Retention / Referral
How to Promote a Referral Program
// a hidden referral program refers no one
By Fracto Solutions
October 31, 2026
The short answer
Promote a referral program by surfacing it where happy customers already are, the post-purchase flow, account area, emails, and order confirmations, and asking at peak satisfaction. Make sharing a single tap with a personal link or code, remind customers it exists, and keep advocates engaged. Visibility and ease, not reward size alone, drive participation.
Why does promotion matter to a referral program?
Because a program nobody can find gets used by nobody. That much is obvious. What we should not claim, and did previously, is that a well-promoted modest reward beats a hidden generous one every time; nobody has tested promotion intensity against reward size.
What has been tested is the effect of effort. Gershon, Cryder and John found that the cost of the sharing action changes which reward structure wins: under low action cost the friend-benefiting reward performed better, and under high action cost the referrer-benefiting one did. So reducing friction is not only a volume lever, it changes the design of the incentive. That is a narrower and more useful claim than the one we were making.
We have also dropped the assertion that most referral programs fail from invisibility. No data on why referral programs fail exists that we could find. The only prevalence figure anywhere in the material is Shopify’s unattributed “healthy referral rate” of 1% to 3%, which is not a failure analysis. If you want to make the invisibility argument, make it from your own audit data and say so.
Where should I surface the program?
At the points where a satisfied customer is already paying attention: the post-purchase flow, the order confirmation, the account area, a dedicated email, and a findable on-site entry. These are placement conventions, not measured rankings, and we have relabeled the table to say so.
Placement
Why we recommend it
Basis
Post-purchase flow
The customer has just had the experience
Practitioner judgment
Order confirmation
Among the most-opened messages you send
Transactional email benchmarks
Account dashboard
Repeat visibility without a send
Practitioner judgment
Dedicated email
Room to explain both sides of the reward
Practitioner judgment
On-site entry
Findable outside a campaign window
Practitioner judgment
Four of five rows are our recommendation rather than a finding. We would rather label that than imply a ranking nobody has measured.
One thing worth adding to the list that is not a placement: a targeting filter. Van den Bulte and colleagues found referrers with under one month of tenure produced referrals whose margin fell below non-referred customers, and recommend recruiting referrers acquired at least six months earlier. Promoting the program to everyone who just bought is the opposite of that.
Whom should I ask, and when?
Ask customers with tenure, and ask the ones who arrived by referral themselves. On timing we have practitioner judgment; on targeting we have real data, and the data is the more actionable half.
Gershon and Jiang’s work on referral contagion is the useful finding here. Across a mobile cashback firm’s 41.2m-customer dataset, with a 500,000-consumer regression sample, referred customers sent invitations at 8.72% against 2.51% for other customers, and referred 31% to 57% more; 19 of 20 companies in a separate dataset showed the pattern. The authors are explicit that the field comparisons are not causal and that referral status is not randomized, so this is a targeting signal rather than a proven multiplier. Even as a signal it is a strong one: your referred customers are your most likely referrers.
On timing, we previously wrote that asking at peak satisfaction converts “far better than an untargeted blast.” We have no published test of referral-ask timing, so that is now labeled as our recommendation. It remains a sensible one, and it pairs naturally with your review requests and post-purchase sequence. One caution on combining the two: if the referral reward is routed to customers you selected for leaving a positive review, you are in the territory of 16 CFR Part 465, which prohibits incentives conditioned on a review’s sentiment.
How should the share mechanism actually work?
Give the referrer a link and let them send it themselves. Do not pre-fill the message, do not have your system email or text the friend, and do not store the friend’s contact details. This is the opposite of the one-tap share that referral advice recommends, including our own earlier version, and the reason is legal rather than aesthetic.
Start with who the friend is. They are a new contact, not your customer, so the ePrivacy soft opt-in does not reach them and Article 13(1) prior consent applies. The UK ICO is specific about the mechanism: you must comply if you send a marketing message “or if you ‘instigate’ someone else to send it,” responsibility attaches where “you must have encouraged them to send it,” and on collecting friends’ contact details the guidance states “we would therefore advise against this type of viral marketing.”
If you do receive the friend’s details you become the controller for that data and GDPR Article 14 applies, including the duty under 14(2)(f) to tell the friend “from which source the personal data originate,” which means naming the referrer, within one month at the latest.
