Somewhere in your store's settings there is probably a referral page nobody visits. Give $20, get $20. A share button, a link, a cheerful graphic. It went live a year ago, it has produced four orders, and two of those were the same customer using a second email address.
Ask the people who run retention for online stores whether these programs work and you will get two answers that sound like opposites. Max Sturtevant, whose agency runs email and text marketing for more than a hundred online brands, put it flatly in a recent breakdown of how stores keep customers coming back: he has never seen a "give $20, get $20" program actually work, and what works is word of mouth. Raul Galera, who has worked at ReferralCandy, one of the longest-running referral apps on Shopify, since 2016, says the opposite, with a condition attached: a referral program should reward "something that's already happening."
They are describing the same fact from opposite ends. A referral program does not make anyone talk about your store. If customers already recommend you, it can make those recommendations happen more often, reach the right people and show up in your numbers. If they do not, it is a megaphone pointed at an empty room, and the first job is working out which of those two stores you are.
Referred Shoppers Are Worth More Because Your Customers Choose Who to Tell
The best evidence on referral programs does not come from the companies that sell referral software. It comes from a study published in the Journal of Marketing in 2011 by Philipp Schmitt, Bernd Skiera and Christophe Van den Bulte, who tracked roughly 10,000 customers of a large German bank for almost three years. About half had come in through the bank's referral program, which paid the existing customer a small voucher for each new account. The rest came through the usual channels: advertising, direct mail and walk-ins.
Compared with similar customers acquired the same month, the referred customers were:
- At least 16% more valuable over their expected lifetime with the bank.
- About 18% less likely to leave at any given point, and that advantage held for the whole study.
- More profitable at the start, though that margin edge shrank over time as the other customers caught up.
Two details matter more than the headline number. First, the authors are careful about why. Their explanation is matching: a customer recommends you to someone they believe will actually like you, and they tend to recommend people similar to themselves. Your existing customer did the screening before the new one ever arrived. The program did not create that judgment. It paid for it, and the reward was small next to the value of the customers it brought in.
Second, the advantage was not universal. Among the bank's low-margin customers, the advantage disappeared. The program paid off where the customers involved were good ones, and not where they were not.
That is the whole argument in miniature. A referral program amplifies the recommendations your good customers already make. Point it at a store whose customers are lukewarm and there is nothing for it to amplify.
Most Stores Never Count How Many Customers Arrive Because a Friend Sent Them
Before you install anything, find out whether the recommending is already happening. You can answer that in an afternoon, with data you mostly have.
Ask every new customer how they heard about you. One question on the order confirmation page, with "A friend or family member" as one of the options and an open box for "Other." Several Shopify apps add this in a few minutes, or you can add a question to your post-purchase email. We have argued before that asking the customer is the one measurement no pixel can match, and this is the clearest case for it: a recommendation over dinner never shows up in an ad dashboard. Run it for a month before you decide anything.
Read the open answers, not just the percentages. "My sister" and "someone in my running club" tell you who your referrers are and where they talk. That matters later, when you write the share message.
Look at the second order. Pull the share of last year's first-time customers who came back to buy again. People who reorder are the ones who recommend, because nobody recommends a product they were not happy enough to buy twice.
Look at your reviews for the right kind of praise. Volume matters, but so does the specific detail in them. "Finally a sunscreen that does not sting my kid's eyes" is a sentence somebody will repeat to a friend. "Arrived fast, good product" is not. If your reviews are thin, getting more of them comes before a referral program, because the friend who clicks a referral link reads them too.
One number to be careful with is the "would you recommend us" score. It measures intent, and intent runs well ahead of action. Galera made the point on the 2X eCommerce podcast: plenty of customers who say they would happily recommend a store never get around to it, not because they changed their minds, but because life got in the way. That gap is the honest reason a referral program exists. It reminds people who already meant to.
And if the survey shows almost nobody arriving through a friend, take that seriously. Galera's warning applies in both directions: a referral program amplifies what customers already feel about you, and if some of them are unhappy, you have handed the unhappy ones a link too.
Your Best Customers Share to Get a Friend the Discount, Not to Get Their Own Reward
The default setup in most referral apps is symmetrical: the friend gets $20 off, the customer who sent them gets $20 in credit. It looks fair. It is usually not the best split, and there is research that says why.
In 2007, Gangseog Ryu and Lawrence Feick published four experiments in the Journal of Marketing on how rewards change people's willingness to recommend. Rewards raised the likelihood of a referral, but not evenly. They had the biggest effect on recommendations to weak ties, meaning acquaintances, and for weaker, less-known brands. In those cases, rewarding the person making the recommendation mattered. For close friends and stronger brands, the pattern shifted: giving at least part of the reward to the person receiving the recommendation worked better.
Think about why. Recommending something to an acquaintance carries a small social risk, and a personal reward covers it. Recommending something to your sister carries almost none. What you want there is for her to get a good deal, and being the person who handed her one is the reward.
