Do Referral Programs Work: What Happens to a Referred Customer When the Referrer Leaves
Do referral programs work? Almost every page answering that question is published by a company selling referral program software, and almost every one of them stops at the same place: referred customers are better, so build a programme.
That is true. It is also about a fifth of what the research found, and the missing parts are the ones that decide whether the programme pays for itself.
Three studies in peer-reviewed marketing journals have tracked this properly — not with a survey asking executives how they feel about referrals, but by following individual customers for years and comparing what they were actually worth. What they found is more specific, more useful, and considerably less flattering than the version you get from a vendor.
The Data Behind the Answer
Two of the three studies share a dataset, and it is worth understanding what it is.
A German bank ran a referral programme paying roughly €25 per successful referral. Researchers obtained its customer records and followed everyone acquired in 2006 through to September 2008 — 33 months of individual-level data.
The first study, Schmitt, Skiera and Van den Bulte in the Journal of Marketing, compared 5,181 referred customers against 4,633 non-referred ones, with 9,495 customers in the analysis sample after removing outliers.
The second, Van den Bulte, Bayer, Skiera and Schmitt in the Journal of Marketing Research, went back to the same bank and asked a different question: not whether referrals are better, but why. It analysed 1,799 referral–referrer pairs against 3,663 randomly drawn and 1,788 demographically matched non-referred customers.
The third is unrelated — Villanueva, Yoo and Hanssens, who tracked an internet company’s customers over 70 weeks.
Two points of honesty before the numbers. The first two studies use the same bank, so they are not independent confirmations of each other; they are one dataset examined twice. And a German retail bank is not every business. Treat the direction as well-established and the exact figures as specific to that setting.
Do Referral Programs Work: The Numbers That Say Yes
On the basic question, the evidence is consistent.
| What was measured | Referred | Non-referred |
|---|---|---|
| Daily contribution margin | €0.65 | €0.54 (€0.48 matched) |
| Churn by end of the window | 9.7% | 14.5% (13.4% matched) |
| Survival at 33 months | 82.0% | 79.2% |
| Six-year customer lifetime value | ~16% higher | baseline of €253 |
Including the roughly €20 lower acquisition cost per referred customer, the lifetime value gap rises to about 25%. Measured over the observed 33 months rather than projected, referred customers were worth about €49 more — roughly 25%, or nearly 35% once acquisition savings are counted.
The paper’s own summary sentence: “The average value of a referred customer is at least 16% higher than that of a nonreferred customer.”
So the headline holds. Now the parts nobody quotes.
The Margin Advantage Expires. The Retention Advantage Doesn’t.
This is the single most important finding for anyone deciding whether a referral bonus makes financial sense, and it almost never appears outside the papers.
The margin gap decays. Referred customers start out contributing meaningfully more per day, and that difference shrinks steadily. The 2011 study estimated the advantage disappearing after roughly 857 days. The 2018 study, using tighter comparison groups, put it between roughly 730 days and 1,350 days depending on which non-referred customers you compare against.
Two to four years, then gone.
The churn gap does not decay. Referred customers keep leaving at a lower rate, and that advantage persists across the whole observation window.
These two facts pull in opposite directions, and which one dominates decides your economics. If your business makes its money on margin per transaction, the referral premium has a shelf life and your payback calculation needs to respect it. If your business makes its money on customer duration — subscriptions, retainers, recurring services — the advantage is the durable kind.
That distinction matters more than the headline percentage, and it is the sort of thing a pricing decision has to account for explicitly rather than assume.
Why Referred Customers Are Better: Two Different Mechanisms
The 2018 study is the one that explains the machinery, and it found that the margin advantage and the retention advantage come from entirely different causes.
Better matching produces the margin. A referrer who already uses the product knows what it is for and tends to bring people it genuinely suits. The referred customer is a better fit from day one, so they use more and contribute more. This is selection, not persuasion.
Social enrichment produces the retention. The referred customer has a relationship inside the product — someone they know who also uses it. That connection is what keeps them, and it has nothing to do with how well the product fits them. It is the same mechanism that governs how work reaches people through the connections they already have, measured from the business’s side of the transaction rather than the individual’s.
Once you separate the two, the practical advice splits as well. If you want margin, you want referrers who understand the product. If you want retention, you want the social tie to survive.
Which leads to the finding that reframes the whole thing.
What Happens When the Referrer Leaves
The 2018 paper measured churn for referred customers depending on whether the person who referred them was still a customer.
Referrals whose referrer is still around churn at roughly 40% lower hazard than non-referred customers.
Referrals whose referrer has left churn at roughly 280% higher hazard.