And the pre-filled message has been litigated. In Jensen v. Capital One (W.D. Wash., February 2025) a court declined to apply Section 230 immunity to a company whose refer-a-friend feature generated the link and a pre-filled text, finding it “the sole author of the content of the text” and that it “substantially assisted its customers in formulating, composing, and sending commercial text messages.” The company’s in-app notice asking the customer to confirm they had consent did not change the result. If referrals go by SMS in the US, the TCPA’s prior express written consent standard sits behind that.
So the compliant design is narrower than the convenient one: a personal link or code, a copy button, share targets that hand off to the referrer’s own email or messaging app with the body left empty or clearly editable, no server-side send to the friend, and no retention of the friend’s address. You lose some conversion to friction. You also lose the exposure.
How do I measure whether promotion is working?
On incremental new customers and their quality against reward spend, segmented by where the referral came from. Share rate and link clicks tell you the promotion is visible, not that it is earning anything.
Track share rate by placement so you can tell which surfaces actually produce sends, then track the conversion and the first-order margin of the customers those sends bring in. The reason to keep those separate is the research: referral value varies enough across segments that Schmitt and colleagues explicitly recommend “a selective approach,” and referrer tenure moves referral quality measurably. A placement that produces many low-quality referrals is not a win.
Then watch the one durable number. In the dyad study, referred customers churned about 40% less while their referrer remained a customer and considerably more once the referrer left, so referral retention is partly a function of referrer retention. That means your referral reporting should follow the referrer as well as the referral, which almost no referral app does by default. It is the same measure-and-improve discipline behind our growth-strategy pillar.
We put the referral program where happy customers will actually use it.
We surface the program across the post-purchase flow, confirmations, account area, and emails, trigger asks at moments of genuine satisfaction, and reduce sharing to a single tap, so a good program actually gets shared instead of sitting unseen.
It’s the same measurement-first approach behind our lifecycle and conversion optimization work, where we rebuild the buying path and measure the result .
// we would rather show up in your reporting than in your inbox
Frequently asked questions
Where should I promote my referral program?
The post-purchase flow, order confirmation, account area, a dedicated email and a findable on-site entry. Four of those five are our recommendation rather than a measured ranking, and we would rather say so. Add a targeting filter too: referrers with under one month of tenure produced referrals whose margin fell below non-referred customers.
Who are my most likely referrers?
Customers who arrived by referral themselves. Across a 41.2m-customer dataset, referred customers sent invitations at 8.72% against 2.51% and referred 31% to 57% more, with 19 of 20 companies in a separate dataset showing the pattern. The authors note the comparison is not causal, so use it to target rather than to forecast.
Should the share be one tap with a pre-filled message?
No. That design was litigated in Jensen v. Capital One, where a court declined Section 230 immunity because the company authored the pre-filled text and substantially assisted in sending it, and an in-app consent confirmation did not help. Give the referrer a link and let them write and send it from their own app.
Can my system email the friend directly?
Not on the referrer’s consent. The friend is a new contact, so the ePrivacy soft opt-in does not apply and prior consent is required. The UK ICO holds you responsible where you instigated the message and advises against collecting friends’ contact details. If you do hold them, GDPR Article 14 requires you to tell the friend where the data came from, naming the referrer, within a month.
Sources
Gershon, Cryder & John (2020), “Why Prosocial Referral Incentives Work: The Interplay of Reputational Benefits and Action Costs,” Journal of Marketing Research. Eight experiments including two field experiments. GiftAMeal: 6,364 customers emailed across control, sender-benefiting, recipient-benefiting, shared (two-sided) and donation conditions. Conversion was 0.94% for recipient-benefiting and 0.94% for shared (p = .99), against 0.39% sender-benefiting and 0.08% control. Game Access: 1,500 assigned, 1,438 emails delivered; conversion 4.77% recipient-benefiting against 1.71% sender-benefiting and 0.61% control. The paper states that the shared incentive “did not outperform the recipient-benefiting incentive.” Harvard Business School research on promote a referral program
Gershon & Jiang, “Referral Contagion: Downstream Benefits of Customer Referrals,” Journal of Marketing Research. Mobile cashback firm, 41.2m registered customers 2012 to 2022, with a 500,000-consumer regression sample, plus a 20-company dataset and a field experiment. Referred customers themselves referred substantially more: invitation rates of 8.72% against 2.51%, and 31% to 57% more referrals, with 19 of 20 companies showing the pattern. The authors state the field comparisons are not causal and that referral status is not randomized, so treat this as a targeting signal rather than a proven effect. Peer-reviewed study (SAGE) on promote a referral program
