Referral Programmes
Date: 2026-08-16
Paying customers to bring customers. It produces the best-retaining customers you can buy, because referred people arrive with a recommendation attached — and it attracts fraud in direct proportion to how well the incentive works.
What it is
A referral programme rewards existing customers for introducing new ones, usually with a two-sided incentive — something for the referrer, something for the referred. It’s the most common growth loop in ecommerce.
Why referred customers are worth more
Consistently the best-performing acquisition source on lifetime value, for two reasons:
- Selection. People refer others they think will like the product, so referred customers are better matched than any targeting algorithm achieves
- Trust transfer. A recommendation from someone known carries weight no advertising does
The practical consequence: referred customers typically retain better and repeat more than paid-acquired ones, which means their higher first-order cost is often recovered — Customer Lifetime Value, Channel Mix.
The economics
Two-sided incentives on the running model:
referrer reward £10 credit
referred reward £10 off first order
cost per referred customer £20.00
compare paid social CAC £42.00
compare blended CAC £22.50
Cheaper than paid, and the customers are better. But three adjustments before believing it:
The referrer reward is only a cost when redeemed. Credit issued and never used costs nothing, so the effective cost is below the nominal.
The referred discount raises CAC and lengthens payback, and it’s paid to referred customers who’d have bought anyway — the same trap as any first-order discount — Payback Period.
Some referrals would have happened anyway. Word of mouth exists without a programme. The incentive captures and accelerates it, and it also pays for it — Incrementality Testing.
The fraud surface
Proportional to how well it works, and it needs designing for rather than reacting to:
- Self-referral — a second email address, a different card
- Referral rings — groups referring each other in circles
- Incentive farming — accounts created solely to harvest rewards, often at scale
- Code leakage to aggregator sites, converting a personal referral into a public discount — the same failure as Affiliate and Partnerships
Defences that work without ruining the experience:
- Reward on the referred customer’s first order, not on signup. Removes most of the incentive to farm
- Delay the referrer reward until the return window closes, so a refunded order doesn’t pay out
- Cap referrals per account
- Match on address and payment method, not just email
- Cap total reward per referrer to make ring behaviour unprofitable
Design that matters
- Reward both sides. One-sided programmes underperform — the referrer needs a reason to act and the referred needs a reason to accept
- Ask at the right moment. Immediately after a good experience — order delivered, positive review, second purchase — not on the confirmation page before they’ve received anything
- Make sharing trivial. One tap, pre-written message, works on mobile
- Non-monetary rewards work where the product has status or scarcity, and cost far less than credit
Measuring it
- Cost per referred customer, net of unredeemed rewards
- Referred customer LTV against other channels, from cohorts. This is where the programme justifies itself — Cohort Revenue
- Participation rate — what share of customers refer at all. Usually small, and concentrated in your best customers
- Fraud rate as a standing metric, not a periodic panic
- Incrementality, by holding the programme back from a random slice. It’s the one number that says whether you’re paying for word of mouth you already had
In plain terms: referral is the best acquisition channel most retailers under-invest in, and the one most likely to be quietly abused. Design the fraud controls first and the incentive second.