem, not just a measurement problem. Because your best customers are the most likely to join, a poorly designed program ends up subsidizing your most committed buyers. You are not creating incremental sales — you are discounting revenue you already had.

Delta learned the speed at which loyalty can evaporate. Its 2023 SkyMiles changes triggered enough backlash that it rolled back several changes within a month. When a program is perceived as taking away earned value, the response is swift and public.

The broader pattern is sobering. McKinsey observes that brand promiscuity is at an all-time high at a time when loyalty programs have proliferated. More programs have not produced more loyalty — they have produced crowded wallets, with the average consumer belonging to roughly 21 loyalty programs but active in only 11.

The common thread across these failures is a program that rewards behavior instead of changing it. Before crediting your program with sales, check whether it actually drives incremental lift:

• Measure against a control group — EY recommends difference-in-difference analysis or A/B testing, not simple member vs. non-member comparisons.
• Budget for the full cost — rewards, technology, operations, and fraud monitoring — not just the headline discount.
• Prioritize redemption over enrollment, since only 56% of members actively use their memberships.
• Design rewards that motivate spend members would not have made otherwise.

The right measurement discipline — the kind Worqd applies when evaluating whether any growth initiative is producing real incremental revenue — separates programs that lift sales from programs that quietly erode margins. A program is only working if it changes behavior, not if it merely rewards it.

How to Measure Real Incremental Lift (Not Vanity Comparisons)

Your loyalty dashboard says members spend 12–18% more than non-members. That number is probably lying to you — or at least, it isn't telling you what caused what.

The problem is selection bias. As EY points out, a company's best and most loyal customers are also the ones most likely to join its loyalty program in the first place. Compare your members against everyone else and you're mostly measuring who your members already were before they enrolled. EY notes these simple comparisons "rightfully incur scrutiny among statistically minded reviewers" — like your finance team.

EY recommends pre-/post-comparisons or difference-in-difference analysis as a more robust, defensible way to measure program impact. Difference-in-difference compares how members' behavior changed after joining against how a similar non-member group changed over the same period — so market-wide trends don't get credited to your points program.

If you can't track purchases before signup, EY suggests A/B testing instead: hold out a control group that never sees the program, and measure the gap. It's the same discipline Worqd applies elsewhere in the funnel — no vanity metrics, only lift you can defend to a skeptical CFO.

Once your comparison is honest, track the numbers that reveal what the program actually caused:

• Customer lifetime value — customers who redeem points show CLV 6.3x higher than non-members, making redemption a leading indicator of value.
• Purchase frequency — 72% of members report buying more often after enrolling; verify that self-report in your own transaction data.
• Average order value — measure whether members' basket size shifts after enrollment, not just versus strangers.
• Redemption rate — with only 56.2% of members actively using their memberships, redemption separates real engagement from dormant signups.
• Churn — members are 70% more likely to keep shopping with a brand, but only a controlled comparison proves your program moved that number.

The cleanest test is simple to describe: what would these customers have spent anyway? Vendor case studies claiming 20–71% of revenue "attributed" to loyalty programs are member vs. non-member comparisons — useful as illustrations, weak as proof. EY also reminds teams to budget the full cost side: rewards, technology, operations, and fraud monitoring, so your ROI claim survives scrutiny. Measure the increment, not the overlap, and you'll finally know whether the program earns its keep.

Five Moves That Make Loyalty Actually Pay

Most loyalty programs don't fail because the idea is wrong — they fail because of how they're measured, designed, and run. The research points to five moves that separate programs that pay from programs that just cost.

First, measure incremental lift, not member-versus-non-member gaps. Your best customers join first, so raw comparisons flatter the program. EY recommends pre/post comparisons, difference-in-difference analysis, or A/B testing before crediting a program with any sales increase. If your finance team wouldn't trust the number, don't build strategy on it.

Second, chase redemption, not enrollment. Customers who redeem points show a customer lifetime value 6.3x higher than non-members, according to program research from Tremendous — yet only 56.2% of members actively use their memberships, per Capital One Shopping's loyalty statistics. Signups are a vanity metric; engagement is where the revenue lives.

Third, test a paid membership tier. Shoppers are 60% more likely to spend more after joining a paid program, versus just 30% for free ones — a gap that suggests real causation rather than self-selection. Amazon Prime demonstrates the model at scale, with members spending multiples of what non-members spend.

Fourth, budget for the full cost — rewards, technology, operations, and fraud monitoring. Target Circle eliminated its 1% cashback in 2024 because, margin-wise, the program wasn't working. A rewards structure that outpaces profit quietly turns a retention tool into a leak.

Fifth, integrate loyalty with personalized offers. McKinsey's pilot programs integrating personalized marketing and pricing delivered a 2–4 percentage point improvement on gross margin dollars versus standard mass offers — with personalization layered on top of broad touches, not replacing them.

A quick checklist before you scale any program:

• Run an A/B or difference-in-difference test before claiming lift
• Track redemption rate, purchase frequency, and churn — not just signups
• Pilot a paid tier with a small segment first
• Model full program costs against margin, including fraud
• Pair loyalty data with personalized offers and pricing

This is the same discipline Worqd applies across retention and pipeline work: one plan, one report, and no vanity metrics. Whether it's reactivating contacts already sitting in your CRM or measuring what a loyalty program actually adds, the standard is identical — count only what you can prove, and cut what you can't.