How to calculate cost per paying customer
Cost per paying customer is the ad spend of a campaign divided by the number of customers that campaign brought who paid and were not refunded. It differs from cost per conversion because the platform's conversion is usually a sign-up or a lead, not a payment.
The formula is simple. The hard part is counting the customers honestly.
What is the formula?
Cost per paying customer = campaign spend in a period ÷ paying customers matched to that campaign in the same period.
Use the same period on both sides, for example the last 28 days. Count a customer once, for one campaign, and only if the payment was not refunded within your refund window. Datarithms uses the last ad click within 90 days and a 14-day refund window.
What does a worked example look like?
| Campaign | Spend | Paying customers | Refunded within 14 days | Counted | Cost per paying customer |
|---|---|---|---|---|---|
| Search, brand terms | $1,240 | 33 | 2 | 31 | $40 |
| Reels, free trial offer | $2,910 | 36 | 2 | 34 | $86 |
| Lookalike, sign-ups | $5,570 | 52 | 4 | 48 | $116 |
$1,240 ÷ 31 = $40. $2,910 ÷ 34 = $85.59, shown as $86. $5,570 ÷ 48 = $116.04, shown as $116.
Which mistakes inflate or deflate the number?
- Counting sign-ups or trials as customers. The cost looks lower than it is.
- Counting refunded customers. The cost looks lower for exactly the campaigns that attract bad fits.
- Adding up every platform's own conversions. Several platforms claim the same customer.
- Mixing periods: spend from this month, customers from last month.
- Ranking campaigns with only a few customers. See how many customers you need.
What about customers you cannot match to a campaign?
Keep them out of every campaign's count and report them as their own line. Spreading them across campaigns in proportion makes every number look better and none of them true.
Common questions
Is cost per paying customer the same as CAC?
It is the paid-ads part of CAC for one campaign. Customer acquisition cost usually also includes salaries, tools and other channels.
Which period should I use?
Use one long enough to collect at least 30 paying customers for the campaigns you want to compare, and the same period for spend and customers.
Related pages
- How many customers you need before judging a campaignWhy Datarithms gives no verdict below 30 paying customers, with an example.
- How to match Stripe payments to ad clicks with exportsStep by step: join ad exports, sign-up records and Stripe payments by hand.
- GlossaryPlain definitions of the ad measurement terms used on this site: paying customer, CAC, attribution window, last click, view-through, ROAS and more.
- How it worksThe matching rules behind the weekly table: last ad click within 90 days, refunds within 14 days excluded, no verdict under 30 paying customers.
Want your own table?
Tell us which ad platforms you use and roughly what you spend. We reply by email within two working days.
Early access: we prepare your weekly table with you from your own exports while the direct connections are built.