Loyalty KPI Dashboard: 13 Metrics, Formulas, and Benchmarks

The Loyalty KPI Dashboard: 13 Metrics With Formulas, Benchmarks, and Diagnostic Interpretation

Most loyalty program dashboards measure what is easy to count rather than what determines whether the program works. Enrollment numbers go up and to the right, the reward catalog gets used, the quarterly deck shows growth, and none of it answers the question the CFO is actually asking: is this program creating value, or moving it around?

That question deserves a blunt starting point. McKinsey's research on loyalty programs found that roughly two-thirds of established programs fail to deliver value, and that many actively erode it. A dashboard that only tracks participation and activity will never detect which side of that line a program is on, because the metrics that reveal value creation are different from the metrics that reveal activity, and because the naive versions of the value metrics are corrupted by selection effects that flatter almost every program.

This guide organizes loyalty measurement into a dashboard of thirteen metrics across four domains that follow the program's own causal sequence: participation (who joins and stays active), engagement (how member behavior differs from non-member behavior), retention and value (whether the relationship deepens and lasts), and financial health (whether the economics work). Each metric comes with its formula, a worked example drawn from a single illustrative program so the numbers reconcile, an honest statement of what benchmark exists and what does not, a measurement cadence, and the diagnostic question the metric actually answers.

The recurring discipline throughout is incrementality. Members spend more than non-members in virtually every program, but most of that gap is selection, not causation: the people who join loyalty programs were already the brand's better customers. A dashboard that attributes the full member-versus-non-member gap to the program will overstate its value, sometimes enormously. The metrics that matter are the ones designed to isolate what the program actually changed.

The Four-Domain Architecture

The thirteen metrics are grouped into four domains that follow the program's causal logic. Reading them in sequence is the point: a weak metric in an early domain usually explains a weak metric in a later one, and treating the dashboard as an unordered scorecard hides those relationships.

Domain 1, Participation. Who enrolls, who stays active, and how much of the business flows through members. This is the top of the funnel: if the active rate is low, nothing downstream can be strong, because a program only affects behavior it actually touches.

Domain 2, Engagement. How member behavior differs from non-member behavior on frequency, order value, and reward usage. This is where selection effects do the most damage, and where incrementality discipline matters most.

Domain 3, Retention and Value. Whether the member relationship lasts and deepens: churn, retention, lifetime value, and share of wallet. This is where the program's long-run economic contribution shows up, if it exists.

Domain 4, Financial Health. Whether the economics work: the outstanding point liability and the program's return on investment, measured honestly on incremental margin.

Domain 1: Participation

Metric 1: Enrollment Rate

Formula: members enrolled divided by total customers, over the same period.

Illustrative: 30,000 enrolled / 60,000 customers = 50%.

Benchmark: there is no reliable universal benchmark. Enrollment rate depends almost entirely on the enrollment model: a program that auto-enrolls at checkout will post a far higher rate than one requiring a separate sign-up, and neither number means much on its own. Compare against your own trailing periods and against the friction of your enrollment path, not an external figure.

Cadence: monthly. Diagnostic: is the enrollment mechanism capturing the customers who transact, or leaking them? A low rate with low friction points to an awareness or incentive problem; a low rate with high friction points to the sign-up path itself.

Metric 2: Active Member Rate

Formula: active members (those who transacted or engaged in the period) divided by total enrolled members.

Illustrative: 18,000 active / 30,000 enrolled = 60%.

Benchmark: this is one of the few participation metrics with external grounding, and the finding is sobering. Bond Brand Loyalty's 2025 research finds the average consumer belongs to 17.4 loyalty programs but is active in only 8.8, roughly half. Deloitte finds a similar gap, with consumers enrolled in about eight programs and active in around five. So an active rate near 50 to 60 percent is typical, not strong, and a program should be measuring its own trend rather than congratulating itself for clearing a low bar.

Cadence: monthly, with the trend line mattering more than the level. Diagnostic: this is the single most important participation metric, because it defines the population every downstream metric can actually influence. A declining active rate is an early warning that precedes financial decline by one to two quarters.

Metric 3: Member Share of Transactions

Formula: transactions attributed to identified members divided by total transactions.

Illustrative: assume 55% of transactions are member-identified.

