When few customers buy again within 90 days, should your referral reward still be a coupon?

When I talk to merchants about referral rewards during demo calls, a reason they give for choosing a coupon is that they want the customer to come back. They see a chance to bring in a new customer and encourage another purchase from the person who referred them.

I understand that. But they’re asking the same reward to do two jobs: encourage a referral while also helping persuade the existing customer to buy again.

My concern is that merchants sometimes expect too much from the coupon. A customer can like what they bought, recommend it to a friend, and have no plans to buy anything else for months because they don’t need another version of the item. They may love it and still have no use for a second, third, or fourth one. So a discount on that next purchase may feel useless, even if they’re happy with the brand.

Then there’s the margin question, which is a legitimate concern. The merchant needs an offer they can afford, which may mean keeping the discount low. If the customer can get a similar 10% or 15% discount without referring anyone, how much extra reason does the referral reward give them to act?

The size of the discount is only part of the decision because a larger coupon still requires the customer to make another purchase to receive its benefit. And this is exactly where the timing of repeat purchases becomes useful to look at.

In our analysis, we measured the percentage of new buyers who placed another recorded purchase within 90 days (yes, 90 days). A merchant might have customers who return year after year and still have a low 90-day repeat rate. For that merchant, it’s worth asking how appealing a discount on another purchase is as a reward for referring someone now.

What we found

For this study, I analyzed about 650 eligible Shopify stores over the 12 months from September 2025 through August 2026. To avoid mixing referral and affiliate programs, I excluded stores with an identified affiliate campaign active during that period.

I grouped stores by their current referral reward: was it a fixed cash amount, a percentage commission, or a coupon toward another purchase? In all three cases, the reward went to the existing customer for referring someone.

Then, for each store, I calculated referral revenue share, which is the percentage of its total revenue attributed to referrals. I calculated the median, or middle value, for each reward group within each repeat-rate band and compared those medians.

Still with me? Okay, good. The results are in the table below, which shows how the cash and commission groups compare with coupon programs in the same band.

Look at that first row. At stores where fewer than 10% of new buyers returned within 90 days, the median referral revenue share for fixed-cash programs was nearly three times that of coupon programs. For commission programs, it was more than eleven times as high.

I find that gap astonishing.

And look at how much it narrows in the next group. Among stores where 10% to below 25% of new buyers returned within 90 days, the fixed-cash comparison was 1.3 times the coupon median, and the commission comparison was 2.9 times. Commission still showed a substantial gap, but the difference was much larger among stores with fewer customers returning within 90 days.

That brings me straight back to those demo calls. A merchant may choose a coupon because they want the customer to return. But when so few new buyers come back within 90 days, I’d want to look closely at whether that reward gives the customer enough reason to refer a friend in the first place.

A separate check measured repeat buying first, then looked at referral performance in a later period. The same pattern held: cash and commission programs led the available comparisons, with a wider gap among stores with fewer customers returning within 90 days. That gives me more reason to investigate how well the reward fits the customer.

What I’d look at in your store

I’d start by looking at your store and checking how many of your new customers actually buy again within 90 days. Just choose a group whose first purchases were long enough ago that everyone has had the full 90 days to return, then calculate what percentage made another purchase within that window.

Giving yourself that full window is important because a customer who first bought last week hasn’t had the same chance to return as someone who bought months ago.

Then let’s look beyond those 90 days. If customers tend to return much later, that matters to how you interpret the number. The window gives you a way to compare your store with this study. In this case, we chose 90 days, but your customers may have a perfectly reasonable buying cycle that takes longer.

Once you’ve understood that pattern, look at the reward from the customer’s side and ask yourself a few questions: What would they want to buy next? How soon would they want it? Would the coupon still be valid? And could they get a similar offer without making a referral?

These questions help separate the benefit you want from the benefit you’re offering. You want another purchase (who doesn’t?), but the customer needs something they can see themselves using. Otherwise, another purchase may just feel like money they don’t need to spend.

And that’s precisely why I think cash can be a better fit in these cases. Cash removes the requirement to buy from your store again to use the reward, and a customer can receive something useful for bringing you a friend even if their own need for your product has already been met. That is why, for a store with a referral program and few customers returning soon, I’d put a cash reward on the list of things worth testing.

Keep the margin question in the decision

Merchants’ concerns about margins still deserve an answer, and this blog post wouldn’t be useful if I ignored the number one concern I hear during calls. If you consider cash, work through what you can afford to pay for a successful referral after accounting for the product, fulfillment, and any offer given to the friend.

Even a small cash reward could be worth testing. If customers value it and the referrals make financial sense for you, both sides benefit.

Now the flip side: a coupon only creates a redeemed discount when someone uses it on another purchase, but a cash payout has a cost without requiring that extra order.

If I were a merchant, I’d want to assess completed referred purchases and the cost of the rewards, as well as whether the referring customers later bought again. Those outcomes let you examine both goals that led you to consider a coupon in the first place.

The numbers in the table give you a reason to investigate, but your store might not look like the typical store in the study. Your own customers and costs determine whether a different reward makes sense for your store.

A satisfied customer may have no use for another item and still know someone who would love their first one, so I’d want the referral reward to make sense for that customer, too.

How the study was conducted

The analysis used ReferralCandy-recorded purchases and revenue from live, non-internal Shopify stores. Included stores had a referral campaign active during the performance period, a currently live campaign with one identifiable reward type, sales records in every month, positive annual sales, at least 100 annual orders, and at least 100 eligible new buyers. Stores with an identified active affiliate campaign were excluded. This is a selected sample of stores with sufficient recorded activity.

The repeat-rate calculation followed customers whose first recorded order had a positive value and occurred from September 1, 2025, through May 31, 2026. It measured whether they had another positive-value recorded order within 90 days. Each buyer had a full observation window. Referral revenue share used the full September 1, 2025–August 31, 2026 period, including days without referral sales.

The timing check followed new buyers from September–November 2025 through their 90-day window, then measured referral performance in March–August 2026. It also applied stricter purchase-record checks. An independent rerun reproduced the main population counts and annual reward comparisons. The additional timing and data-quality checks were reviewed by the auditor but were not independently rerun.

The results describe associations across stores. Reward amounts, product categories, program promotion, and customer mix may differ between the groups. Reward classification uses current settings, which may have changed during the measurement period. The study did not measure customers’ reward preferences or coupon redemption.

The cash and commission groups were much smaller than the coupon groups, especially in the middle repeat-rate band. The large commission ratio also reflects a low coupon baseline, and the analysis did not estimate confidence intervals or formally test whether the reward gap changes with repeat rate. At repeat rates of 25% or more, there weren’t enough comparable cash or commission stores to establish which reward performed better. Ninety days is an observation window, not a demonstrated deadline for a coupon to be useful.