Loyalty Card Completion Rates: A 23,000-Stamp Study

We analysed 23,296 loyalty card stamps from 10,690 unique customers across 1,013 small businesses over four months in 2026. We wanted to answer a question that has been argued about for two decades without much data behind it: what actually makes a loyalty card get completed?

The answer is not what you would expect from reading the existing literature. It is not the stamp count. It is not the reward. It is not the industry. There is one variable that matters: how long the customer has to wait for the reward. Most small businesses get this badly wrong.

This post is the full study. Methodology, tables, caveats, and the practical implications for anyone running a loyalty programme. If you want the same numbers in quick-reference format with an interactive wait-time calculator, see the Loyalty Card Completion Rate Benchmark — kept evergreen and updated quarterly.

Quick summary: five findings worth citing

1. The 14× spread. Loyalty cards designed to complete in 2-4 weeks see an 11.4% completion rate. Cards designed to take 16+ weeks see under 1%. Same product, same psychology, fourteen times the result.

2. The 8-week threshold. Completion rates collapse beyond 8 weeks of expected wait time. Under 8 weeks: 9-11% completion. Eight to sixteen weeks: 4%. Over 16 weeks: under 2%.

3. The halving rule. Every doubling of expected time-to-reward roughly halves the completion rate. This is monotonic and dramatic across our entire dataset.

4. Visit-frequency mismatch is the silent killer. Coffee shops (daily visits) hit 10.2% completion. Salons (monthly visits) hit 1.1%. The gap is not because daily-visit businesses are better at loyalty. They are using the same 8-10 stamp cards, but those cards translate to a 2-week wait instead of a 10-month one.

5. More than half of small business loyalty cards are mathematically set up to fail. 56% of cards require more than 8 weeks of repeat visits to complete. 12% require more than 8 months. The merchant chose the design; the cycle was over before it started.

The 8-Week Threshold

We need a single concept to make this practical. Here it is: the 8-Week Threshold.

Below 8 weeks of expected wait time between the first stamp and the reward, loyalty cards behave the way the marketing textbooks suggest. Customers come back, the streak builds, and roughly one in ten cycles finishes with someone redeeming the reward. Above 8 weeks, the relationship breaks. The card stops being a habit. By 16 weeks, you have a piece of paper (or a digital pass) that nobody thinks about anymore.

The 8-Week Threshold is the most actionable finding in our study, because every merchant can calculate where they sit on it with one line of arithmetic. We’ll get to that in §2.

Section 1: Time-to-reward determines completion

For each active loyalty card in our dataset, we computed the expected weeks to reward as:

(days between typical customer visits × total stamps required) ÷ 7

We then bucketed cards and measured actual completion behaviour from the underlying stamp events. A completion is defined as a customer reaching the required stamp count within a given cycle.

Table 1: Completion rate by time-to-reward (n = 5,277 customer-cycles across 128 cards)

Weeks to reward	Cards	Customer-cycles	Completion rate
2-4 weeks	35	1,320	11.4%
4-8 weeks	41	1,905	9.3%
8-16 weeks	29	1,035	4.0%
16-32 weeks	11	648	1.2%
32+ weeks	12	369	0.8%

The pattern is monotonic and the cliff is sharp. The completion rate at 2-4 weeks (11.4%) is 3× higher than at 8-16 weeks (4.0%) and 14× higher than at 32+ weeks (0.8%).

Here is the cleanest single-sentence version: loyalty cards designed to complete in under 8 weeks see roughly 3× higher completion than cards taking 8-16 weeks, and 10× higher than cards taking 4+ months.

The 16+ week buckets should be read as floors, not actuals. Our observation window is four months, so some in-flight cycles will complete after our window ends. The 2-4 / 4-8 / 8-16 week buckets are reliable; the longer buckets cannot rise above what time-to-reward physically allows.

Section 2: The Wait-Time Formula

The practical takeaway from Section 1 fits on a napkin:

Wait-Time Formula: days between visits × stamps to reward ÷ 7 = expected weeks to complete.

If the result is over 8, completion rates collapse.