Van den Bulte, Bayer, Skiera & Schmitt (2018), “How Customer Referral Programs Turn Social Capital into Economic Capital,” Journal of Marketing Research. Same bank, now at dyad level: 1,799 referrer-to-referral dyads (3,598 customers) against 3,663 non-referred. Referred customers churned about 30% less after controls (9.7% against 14.5% by the end of the window), but the paper finds they “exhibit lower churn only as long as their referrer has not churned.” Churn was about 40% lower while the referrer remained a customer and markedly higher once the referrer left. Referrers with under one month of tenure produced referrals whose margin was below that of non-referred customers, which is why the authors advise recruiting referrers “among their customers who have been acquired at least six months ago.” Peer-reviewed study (SAGE) on promote a referral program
Schmitt, Skiera & Van den Bulte (2011), “Referral Programs and Customer Value,” Journal of Marketing 75(1):46-59. One large German retail bank, anonymized, paying referrers a EUR 25 voucher. 5,181 referred and 4,633 non-referred customers acquired in 2006, analyzed sample 9,495 after removing influence points; observation window January 2006 to September 2008 (33 months). Referred customers were worth about EUR 40 (16%) more on a six-year lifetime value basis and EUR 49 (about 25%) more within the window; acquisition cost was about EUR 20 lower; churn hazard was about 18% lower, with 82.0% of referred against 79.2% of non-referred still active at 33 months. The margin advantage shrinks by 23.1 cents per 1,000 days and the paper states that “the expected contribution margin advantage of a referred customer disappears after 857 days,” while on retention “there is no such erosion.” The abstract adds that “the size of the value differential varies across customer segments; therefore, firms should use a selective approach for their referral programs.” Wharton research on promote a referral program
ePrivacy Directive 2002/58/EC, Article 13. Article 13(1): electronic mail for direct marketing “may only be allowed in respect of subscribers who have given their prior consent.” Article 13(2)’s soft opt-in applies only where the contact details were obtained from that customer “in the context of the sale of a product or a service,” which is never true of a referred friend. EU ePrivacy Directive
UK ICO, electronic mail marketing guidance under PECR. You must comply “if you send a marketing message, or if you ‘instigate’ someone else to send it,” and the guidance is explicit that you are only responsible where “you must have encouraged them to send it.” The soft opt-in for previous customers “does not apply to prospective customers or new contacts,” which by construction is the referred friend. On asking customers for their friends’ contact details the ICO states: “we would therefore advise against this type of viral marketing.” UK ICO guide to email marketing (PECR)
GDPR Article 14 (information where data are not obtained from the data subject). When you receive a friend’s email or phone number from a referrer you are the controller for that data. Article 14(2)(f) requires you to tell the friend “from which source the personal data originate,” and Article 14(3)(a) requires it “within a reasonable period after obtaining the personal data, but at the latest within one month.” GDPR Article 14
Jensen v. Capital One Financial Corp., 2025 WL 606194 (W.D. Wash., February 25, 2025). A putative class action over a “Refer a Friend” text where the company generated the referral link and a pre-filled message and directed the customer to send it. The court held that Section 230 did not immunize the company, finding it “the sole author of the content of the text” and that it “substantially assisted its customers in formulating, composing, and sending commercial text messages.” An in-app notice asking the customer to confirm they had the recipient’s consent did not change the outcome. Cited because it is the exact pre-filled, one-tap share design that referral advice usually recommends.
FTC Rule on the Use of Consumer Reviews and Testimonials, 16 CFR Part 465 (89 FR 68034, effective October 21, 2024). This is a rule, so violations carry civil penalties of up to $53,088 each under FTC Act §5(m)(1)(A) (16 CFR 1.98(d), for penalties assessed after January 17, 2025). §465.4 prohibits providing compensation or other incentives conditioned, expressly or by implication, on a review expressing a particular sentiment. §465.1 defines the compensation broadly enough to cover discounts, coupons and store credit. Relevant here because a program that routes a referral reward to customers selected for leaving a positive review moves from Part 255 into Part 465. FTC rule on consumer reviews and testimonials
Shopify, “Referral Marketing Statistics.” Vendor guidance, cited only for the figures it carries and attributes, each of which lacks a stated sample: referral traffic converting at 2.5% to 3.5% against 0.5% to 1% for paid social (Elogic Commerce); a “healthy referral rate” of 1% to 3% (unattributed); and merchant anecdotes from the Shopify Masters podcast. It does not address friend-side rewards, referral fraud, FTC disclosure or program promotion. Shopify referral marketing statistics