Read that twice. A referred customer is not permanently better. They are better while the relationship that brought them in still exists. When the referrer leaves, the referred customer does not revert to average — they become substantially worsethan someone who arrived through advertising.
The authors’ explanation is that the churn advantage was never a property of the referred customer. It was a property of the connection. Remove the connection and you remove the thing that was doing the work, and what remains is a customer whose anchor just walked out.
The operational consequence is unusual and almost nobody acts on it: retaining your referrers is also a retention strategy for the customers they brought. Losing one good referrer can cost you several customers downstream, and standard churn reporting will never show you that, because it treats each account as independent.
Not Every Customer Should Be Asked to Refer
The same study looked at who makes a good referrer, and the answer is specific enough to act on.
| Referrer tenure when they referred | Effect on the referral’s margin |
|---|---|
| Less than 1 month | Margins lower than a non-referred customer |
| 1 to 6 months | About 25% of the gap produced by experienced referrers |
| More than 6 months | The full effect |
A customer who refers someone in their first month produces a referral that is worse than a stranger who found you through an ad. They do not know the product well enough yet to match it to the right person, so they bring the wrong people.
The authors’ recommendation is direct: recruit referrers who have been customers at least six months, have high margins, and are unlikely to churn.
Which is the opposite of how most referral programmes are built. The standard design fires an invitation at the moment of purchase, when enthusiasm is highest and product knowledge is lowest. The research says that is precisely the wrong moment — and that the excited new customer is the one whose referrals will underperform.
One more detail: referral margins rose with the referrer’s own margin and with their age. But no referrer characteristic predicted referral churn at all. You can select for a profitable referral. You cannot select for a loyal one — that depends entirely on whether the referrer stays.
Where the Advantage Disappears Completely
The 2011 study checked whether the referral premium held across customer segments. It did not.
The value advantage held for retail and non-retail customers and for high-margin customers. It was absent for low-margin customers and absent for customers over 55.
That is a limit no referral software vendor will mention, and it has an obvious implication: a programme aimed at your cheapest customers may produce referrals carrying no premium at all, while still costing you the reward you paid for each one.
Word of Mouth Against Paid Acquisition
The third study separates customers by how they arrived and tracks what each cohort is worth over time.
Customers acquired through marketing contribute more in the short run. Customers acquired through word of mouth contribute nearly twice as much long-term value.
The timing is the interesting part. The effect of marketing-driven acquisition settles after about three weeks. Word-of-mouth impact runs roughly six weeks, and by week 10 its cumulative impact is almost double.
Paid acquisition buys a spike. Word of mouth buys a slope. Anyone deciding where a limited budget goes is choosing between those two shapes, not between two versions of the same thing — a distinction that matters most for the kind of small operation where the budget is genuinely limited.
What This Changes About How You Run a Referral Programme
Five things follow from the research rather than from opinion.
Ask at six months, not at checkout. Referrals from customers in their first month underperform non-referred customers outright. The enthusiasm at purchase is not the same thing as the product knowledge that makes a referral valuable.
Pick who you ask. High-margin, long-tenured, low-churn-risk customers produce measurably better referrals. A programme open to everyone is a programme buying a lot of referrals that carry no premium.
Keep your referrers. Their departure is not just one lost customer — it is a sharp increase in churn risk for everyone they brought. This is the single most actionable finding here and the one most likely to be invisible in your existing reporting.
Match the programme to how you earn. Margin-driven businesses are buying an advantage that expires in two to four years. Duration-driven businesses are buying one that persists. Build the payback calculation accordingly.
Don’t aim it at your cheapest segment. The premium was absent for low-margin customers. The reward is not.
None of this requires software. It requires knowing which customers to ask and when — which is the same problem as finding the first ten buyers for anything, approached from the other end. That matters because the business this research describes is not the typical one: the SBA’s small business figures put average survival across cohorts from 1994 to 2022 at 67.7% after two years and 49.2% after five. Most firms reading this will never run a referral programme at the scale a German bank does. They will ask ten people, and which ten is the entire decision.
What the Research Does Not Settle
Four honest limits.
Two of the three studies are one bank. The 2011 and 2018 papers analyse the same German retail bank. They are a deep look at one case, not two independent confirmations.
Reward size was never tested. The programme paid about €25 per referral with no conditions beyond opening an account. Whether a larger reward produces worse matches is raised by the authors as an open question, not answered. Anyone promising you an optimal reward amount is not citing this research.
The data is from one period. Customers acquired in 2006, observed to 2008. The mechanisms — matching and social enrichment — are structural and unlikely to have reversed. The specific euro figures are not current prices.