Three worked examples, using the actual customer-visit cadences declared by merchants in our dataset:

• Coffee shop, 2 days between visits, 10 stamps: 2 × 10 ÷ 7 = 2.9 weeks. Sweet-spot zone, double-digit completion likely.
• Restaurant, 14 days between visits, 8 stamps: 14 × 8 ÷ 7 = 16 weeks. Failure zone, completion under 2%.
• Hair salon, 30 days between visits, 8 stamps: 30 × 8 ÷ 7 = 34 weeks. Mathematically dead. The card will be in a drawer somewhere by stamp 3.

If you run a small business with a loyalty programme right now, run the formula on your own card. You may discover, as 56% of the merchants in our dataset have, that the card was over before it started.

Section 3: The visit-frequency mismatch

There is a related and equally large effect hiding inside the time-to-reward story. We segmented cards by the parent business’s declared visit frequency and measured completion separately.

Table 2: Completion rate by declared visit frequency (n = 5,277 customer-cycles)

Visit frequency	Cards	Customer-cycles	Completion rate
High (~daily, ≤ 3 days between visits)	75	3,204	10.2%
Medium (~weekly, 4-14 days between visits)	31	1,105	3.9%
Low (~monthly, 15-35 days between visits)	22	968	1.1%

A 9× gap between daily-visit and monthly-visit businesses.

This is the cleanest demonstration that the wait time, not the customer, is the variable. A coffee shop customer is not “more loyal” than a salon customer; they simply complete the card before they have time to forget it exists. The same person, with the same intent, would forget a 10-stamp salon card in three months.

Most marketing advice on loyalty cards talks about engagement and customer commitment. Our data suggests that is the wrong frame. The variable that matters is customer memory under a wait constraint, and memory is fragile. Eight weeks is a lot to ask of it.

Section 4: How most merchants actually set up their cards

Here is where the picture gets uncomfortable for the industry. The vast majority of small business loyalty cards are designed with no apparent reference to the visit frequency of the underlying business.

Cross-tab: visit frequency × stamps required (active cards, n = 2,008)

Visit frequency / Stamps required	1-4	5	6	7	8	9	10	Row total
High (daily-ish)	<50	<50	<50	<50	285	<50	271	679
Medium (weekly-ish)	<50	<50	162	<50	59	<50	102	390
Low (monthly-ish)	175	81	75	<50	58	<50	118	514
Not set	<50	<50	<50	<50	194	<50	123	397
Column total	237	160	325	37	597	32	620	2,008

(Cells under 50 are masked for k-anonymity.)

Look at the low-frequency row. Of the 514 monthly-visit businesses with active loyalty cards, 118 of them chose a 10-stamp card, which mathematically requires 10 months to complete at the customer’s natural visit cadence. Another 58 chose an 8-stamp card (8 months). These cards do not have a completion problem because of bad luck. They have a completion problem because no one redeems a card they have been carrying for nine months.

Distribution of expected wait time across the platform

Of the 1,611 active cards with declared visit frequency:

Weeks to reward (declared)	Cards	Share
Under 2 weeks	22	1%
2-4 weeks	391	24%
4-8 weeks	324	20%
8-16 weeks	350	22%
16-32 weeks	329	20%
32+ weeks	195	12%

• 44% (715 cards) fall inside the 8-week sweet spot. These cards are likely seeing double-digit completion.
• 22% (350 cards) require 2-4 months. Borderline; about 4% completion in our sample.
• 32% (524 cards) require more than 4 months. The majority will see under 2% completion.

The headline-sized version: roughly one in three small business loyalty cards is mathematically set up to take more than 4 months to complete, and one in eight requires more than 8 months.

Section 5: Why this happens (the behavioural science)

The wait-time finding is not new in academic literature, but it has rarely been measured in the wild at this scale. Two classic studies are worth reading alongside our results, because they explain why time-to-reward dominates the completion equation.

The Goal-Gradient effect (Kivetz, Urminsky and Zheng, 2006)

In a now-famous field experiment, Ran Kivetz and colleagues handed coffee shop customers one of two loyalty cards: a 10-stamp card with no head-start, or a 12-stamp card with 2 stamps already filled in. Both cards required 10 new stamps. The 12-stamp card with the head-start was completed in 12.7 days on average; the empty 10-stamp card took 15.6 days. Customers also accelerated their purchases as they got closer to the reward. This is the “goal-gradient” effect, borrowed from older behaviourist work.