Most small-store referrals go to close ties: partners, siblings, friends, the parent at pickup. So lean the offer toward the friend. Galera described two subscription brands that did exactly that: one gave the friend 70% off the first box and the referring customer $5 off their next one, and another gave the friend a free first month and the existing subscriber half off their next. Both work because the referring customer gets to make a generous offer, which is the thing they actually wanted to do.
Then check that the friend's offer is the best one on your site. Galera described a store whose referrals fell every weekend and recovered every Monday. The cause turned out to be a weekend sale with a bigger discount than the referral offer, so friends used the sale code instead, the referral never registered, and the customers who sent them wrote in asking where their reward was. The same thing happens quietly with the popup. If anyone on your site can get 15% off by typing in an email address, a 15% referral offer is not a gift from a friend. It is the price of an email address.
Store Credit Only Pays Off When the Referring Customer Shops Again
The friend's side of the offer is almost always a discount on the first order. The referring customer's side should depend on one question: will this person buy from you again soon?
They buy often: give credit toward the next order. Coffee, supplements, skincare, pet food, anything on a subscription. Credit brings the referring customer back, so the reward doubles as retention. There is also a good case that a balance feels more real than a coupon. The host of the Shopify1Percent podcast put it well in his interview with Galera: a $20 balance in your account feels like money you would lose by not spending, while an unused 20% coupon just feels like a coupon you did not use. That is an observation from running a store, not a study, but it matches what most of us have felt looking at a gift card balance.
They buy rarely: give cash or a gift card for somewhere else. A mattress, a sofa, a bike, a piece of jewelry. Galera's point here is sharp: offering store credit to someone who just made a purchase they will repeat once every few years asks them to spend more money with you to collect a reward they cannot use, and it can read as faintly insulting. Picture someone who just spent $2,000 on a mattress being offered $50 toward another one. A gift card they can spend anywhere, or a straight cash reward, is the honest version. Friendbuy, another referral platform, says its mattress and furniture clients use gift cards for exactly this reason.
They would rather have a thing. Sometimes neither money nor credit is the right reward. Galera told the story of a toddler-toy brand whose referrers were mostly parents sharing in parenting groups. They did not want another discount on a toy their child already had, and $10 cash moved nobody. A small collectible gift, something the kids looked forward to, did.
Whatever form it takes, size the reward against what a new customer already costs you. Here is a worked example for a store with a $60 average first order:
- Meta ads currently bring in a new customer for about $45 in ad spend.
- The referral offer gives the friend $15 off their first order and the referring customer $10 in credit, so $25 at face value.
- Even treating every dollar of both rewards as a full cost, which overstates it because the credit is spent on products you sell at a margin, a referred customer costs about $20 less than one bought through ads, before you count the app's monthly fee.
That gap is what you can afford to spend on the reward, and it is usually more than stores assume. The bank in the 2011 study paid its referrers a small voucher, and its authors concluded the reward was probably set too low for the value it brought in. Some stores put that math right on the referral page. Galera quoted a share page that read, in effect, "instead of spending this money on Meta, we would rather give it to you." It is honest, and it tells the customer exactly why the offer is generous. Working out what a new customer is worth to you, and when that money actually comes back, is the kind of math we run every week when we help online stores decide where their next customer should come from.
Nobody Recommends a Product Before It Arrives. The Thank-You Page Asks Too Soon.
Most stores ask for the referral in one place: the thank-you page, right after checkout. It is not a bad moment. People are excited about what they just bought, and some of them are already texting a friend about it. But it is an incomplete one, because the customer has not used the product yet. They can share excitement, not a recommendation.
Spread the ask across the moments when a customer actually has something to say:
- On the thank-you page, briefly, for the people who just told a friend they finally bought it.
- A few days after delivery, in its own email. Galera described this timing as consistently stronger: the customer has opened the box, tried the product and liked it. Now they can recommend it, not just mention it. This belongs in your post-purchase flow, the series of emails that runs after the order confirmation and decides whether there is a second order.
- Right after a good review. Someone who just gave you five stars and wrote two sentences about why is the warmest referrer you will ever have. Goyal recommends connecting the referral program to your review app for exactly this reason, so a positive review can prompt the referral ask.
- In the box. One card with one ask, not a card that also asks for a review, a follow and a photo. We wrote about why a packaging insert gets one request answered and five ignored, and the referral offer is a strong candidate for that one request.
There is a second benefit to the post-delivery email that Galera heard from retention managers. Anyone emailing customers several times a month runs out of reasons that are not "here is another sale," and customers learn to tune those out. A referral email is one of the few messages that hands the customer something instead of asking for money. It gives your list a reason to keep opening your emails.
Then make the link easy to find later, because the real recommendation rarely happens on your site. Manish Goyal, who founded the referral platform Friendbuy, calls these impulse referrals: two friends at dinner, your product comes up, and one of them wants to send a link right now. If finding it takes logging in, digging through old emails and guessing which menu it lives in, the moment passes. Put the referral link in the account page, in the footer of every email, and in the order confirmation. If a customer cannot find their link in about ten seconds on their phone, it might as well not exist.
