Referral marketing vs paid ads, compared like for like
Compare referral and paid ads fairly: same customers, same window, all costs, each sale counted once, and what is genuinely extra.
Someone shows you cost per sale for referral and cost per sale for paid ads. Before you act on the gap, check whether the two numbers describe comparable customers, the same period and the same cost categories.
All figures are invented to show the arithmetic. They are not benchmarks, and they do not show which channel is cheaper for you.
Comparable customers, same window, same costs
A fair comparison starts with three matches:
- Comparable customer groups. Use clear cohort definitions for eligible customers in each channel, not one channel's best segment against the other's average.
- The same time window. Count customers and costs over the same period.
- The same cost categories. Include everything each channel costs: ad spend, agency or platform fees, referral rewards, tool fees and the time spent running each.
A note on terms. Searches for this topic return many pages about affiliate and cost-per-action marketing. Those are related but different arrangements. This guide is about customer referral, where existing customers are rewarded for introducing new ones.
Comparing costs does not prove referral is cheaper or paid is better. It tells you what you are comparing.
Cost per customer versus cost per extra customer
Cost per customer divides spend by every customer a channel gets credit for. Cost per extra customer divides spend by only those you would not have won without the channel. The second is harder to know and often matters more.
Worked example (invented assumptions):
| Paid ads | Referral | |
|---|---|---|
| Total channel cost | £5,000 | £800 |
| Customers credited | 100 | 40 |
| Cost per customer | £50 | £20 |
| Assumed customers who would have bought anyway | 30 | 25 |
| Extra customers | 70 | 15 |
| Cost per extra customer | about £71 | about £53 |
For this calculation alone, assume these are the total comparable channel costs and every omitted charge is zero. A real comparison must include fees, administration and other relevant costs on both sides.
The gap narrows once the question "would they have bought anyway?" is asked of both channels. The figures of 30 and 25 customers are assumptions, and you cannot read them off a dashboard. This is scenario arithmetic, not measured incrementality. It shows why the question matters, not what the answer is for you.
Counting each sale once
If each channel reports a conversion for the same buyer, the totals can exceed your real customer count. That is an accounting problem, separate from the question of what is extra.
Under the assumptions here, suppose you won 100 distinct new customers, each counted once. If the ad platform reports 70 and the referral tool reports 40, the reports add to 110, so at least 10 customers are being claimed by both.
Two practical points:
- An optional "How did you hear about us?" question can help you reconcile claims. Even optional, it may add some friction, so keep it proportionate, and remember that self-reported answers are imperfect.
- A rule such as "a referral code beats an ad click" is a reporting policy. It keeps your totals tidy. It is not proof about what caused the sale.
Ways to estimate what is extra
Several designs can estimate what a channel adds. A controlled holdout, where a comparable group does not receive the channel, or a randomised test, is one possible method, and our guide to measuring a rewards-based campaign explains holdout groups. Other designs exist.
A simple pause-and-compare, where you switch a channel off and watch what happens, can be confounded by season, promotions, price changes and everything else that moves at once. Report a range rather than a single number, and record the method you used so the next comparison is consistent.
None of these methods is declared best here, and none was tested for this article. Reliable incrementality measurement may need a separate study.
Comparison checklist
This checklist supports an initial comparison using a consistent period of data. It does not replace a separate incrementality study.
For the wider cost picture, see the acquisition economics playbook and the paid-growth saturation test.
Where Zenko fits
To ask about a campaign, contact Zenko or see the for marketers page. Availability and terms are confirmed for each campaign.
Related referral guides
- How to set up a referral programme one decision at a time
- Why referral programmes fail and where to look first
Frequently asked questions
Is referral marketing cheaper than paid ads?
It depends on your business, and comparing costs alone cannot settle it. Compare comparable customers over the same window with all costs included, then ask how many of each channel's customers were genuinely extra.
Is referral the same as affiliate marketing?
No. Customer referral rewards existing customers for introducing new ones. Affiliate and cost-per-action marketing usually involve external partners paid per action. They can be compared, but they should not be treated as one thing.
How do I stop two channels claiming the same sale?
Count each customer once in your order system, and agree a written reporting rule for overlaps. An optional "How did you hear about us?" question can help, though it may add some friction. This tidies the accounting but does not show what caused the sale.
Sources
- Zenko, How to measure a rewards-based campaign: https://www.zenkoprotocol.com/insights/measure-rewards-based-campaign/
- Zenko, The acquisition economics playbook: https://www.zenkoprotocol.com/insights/acquisition-economics-playbook/
- Zenko, The paid-growth saturation test: https://www.zenkoprotocol.com/insights/paid-growth-saturation-test/
General information. Worked examples use invented figures and are not measured results.