The implication: proximity to the reward changes customer behaviour. The further the reward feels, the slower the customer moves; the closer the reward feels, the more they accelerate. Long cards remove this acceleration entirely because the reward is never near.

Published in Journal of Marketing Research: The Goal-Gradient Hypothesis Resurrected.

The Endowed Progress effect (Nunes and Drèze, 2006)

In a parallel study at the same time, Joseph Nunes and Xavier Drèze ran an experiment with a car wash loyalty card. One group received an 8-stamp empty card; another received a 10-stamp card with 2 free stamps already given. Both groups had to collect 8 real stamps. The group with the head-start completed the card at a 34% rate; the empty 8-stamp group completed at 19%.

The implication: a card that already looks “started” is psychologically much closer to done than a blank one. This is why the first stamp is the highest-leverage marketing moment in any loyalty programme.

Published in Journal of Consumer Research: The Endowed Progress Effect.

Why our field data goes further

Both classic studies tested psychological framing on cards of similar length. Our study measures the opposite end: what happens when card length itself changes by an order of magnitude, in real businesses, in 2026. The two findings combine cleanly:

• The 8-Week Threshold is the upper bound on goal-gradient acceleration. Beyond it, customers are too far from the reward for the acceleration to kick in.
• The endowed-progress finding suggests merchants can claw some completion back by giving first-visit bonus stamps, but only if the underlying card is short enough that the reward is in psychological reach to begin with.

Long cards do not survive even with the best behavioural-science tricks layered on top. The card has to be short first.

Section 6: Recommendations by industry

Combining the time-to-reward data, visit-frequency data, and the behavioural science yields a defensible per-industry recommendation. We do not believe in “8 stamps is the universal answer”. The right number is whatever produces a 2-8 week expected wait.

Industry archetype	Typical visit cadence	Recommended stamp count	Expected completion window
Coffee shop / cafe / fast food	Daily or every 2-3 days	8-10 stamps	2-4 weeks
Fast-casual restaurant	Weekly	5-6 stamps	5-6 weeks
Hair salon / barber shop	Monthly (4-8 weeks)	3-5 stamps	3-5 months (still risky; consider tiered or visit-based rewards instead)
Nail salon	2-3 weeks	5-6 stamps	10-18 weeks (use endowed-progress: 2 free stamps to start)
Med spa / wellness	4-8 weeks	3-4 stamps	12-24 weeks (use scheduled rewards rather than stamp-card model)

A general principle from our data: if the right number of stamps for your industry produces an expected wait of more than 8 weeks, you probably do not want a stamp card. You want a visit-based reward or a tiered programme. Stamp cards are a high-frequency mechanism. Forcing them onto low-frequency businesses is the single most common mistake on the platform.

You can see this applied across our coffee shop loyalty programme guide and the hair salon loyalty programme guide. Same underlying product, very different stamp-count recommendations because the wait-time arithmetic forces it.

Mitigation: spread the reward, do not stack it

If the Wait-Time Formula tells you your card is in the failure zone but you cannot easily switch business models, there is a partial fix: stop treating the final stamp as the only reward.

Place small rewards along the card. A 10-stamp card might give a 10% discount at stamp 3, a free side at stamp 6, and the full reward at stamp 10. This converts one long wait into three shorter ones. Each milestone also re-activates the goal-gradient effect described in §5: customers move faster as they approach each reward, not just the final one. The Endowed Progress effect compounds this further if the first reward sits close to the start of the card.

On FaveCard, rewards can be assigned to any stamp position (1, 2, 3, etc.) with no limit on how many rewards a single card carries. This is the case on both the Free and Pro plans. The one practical constraint to plan for: stamp count and reward positions are fixed at card creation. The card’s name, design, and reward text remain editable, but to restructure the stamps or move reward positions you have to create a new card and bring customers across. So decide your tiered structure before you launch.

This is not a product pitch. Multi-position rewards are mechanically possible on most paper cards and on several competing digital platforms. For any merchant whose Wait-Time Formula puts the final reward more than 8 weeks away, redistributing the reward across the card is the single highest-leverage change available short of redesigning the entire programme.

Section 7: Platform-scale context

For reference, the totals behind everything above (FaveCard platform data, 20 January – 18 May 2026):

• 23,296 stamp events recorded
• 10,690 unique customers received at least one stamp
• 1,013 unique small businesses actively used loyalty cards in the window
• 10,947 unique loyalty cards (digital passes) received at least one stamp
• 1,099 completed cycles observed

These are floors. They reflect what landed in our event store during the four-month window, not all-time totals.