If you want fewer, more deliberate referrals instead of more, give your best customers a small number of passes rather than an unlimited link. We covered that approach in our piece on gift cards, and it suits stores with a high price point or a product that only fits certain people.
A Referral Link Without a Disclosure Is an Ad Pretending to Be a Friend
Here is the part almost no referral app tutorial mentions. When a customer posts their referral link and earns something for it, they are not just a fan sharing something they like anymore. In the FTC's view, they are making a compensated endorsement.
The FTC's guidance on endorsements, in its plain-English question-and-answer page for businesses and the public, addresses this directly. If you share how much you like something you bought on your own and have no relationship with the company, no disclosure is needed. "However, if you're doing it as part of a sponsored campaign or you're being compensated - for example, getting a discount on a future purchase or being entered into a sweepstakes for a significant prize - a disclosure is appropriate." A referral reward is a discount on a future purchase. That is the example the FTC chose.
Your customers are not going to read the Endorsement Guides, and you should not expect them to. So write the disclosure into the message they share. Wherever you can set the default message that goes out with the link, make it honest and friendly at the same time:
"I've been using these for six months and love them. Here's $15 off your first order. Full disclosure, I get a credit too if you buy."
That sentence costs you nothing. It reads as more trustworthy, not less, because it is exactly how a real person talks when they have nothing to hide. And it keeps your customers, and your store, on the right side of the FTC. If you run larger campaigns or work with creators, it is worth a proper review with someone who knows the rules; this is the practical floor, not legal advice.
The other thing to plan for before launch is the people who will treat your program as a free-money machine. The common ones:
- Customers referring themselves with a second email address to stack the friend discount on their own order.
- Existing customers using a friend's code to get a first-order discount they are not entitled to.
- Links leaking to coupon sites and public forums, where strangers paste them for anyone to use. Galera's line between a referral program and an affiliate program fits here: a referral is one person sending a link to someone they know, while a link dumped in a public group is just a discount code with extra steps.
The fixes are simple, and referral apps typically include at least some of them: make the friend offer valid only for first-time customers, cap how many rewards one customer can earn, and hold the reward until your return window has closed so nobody gets paid for an order they send back. The guardrails cost nothing to set up before launch and are miserable to add after a coupon site finds you.
Your Referral Program Will Take Credit for Friends Who Were Coming Anyway
Six months in, your referral app will show you a number: referred orders, maybe with revenue attached. It will look good. Be careful with it.
The app counts every friend who used a referral link or code. Some of those friends would have bought anyway, because their sister had been raving about you for a month and they were going to order regardless. The code just showed up at checkout. The app has no way of knowing which friends the program actually brought in and which it simply tagged. It is the same trap we described in our piece on loyalty programs, where a dashboard counts every order a member places, including the ones they would have placed without the points.
The way around it is the survey you started before launch. If "a friend or family member" was 8% of new-customer answers before the program and 14% six months after, the program plausibly caused something close to that difference. If it was 8% before and 8% after, while the app reports hundreds of referred orders, the program is mostly attaching discounts to recommendations that were already happening. That is not free, either. You are now paying a reward on orders you used to get at full price.
Beyond that, keep a short list and check it monthly:
- Share rate: of customers who saw the offer, how many sent a link. If almost nobody shares, the offer or the timing is wrong.
- Referred first orders per month, and their share of all new customers.
- Cost per referred customer: both rewards plus the app's monthly fee, divided by referred customers. Compare it with what ads cost you per new customer.
- Second orders within 90 days, for referred customers versus everyone else. The 2011 bank study found referred customers stayed longer. If yours do not, the program is pulling in deal-seekers instead of the friends your customers would have picked.
That last number is the one that tells you whether the program is doing what the research says referral programs can do. A referred customer who never comes back is just a discounted order.
A Referral Program Can't Fix a Product Nobody Talks About
Sometimes the honest answer is not yet. A referral program is probably premature if:
- Very few customers come back for a second order.
- Your reviews are thin, generic or mixed.
- The "how did you hear about us" answers almost never say a friend.
- You are taking so few orders that a dozen referrals a year would be the whole program.
None of those are problems a referral app solves, and all of them make one a waste of a monthly fee. The work in that case is the work that makes people want to talk: a product that delivers, a first delivery that arrives the way you promised, and a post-purchase experience that makes a customer glad they bought. Galera's own advice to stores is the same: get the customer experience right first, then layer the referral program on top of the word of mouth it creates.
When the recommending is already happening, though, a referral program is one of the best-value acquisition channels a small store can run. The customers it brings are pre-screened by people who know them, and they tend to stick around. Lean the reward toward the friend, ask after the product has arrived, write the disclosure into the share message and measure the program against the referrals you were already getting.
If you would like help setting it up, or working out whether your store is ready for one, this is the kind of thing we look at with online stores every week. And if you would rather build it yourself, everything you need is above.