Benchmark: no universal benchmark; the meaningful comparison is your trend and your identification capability. This metric doubles as a data-quality measure: a low share may mean members are not transacting, or simply that the program is failing to identify members at the point of sale, which is a very different problem with a very different fix.

Cadence: monthly. Diagnostic: how much of the business does the program actually see? Everything the program cannot identify, it cannot measure, personalize, or influence.

Domain 2: Engagement

Metric 4: Purchase Frequency and the PF Ratio

Formula: average purchases per member in the period, and the PF ratio, member frequency divided by non-member frequency.

Illustrative: members purchase 4.0 times a year, non-members 2.0. PF ratio = 2.0x.

Benchmark and the selection warning: resist the temptation to treat a PF ratio above 1.0 as proof the program works. Members purchase more partly because the program encourages it and partly because frequent buyers are the ones who enroll. There is no credible universal PF-ratio benchmark, and any that is quoted almost certainly conflates the program effect with selection. The defensible external anchor is McKinsey's finding that top programs lift revenue from redeeming members by 15 to 25 percent annually through frequency or basket size, a figure scoped specifically to the incremental lift, not the raw gap.

Cadence: quarterly. Diagnostic: is frequency rising within the member cohort over time (a plausible program effect), or is the member-to-non-member gap simply stable (more consistent with selection)? Cohort trend is more informative than cross-sectional comparison.

Metric 5: Average Order Value and AOV Lift

Formula: total member revenue divided by member transactions, compared to the non-member equivalent.

Illustrative: member AOV $70, non-member AOV $50, a 40% higher member AOV.

Benchmark: no reliable universal benchmark, and the same selection caveat as frequency applies with full force. A 40 percent AOV gap is not a 40 percent program effect. Higher-spending customers self-select into loyalty programs. The only trustworthy version of this metric comes from a matched-control or pre-post cohort design that isolates the change attributable to the program.

Cadence: quarterly. Diagnostic: does member AOV grow after enrollment relative to a matched control, or is the gap purely a starting-point difference the program inherited?

Metric 6: Points Redemption Rate

Formula: points redeemed divided by points issued, over the period. Note this is distinct from reward redemption rate, the share of members who redeem any reward, which answers a different question.

Illustrative: 130 million points redeemed / 200 million issued = 65%.

Benchmark: genuinely sector-dependent, which is why a single benchmark misleads. Published redemption figures span roughly 15 to 50 percent across program types, with high-frequency retail at the upper end and infrequent-purchase categories much lower. Rather than compare to a headline number, read redemption in two directions at once: too low signals rewards that are unattractive, hard to reach, or hard to use, which erodes the program's motivational value; extremely high can signal reward economics that are too generous to be sustainable.

Cadence: monthly. Diagnostic: are members earning currency they never use? Low redemption is not a cost saving. It is a signal that the reward is not doing the behavioral work it was designed for.

Metric 7: Breakage Rate

Formula: the complement of the points redemption rate. Breakage rate = 1 minus redemption rate, on the same issued-points basis, a relationship formalized in loyalty actuarial practice.

Illustrative: 1 minus 65% = 35%.

Benchmark: breakage is a financial-accounting input governed by ASC 606 in the US and IFRS 15 internationally, under which the liability is recognized on the expected value of future redemptions. It varies enormously by sector: roughly 20 to 30 percent in high-frequency retail, and as high as 70 to 85 percent in travel and some B2B programs where points accumulate faster than members redeem. There is no target breakage rate, and treating breakage as profit is a trap: high breakage flatters the current-period economics while signaling that members are not engaged enough to return for rewards they already earned.

Cadence: quarterly, aligned to financial reporting. Diagnostic: is the program's apparent profitability being funded by members forgetting their points? That is not a sustainable model; it is a countdown.

Domain 3: Retention and Value

Metric 8: Member Churn Rate

Formula: members lost during the period divided by members at the start of the period. State the period explicitly, because a monthly and an annual churn rate are wildly different numbers and mixing them is a common dashboard error.

Illustrative: on a monthly basis, if 342 of 18,000 active members lapse in a month, monthly churn is 1.9%. Compounded, that is roughly 20% annually, not 1.9% annually, and the distinction matters.

Benchmark: no universal benchmark; churn is meaningful against your own trend and your definition of a lapsed member. What matters is consistency of definition and direction of travel. Behavioral churn (a member who stops purchasing) leads explicit churn (a member who unsubscribes) by months, so track the behavioral version as the early signal.

