Do loyalty programs actually work?

We analyzed transaction data across seven merchants on the Oshi network: over 40,000 customers and tens of thousands of orders. These seven were selected because they had the largest customer bases, the longest program history, and enough data to produce reliable comparisons. They span food and drink, health and wellness, and specialty consumer goods.

At each merchant, customers who engage with the Bitcoin rewards program are classified as “enrolled.” Enrollment means the customer opted into marketing, received a reward notification, and actively claimed the reward. Depending on the merchant, roughly 19–38% of customers have done this. The rest either never engaged with the reward or opted out of marketing entirely.

That gives us two groups to compare at every store: customers who enrolled in the rewards program and customers who didn’t. That comparison is the starting point for every loyalty program case study in the industry. It’s also where most of them stop.

We didn’t stop there. Every loyalty platform publishes case studies showing that their members spend more, but none of them test whether the program actually drove that difference. We wanted to find out. So we applied six layers of analytical correction that no other platform uses and published everything.

To our knowledge, this is the most rigorous loyalty program effectiveness study published by any rewards platform. This is what we found.

This is an observational study. Customers who claim rewards self-selected into the program. Our analysis controls for lifecycle stage, calendar timing, and pre-enrollment spending, but does not claim that bitcoin rewards cause increased spending. This study measures the association between reward engagement and spending growth within a bitcoin rewards program. Whether the bitcoin element itself contributes to the effect, through novelty, appreciation potential, or attracting a different customer profile, is a question our data cannot answer separately.

We could tell you that our loyalty members spend 3x more. That they come back twice as often. That they generate over half of total revenue. And it would all be true.

It’s also exactly what every other loyalty platform will tell you. These are real figures from our own merchants, the same type of numbers you’ll see in any loyalty program case study:

These are impressive numbers. And if we operated like other loyalty platforms, this is where we’d stop. We’d put these in a case study, slap on a client logo, and call it proof that our program drives results.

But these are vanity metrics. They describe a correlation between program membership and spending. They don’t tell you whether the program caused the spending, or whether high-spending customers were simply more likely to join. Your most engaged customers are the ones who sign up. They were already your best buyers. And because they shop more often, they see the reward offer more often, giving them more chances to enroll. The program gets credit for spending that most likely would have happened anyway.

So is the program actually changing behavior, or is it just identifying customers who were already great? The only way to know is to test it properly.

Instead of comparing members to non-members, which compares fundamentally different groups of people, we compared the same customers to themselves. We looked at what each customer spent in the 90 days before they enrolled, then what they spent in the 90 days after. Same person, same store, different time periods.

But even a before-and-after comparison isn’t enough on its own. Customer spending naturally changes over time. Prices change, seasons change, promotions come and go. So we applied six layers of correction to isolate the signal from the noise.

Same customer, before and after

Each enrolled customer is their own baseline. We measure their spending in a 90-day window before enrollment and a 90-day window after. No cross-customer comparisons. The question is simple: did this individual customer spend more after enrolling than before?

A matched control group

Non-enrolled customers who reached the same purchase stage at the same time provide a natural baseline. If your third-time buyers typically increase spending by 10% regardless of any program, we need to know that. The control group tells us what would have happened without the reward.

The enrollment purchase is excluded

The reward is issued at the moment of purchase. That purchase cannot be influenced by a reward that was given during it. So we exclude it from both groups. Only purchases made after the first reward count toward the result.

Calendar-time matching

If a merchant raised prices, ran a holiday promotion, or changed their product mix, we need the control group to reflect those same conditions. We matched each enrolled customer’s control group to within 45 days of their enrollment date, so both groups are measured in the same market environment.

Pre-spend matching

Two customers at the same purchase stage can still have very different spending levels. A customer who spent $200 in the pre-period is not a fair comparison for one who spent $50. We added a spending similarity requirement so enrolled customers are compared to controls with similar pre-enrollment spending.

Pre-trend check

Were future claimers already on an upward spending trajectory before they claimed? If so, the post-enrollment lift might just be a continuation of a pre-existing trend. We checked. They weren’t. Both groups had similar spending trajectories leading into the enrollment moment.

That doesn’t make their numbers wrong. It makes them incomplete. And if you’re making business decisions based on incomplete numbers, that can be expensive.

After every correction we could apply, the result was the same: positive. Enrolled customers showed meaningfully stronger spending growth than comparable non-enrolled customers at the same lifecycle stage and calendar period. After subtracting natural spending patterns from a matched control group, enrolled customers still spent significantly more.

To put that in concrete terms: a merchant spending $500 per month on bitcoin rewards at a 1% rate would be associated with roughly $5,000 or more in additional customer spending beyond what similar non-enrolled customers generated naturally.

We ran five independent analytical methods with progressively stricter controls: calendar-matched, enriched matching, regression-adjusted, multi-merchant pooled, and a pre-trend falsification check. All five came back positive. Normalized to a standard 1% reward rate, the range across merchants was $5–$22 per $1.

Why the range varies

The $5–$22 range reflects real differences between merchants. A high-ticket brand where customers spend $200 per order will see a different dollar impact than a snack company with $50 average orders, even if the behavioral lift is similar. Purchase frequency, customer mix, and product category all affect the magnitude. The direction, however, was consistent everywhere we looked.

A finding from one merchant could be a fluke. A finding from two could be a coincidence. But when the same pattern appears across all seven merchants, in different industries, with different customer bases and price points, it becomes very difficult to dismiss.

Most merchants evaluate loyalty programs based on vanity metrics because that’s all anyone gives them. The platforms don’t dig deeper, so the merchants can’t either. That means investment decisions are being made on numbers that look impressive but don’t hold up to scrutiny.

The merchants in our study spend 1–5% of purchase value on bitcoin rewards. At a 1% rate, that’s $1 for every $100 a customer spends. Even at the low end of our estimates, that’s a compelling return for a program that runs passively in the background.

Why now

The number of merchants offering Bitcoin rewards is still small. Most stores are running the same traditional points programs that have been around for decades, backed by the same unverified metrics. That creates an opening for merchants willing to try something different and hold themselves to a higher standard of evidence.

As Bitcoin adoption continues to grow, more merchants will eventually move in this direction. The ones who start now build relationships with this customer base first. They establish trust, earn loyalty, and show up in rewards networks before the space gets crowded.

Oshi connects to your existing store and lets you offer Bitcoin rewards to every customer. No technical setup. No minimum commitment. Just better data and a better rewards experience.