This is association in observational data, with controls. The researchers matched on demographics and controlled extensively, but nobody randomly assigned customers to be referred. The mechanism findings are well-argued, not experimentally proven.
Frequently Asked Questions
Do referral programs work, in plain terms? Yes, measurably. Referred customers in the bank study had around 16% higher six-year lifetime value, rising to roughly 25% once their lower acquisition cost is counted, with about 18% lower churn. The complication is that the margin half of that advantage decays over roughly two to four years while the retention half does not.
How much is a referred customer actually worth? In the studied bank: €0.65 in daily contribution margin against €0.54 for non-referred customers, churn of 9.7% against 14.5%, and about €20 less to acquire. Measured over 33 months, roughly €49 more per customer.
When is the best time to ask a customer for a referral? After about six months. Referrals made by customers with under a month’s experience produced margins lower than non-referred customers, and those from one-to-six-month customers produced only about a quarter of the gap that experienced referrers did.
What happens if the person who made the referral cancels? The referred customer’s churn risk rises sharply — roughly 280% higher hazard than a non-referred customer, against roughly 40% lower while the referrer is still active. The retention advantage belongs to the relationship, not to the customer.
Are referrals better than paid advertising? They work on different timescales. Marketing-acquired customers contribute more in the short run, with the acquisition effect settling in about three weeks. Word-of-mouth customers add nearly twice as much long-term value, with cumulative impact almost double by week 10.
Does a referral programme work for every type of customer? No. In the bank data the value advantage was absent for low-margin customers and for customers over 55, while holding for high-margin, retail and non-retail customers.
How big should the referral reward be? The research does not answer this. The programme studied paid about €25, and the authors explicitly list the effect of larger fees as an open question. Treat specific reward recommendations elsewhere as opinion.
Do I need referral software to do this? Nothing in the findings depends on tooling. What the research supports is choosing who to ask, asking after about six months, and tracking whether your referrers stay. That is a list and a calendar reminder before it is a product.
Conclusion
Do referral programs work? Yes — and the useful version of that answer is almost entirely in the detail that the one-line version throws away.
A referred customer is worth meaningfully more, but for two different reasons that behave differently over time. The margin advantage comes from better matching and expires within a few years. The retention advantage comes from a social connection, persists while that connection exists, and inverts when it breaks.
The practical shape that produces is not the one most programmes are built in. Ask later rather than at the moment of purchase. Ask fewer people rather than everyone. Watch your referrers as carefully as you watch the customers they brought, because the research says those two numbers are not independent.
And treat the version of this that ends at “referrals are better, buy our software” as what it is — the first sentence of a longer finding, quoted by people who benefit from you not reading the rest.
A Note from Lucas
I expected this to be a short piece confirming something obvious, and the 280% figure stopped me — a referred customer whose referrer leaves becomes worse than one who came from an ad, which is the exact opposite of how I had been thinking about referrals. What I keep turning over is that nobody reports it, because churn dashboards treat every account as independent and the link between two customers is invisible unless you deliberately look for it. The other finding I would argue with anyone about is the six-month one: every referral programme I have ever been asked to join fired its invitation seconds after I paid, which the research says is the single worst moment to ask. If you take one thing from this, make it that asking fewer customers, later, beats asking everyone immediately.
About the Author
Lucas Palma is the founder and editor of Info Is The Key. He writes about online business, entrepreneurship, and the practical mechanics of building income independently, working from primary sources and official data rather than recycled advice.
Last Updated: October 2026 | Reading Time: 11 minutes
Editorial note: This article summarises three peer-reviewed marketing studies accessed in October 2026. Two of them analyse the same German retail bank dataset and are not independent replications. All figures are observational with statistical controls, not randomised experiments, and the monetary values are from 2006–2008 data in euros. Findings from one bank and one internet company may not generalise to other business types, markets or periods. Nothing here is financial or business advice.
Sources
- Schmitt, P., Skiera, B., & Van den Bulte, C. (2011). Referral Programs and Customer Value. Journal of Marketing, 75(1), 46–59
- Van den Bulte, C., Bayer, E., Skiera, B., & Schmitt, P. (2018). How Customer Referral Programs Turn Social Capital into Economic Capital. Journal of Marketing Research, 55(1), 132–146
- IESE Business School — Villanueva, J., Yoo, S., & Hanssens, D. M., on marketing-induced versus word-of-mouth customer acquisition, Journal of Marketing Research
- U.S. Small Business Administration, Office of Advocacy — Frequently Asked Questions About Small Business, 2026
One Comment