Cadence: monthly. Diagnostic: is the active base eroding faster than it is being replenished? Rising churn is the earliest financial warning on the dashboard.

Metric 9: Member Retention Rate

Formula: 1 minus the churn rate over the same period, or equivalently, members retained divided by members at the start.

Illustrative: on an annual basis, a 28% annual churn implies a 72% annual retention rate.

Benchmark: rather than a target level, the grounding worth citing is the economics of moving the number. Bain and Company's foundational retention research, associated with Frederick Reichheld, established that a 5 percent improvement in customer retention can increase profits by 25 to 95 percent, depending on industry margin structure. That is the reason retention sits at the center of most loyalty ROI models: small improvements compound.

Cadence: quarterly. Diagnostic: is the program lengthening the customer relationship, and is that improvement showing up in profit at the rate the retention economics predict?

Metric 10: Customer Lifetime Value and CLV Lift

Formula (as used here): average order value times purchase frequency times customer lifespan. Compare member CLV to non-member CLV for the lift.

Illustrative: member CLV = $70 x 4 x 3 = $840. Non-member CLV = $50 x 2 x 2 = $200. Member CLV is 4.2 times non-member CLV.

An important rigor note: the formula above is revenue-based. It is a useful directional comparison, but it is not a value figure a CFO should take into a business case unmodified, because it ignores gross margin and the time value of money. A defensible CLV for investment decisions applies the gross margin rate to revenue and discounts future years to present value. Revenue-based CLV overstates the number, often substantially, and the 4.2x lift still carries the full selection effect: the higher-value customers enrolled in the first place.

Benchmark: no universal benchmark; CLV is intrinsic to your margins, purchase cycle, and category. Its value is as a longitudinal and cross-segment comparison, not against an external number.

Cadence: quarterly or semi-annually. Diagnostic: is the modeled lifetime value of a member growing over time, on a margin basis, relative to a matched non-member?

Metric 11: Share of Wallet

Formula: member spend with the brand divided by that member's estimated total category spend.

Illustrative: if members spend an estimated 45% of their category budget with the brand, share of wallet is 45%.

Benchmark: no universal benchmark, and honest measurement is hard, because total category spend has to be estimated from survey or panel data rather than observed directly. Treat the level as approximate and the trend as the signal. Share of wallet is nonetheless one of the most strategically important metrics, because it captures whether the program is winning a larger portion of an existing customer's spending rather than simply retaining their current level.

Cadence: semi-annually. Diagnostic: is the program shifting spend the member was giving to competitors, or just rewarding spend the brand already had?

Domain 4: Financial Health

Metric 12: Outstanding Point Liability

Formula: outstanding points balance times point value times the expected redemption rate.

Illustrative: 200 million outstanding points x $0.01 per point x 65% expected redemption = $1.3 million liability.

Benchmark: this is an accounting figure, not a performance benchmark. Under ASC 606 and IFRS 15 the liability is carried at the expected value of future redemptions, which is why the redemption and breakage assumptions are not just marketing inputs but audited financial ones. The number to watch is the trend: a liability growing faster than redemptions indicates points accumulating without being used, which is the balance-sheet shadow of a disengagement problem the participation metrics should already be flagging.

Cadence: quarterly, in coordination with finance. Diagnostic: is the liability under control and consistent with the breakage assumption filed with the auditors?

Metric 13: Program ROI

Formula: (incremental gross margin generated by the program minus total program cost) divided by total program cost, expressed as a percentage. The critical word is incremental.

Illustrative, and why the honest number is lower than it looks: suppose the program has 18,000 active members and total annual cost of $1.0 million. The naive approach multiplies the member-versus-non-member revenue gap ($280 minus $100, or $180) by 18,000 members to claim $3.24 million in program revenue, a tempting 3.2x multiple against cost. That number is close to meaningless, because most of the gap is selection. A matched-control analysis that isolates the genuine program effect might show, say, $130 of incremental revenue per active member, or $2.34 million in incremental revenue. Apply a 45% gross margin and the program produces about $1.05 million in incremental gross margin. Against $1.0 million of cost, that is roughly a 5% return on a margin basis.