Methodology

We are sharing the full method so the numbers can be challenged or replicated.

Sources. Two datasets:

1. Loyalty card configuration data from the FaveCard production database (2,008 active cards as of 18 May 2026, scanned with read-only credentials).
2. Behavioural event data from PostHog (favecard.co project), covering all stamp_added events between 20 January and 18 May 2026.

Filter. Active cards only (excluded 516 retired ones). For visit-frequency analysis: 1,611 cards whose parent business had declared a visit cadence. For completion analysis: 159 cards with ≥15 customer-cycles observed in the window (below this threshold, per-card completion rates are too noisy to publish).

Definitions.

• Customer-cycle = a unique (customer, card, cycle number) tuple observed in stamp_added events. Multiple cycles per customer are counted separately.
• Completion = a customer-cycle where the maximum stamp count reached or exceeded the card’s required stamp count.
• Time-to-reward = (declared days between visits × card stamp count) ÷ 7, in weeks.

Aggregation. All metrics are aggregate counts and percentages. k-anonymity applied at 50 cycles per published bucket. No identifiers (companyId, customerId, templateId, business names, geo) appear in this study.

Biases we are aware of.

1. The 4-month observation window means long-cycle cards (16+ weeks) under-report completion. Some in-flight cycles will complete after our window. The 16-32 week and 32+ week buckets are floors, not actuals. The headline finding (the 8-Week Threshold) is robust to this bias because it concerns the short-cycle buckets where cycles fully fit in the window.
2. Completion analysis runs on the 159 most active cards (≥15 cycles in window). These are the busier half of the platform; the full-platform completion rate is likely lower, not higher. Published rates are ceilings for the full population.
3. The 5-stamp card finding (27.9% completion) is striking but rests on only 8 cards. Directionally robust; precise percentage worth re-running with more data later.

Reproducibility. The aggregation scripts, masking logic, and PostHog queries that produced these numbers live in our private data-request system. The methodology is fully transferable to any loyalty card platform with stamp-event logs; the same queries would produce comparable results elsewhere.

How to cite this study

If you are writing about loyalty programmes and would like to reference this data, use:

FaveCard (2026). Loyalty Card Completion Rates: A 23,000-Stamp Study. Analysis of 23,296 loyalty card stamp events from 10,690 unique customers across 1,013 small businesses, 20 January – 18 May 2026. Retrieved from https://www.favecard.co/en/blog/loyalty-card-completion-rate-study/

We re-run this analysis quarterly. The next update is scheduled for September 2026, when we will have a longer observation window for the 16+ week cohort.

What this means if you run a small business

Three things, in order of impact:

1. Run the Wait-Time Formula on your own card right now. Days between typical customer visits × stamps to reward ÷ 7. If you are over 8 weeks, you are in the failure zone, and no marketing tactic will rescue the design.

2. Cut stamps, do not add features. The instinct, when a loyalty card underperforms, is to make the reward bigger. Our data suggests the better lever is making the path shorter. A smaller reward at 5 stamps will outperform a bigger reward at 10 stamps for any business that visits its customers less than weekly.

3. If your business is genuinely low-frequency, do not run a stamp card. Use a digital stamp card only where stamp cadence matches visit cadence. For low-frequency relationships (salons, med spas, dentists), consider scheduled rewards, tiered programmes, or anniversary-based incentives. See how to create a loyalty programme for the longer version of this decision tree.

---

The deeper point is that loyalty cards are not “set and forget” marketing. They are an explicit deal with your customer: come back N times within a reasonable window, and we will make it worth it. When the window stops being reasonable, the deal breaks. Silently, and at scale. More than half of the small businesses we measured are running cards in that broken zone. Most of them do not know it, because no one ever ran the arithmetic.

Run the arithmetic.

---

Lukasz Fryc is the founder of FaveCard. This study reflects FaveCard platform data only; it is not intended as a benchmark for the full small business loyalty industry. We will update the dataset and republish quarterly. If you would like to be notified when the next iteration is published, or if you have a complementary dataset and would like to collaborate, please get in touch via our contact page.