Loyalty program KPIs: the metrics that actually matter
Most loyalty dashboards celebrate the wrong number. Registrations climb, the launch review applauds — and six months later nobody can say whether the program moved a single incremental carton. This guide sets out the metric ladder for channel programs — from enrolment through activation to incrementality and payback — with the realistic ranges Indian trade programs actually hit, and the dashboard cadence that keeps a retailer loyalty program honest from war room to boardroom.
The metric ladder: from vanity to value
KPIs form a ladder, and each rung only means something if the rung below is healthy. Enrolment is a field-force output. Activation is trade belief. Frequency is habit. Share-of-wallet is behaviour change. Incrementality is money. Payback is the only number the CFO ultimately cares about. Programs get into trouble when they report from the bottom of the ladder ("2 lakh registrations!") while the top rungs are empty — and the reverse discipline, refusing to celebrate enrolment until first scans follow, is what separates programs that compound from programs that need the treatment described in our guide to reviving failing programs.
The nine KPIs, with realistic ranges
Enrolment coverage — % of addressable universe registered
Registered users ÷ addressable universe (counters in coverage, estimated influencer population). Targets: 60–70% of active counters in a district within 4–6 weeks of a drive; influencer universes take longer because they are estimated, not listed. The trap: enrolment is a KYC form, not a customer — it measures field-force effort, nothing else. Never bonus the field on enrolment alone; pay on enrolled-AND-activated.
Activation — % enrolled with a first scan in 14–30 days
The single best early predictor of program success. Healthy: 50–70% of enrolments scanning within 30 days. Below 30% points to onboarding friction, un-coded stock still flushing through the channel (allow 4–8 weeks post-launch), or rewards too small to bother with. Cohort it by enrolment week so you can see whether fixes are working — a blended activation number hides everything useful.
30-day active % — the program's heartbeat
Users with at least one scan in the trailing 30 days ÷ total enrolled. Steady-state health: 35–50% for counter programs, 25–40% for influencer programs (site work is lumpy). Watch the trend, not the level — a program drifting from 45% to 38% over a quarter is telling you about scheme fatigue months before revenue shows it. Segment by tier and geography; a healthy national average routinely hides dead states.
Scan frequency and depth per active user
Scans per active per month, and value scanned as % of the user's estimated brand purchases. Frequency norms differ wildly by category — an electrician might scan 8–15 coils a month, a cement counter 60+ bags — so benchmark against your own early cohorts. Depth matters more than frequency: a counter scanning 20% of what it buys from you is either not habituated or leaking codes to someone else, and both need attention. Rising frequency with flat purchases can also be a fraud tell, which is where KPI 7 comes in.
Share-of-wallet lift in enrolled counters
Your brand's share of the counter's category purchases, before vs after enrolment — estimated from scan volumes, invoice OCR and field audits. This is the behaviour the always-on budget exists to buy. Realistic movement: 5–12 percentage points over 9–12 months in contested categories. A worked check: a counter buying ₹2L/month of category at 32% your share moving to 40% is ₹16,000/month of incremental revenue against perhaps ₹1,500 of rewards — the ratio that makes the whole model work.
Incremental lift vs matched control counters
The honest version of KPI 5. Match enrolled counters to non-enrolled counters on pre-program volume, geography and outlet type; the growth gap between groups is your incrementality — everything else is selection bias, because the counters that enrol first were always your friendlier ones. Where holding out counters is politically impossible, stagger district rollouts and use later waves as temporary controls. Design this measurement into the pilot (see our pilot design guide) — it cannot be reconstructed afterwards.
Redemption rate and breakage
Redeemed value ÷ earned value. Healthy catalogue programs: 70–85% redemption (15–30% breakage). Read low redemption as a dying-engagement alarm, not savings — points nobody redeems are points nobody values, and the unredeemed liability quietly compounds on finance's books. Track time-to-first-redemption too: a user who redeems once trusts the program roughly twice as much as one who has only earned. Instant-UPI designs bypass the metric; value moves at scan time.
Fraud-flag rate and confirmed leakage
Flagged scans ÷ total scans, and confirmed fraud as % of reward spend. Normal: 2–5% flagged, under 1% of spend confirmed lost after geo-clustering, velocity rules, dealer-pattern detection and UPI-name-to-PAN checks. Two failure modes: a near-zero flag rate (the engine is asleep, not the channel honest) and confirmed leakage above ~3% of spend (dealer bulk-scanning or code harvesting is outrunning controls). Fraud KPIs belong on the main dashboard, not a side report — leakage is budget walking out of the door.