The distinction that the dashboard must not blur: that same program shows a 2.34x revenue multiple (incremental revenue over cost) and a 5% ROI (incremental margin over cost). Those are different lenses on the same program, and calling a revenue multiple an ROI overstates performance dramatically. Report ROI on a margin basis; if you cite a revenue multiple, label it as such. And notice how close to the line even this illustrative program sits: if the incremental-revenue estimate were even modestly optimistic, the margin-basis return would go negative. That is precisely the mechanism behind McKinsey's finding that around two-thirds of loyalty programs erode value. The naive revenue math hides it; the incremental margin math reveals it.

Cadence: quarterly for the operating read, annually for the audited business case. Diagnostic: after controlling for selection and measuring on margin, is the program creating value, and is that value funded by genuine incremental behavior rather than by breakage?

The Dashboard at a Glance

Reading the Metrics Together: Five Diagnostic Combinations

No single metric diagnoses a program. The value is in the combinations, where two metrics that each look acceptable in isolation reveal a problem when read together. These five patterns recur across programs.

Pattern 1: High Enrollment, Low Active Rate

Enrollment climbs while the active rate falls. The program is good at sign-ups and bad at giving members a reason to stay engaged. This is the most common loyalty failure pattern, and it usually traces to a weak or slow first-reward experience: members join, do not reach a meaningful reward quickly enough to build a habit, and lapse. The fix is upstream of every downstream metric, which is why active rate is the participation metric to watch.

Pattern 2: Healthy Redemption, Rising Breakage

Redemption looks fine in aggregate while breakage climbs quarter over quarter. This usually means a subset of members is redeeming actively while a growing tail earns and forgets. The aggregate redemption rate masks the divergence. Segment redemption by cohort and tier: the aggregate can look stable while the newest cohorts are disengaging, which is the leading edge of a churn problem the point liability will confirm a quarter later.

Pattern 3: Strong PF Ratio, Flat Incremental Revenue

Members purchase far more often than non-members, yet a matched-control analysis shows little incremental revenue. This is selection wearing a disguise: the program enrolled the frequent buyers and is now taking credit for behavior that predated it. A high PF ratio with weak incrementality is not a program that works; it is a measurement that has not controlled for who joined. This is the single most expensive misread on the dashboard, because it justifies continued spend on a program that is not changing behavior.

Pattern 4: Positive Revenue Multiple, Negative Margin ROI

The program clears a comfortable revenue multiple and still loses money once ROI is measured on incremental gross margin. The revenue lens counts top-line dollars the program may not have caused and ignores the cost of the margin given away in rewards. When these two readings diverge, the margin-basis number is the true one, and the divergence is the everyday form of McKinsey's value-erosion finding. A program can be simultaneously popular, growing, and value-destroying.

Pattern 5: Rising Engagement Costs, Falling Active Rate

Reward spend per member rises while the active rate falls. The program is buying activity from a shrinking core rather than broadening engagement, concentrating cost on members who were already loyal. This pattern precedes a financial deterioration that the ROI metric will register only later, which is why the participation and engagement domains are read first: they are the leading indicators the financial domain lags.

Conclusion

A loyalty KPI dashboard is only as useful as its willingness to deliver bad news. The metrics that are easy to count, enrollment, activity, redemption volume, tend to move in encouraging directions even when a program is quietly eroding value, which is why the majority of programs that fail McKinsey's value test still produce reassuring dashboards right up until the business case is questioned.

The thirteen metrics here are structured to prevent that comfortable blindness. Read in causal order, they connect a falling active rate to the incremental revenue shortfall it predicts, a rising breakage rate to the churn it precedes, and a healthy revenue multiple to the negative margin ROI hiding underneath it. The discipline that runs through all of them is incrementality: the refusal to credit the program for behavior that selection, not the program, produced.

The programs that create value are not the ones with the best-looking dashboards. They are the ones whose dashboards are built to distinguish genuine incremental contribution from the flattering arithmetic of selection and breakage, and whose owners measure ROI on the margin that survives that scrutiny. That is a harder number to report, and it is the only one worth acting on.

A note on benchmarks. For several metrics in this guide, no reliable universal benchmark exists, because loyalty performance varies sharply by sector and program design. Where that is the case, the guide says so and directs the reader to their own trailing baseline and sector peers rather than a published range. Figures presented as illustrative are drawn from a single hypothetical program for internal consistency and are not benchmarks.