Cost per incremental rupee and payback period
Total program cost ÷ incremental contribution margin (from KPI 6, at contribution margin, not revenue). Well-run programs reach monthly contribution breakeven in 9–15 months — year one near-neutral as serialisation, platform and launch costs front-load; year two clearly positive as lift deepens against amortised cost. Run the arithmetic in the ROI calculator and report it quarterly with the same matched-control rigour every quarter, or the number decays into advocacy.
Compliance and health metrics that ride alongside
TDS 194R tracking. Number of participants approaching and past the ₹20,000 per-PAN annual benefit threshold, TDS deducted or grossed up this quarter, and payouts blocked on PAN mismatches. The brand carries the compliance risk, and stuck payouts are also a top-three driver of helpdesk tickets — so this is simultaneously a finance metric and a trust metric. Payout success rate (target: above 98% first-attempt) and median scan-to-money time belong here too: in trade programs, payment latency is the loudest quality signal the user ever experiences. Helpdesk ticket mix rounds out the set — a spike in "invalid code" tickets is a printing-batch problem surfacing; a spike in "payout pending" is the rail or the TDS logic misfiring.
Dashboard cadence: who sees what, how often
- CXO — monthly, five numbers. 30-day active %, share-of-wallet lift, incremental contribution vs control, total spend vs budget, confirmed fraud %. One page. The moment a CXO review needs a data dictionary, the review stops happening.
- Trade marketing — weekly, the full ladder. Enrolment and activation cohorts, frequency and depth by segment, redemption and catalogue mix, booster performance, fraud queue, TDS pipeline. This is the operating review where point values and scheme rules actually get tuned.
- Field / RSM — daily-to-weekly, territory cut. Which enrolled counters never activated, which actives went silent for 21+ days (the win-back list for the next beat visit), leaderboards for the salesman's own patch. The dashboard the field uses is the one written in the field's own units — counters and visits, not percentages.
- Launch phase — daily war room, as described in the 90-day launch playbook: enrolments, first scans, payout success, tickets, fraud flags, until hypercare exits.
One design rule across all of them: every metric gets an owner and a threshold that triggers action. A dashboard nobody is obliged to act on is a screensaver. And resist metric sprawl — the nine KPIs above, plus the compliance set, are sufficient to run a national program; the fiftieth metric is where attention goes to die.
Frequently asked questions
What is a good activation rate for a channel loyalty program?
Healthy trade programs convert 50–70% of enrolled users to a first scan within 30 days, and hold 35–50% of the enrolled base as 30-day actives at steady state. If activation sits below 30%, the problem is usually onboarding friction, un-coded stock still in the channel, or a reward too small to bother with — not a lack of communication.
How do you measure incremental lift from a loyalty program?
Compare enrolled counters against matched non-enrolled counters — matched on pre-program purchase volume, geography and outlet type — and read the difference in growth between the two groups as incrementality. Where a control group is politically impossible at national scale, stagger the rollout by district so later waves serve as temporary controls for earlier ones.
What redemption rate should a program target?
In catalogue-based point programs, healthy redemption sits around 70–85% of earned value, implying 15–30% breakage. Very low redemption is a warning sign of dying engagement, not savings — points nobody redeems are points nobody values. Instant-UPI designs sidestep the question because value transfers at scan time.
What fraud-flag rate is normal?
Mature programs typically see 2–5% of scans flagged for review, with well under 1% confirmed as fraud after checks like geo-clustering, velocity rules and UPI-name matching. A near-zero flag rate usually means the engine is too lax rather than the channel too honest; a confirmed-fraud rate persistently above 3% of reward spend means controls need tightening before budgets leak.
How long should payback take on a loyalty program?
Most well-run channel programs reach monthly contribution breakeven within 9–15 months: incremental contribution margin from share-of-wallet lift covering total monthly program cost. Year one is typically near-neutral as fixed costs (serialisation, platform, launch) front-load; from year two the same lift against amortised costs turns clearly positive.
Should TDS deductions appear in program dashboards?
Yes. Track the number of participants approaching and crossing the ₹20,000 per-PAN annual benefit threshold, TDS deducted or grossed up, and pending PAN mismatches that block payouts. Under Section 194R the brand carries the compliance risk, and payouts stuck on KYC mismatches are also a leading driver of helpdesk tickets and trust erosion.