The CAC number most brands get wrong

Platform-reported CAC, blended CAC and new-customer CAC are three different numbers. Confusing them is how budgets get misallocated for months.

Ask five people at a growth meeting what CAC is and you will get five numbers. That is not pedantry — it is the reason budgets get pointed at the wrong channel for a quarter at a time.

There are three numbers worth naming clearly.

Platform CAC

What Meta, Google or TikTok says it cost to acquire a conversion, using its own attribution window. This is useful for one thing only: optimising inside that platform. It is not comparable across platforms, because each one credits itself generously and the windows differ.

Sum the revenue claimed by every platform you run and compare it to your actual revenue. The overlap is often 20 to 40 percent. Nobody is lying; they are all counting the same conversion.

Blended CAC

Total marketing spend divided by total new customers, over the same period. Crude, honest, and difficult to argue with, because it reconciles with the bank account.

Blended CAC is the number to run the business on. It will not tell you which channel to cut, but it will tell you immediately whether the machine as a whole is working. When platform CAC improves and blended CAC does not, you have found reallocation dressed up as growth.

New-customer CAC

Spend divided by first-time customers only. This is the one most brands never calculate and most need. If 40 percent of your orders are repeat purchases from an owned-channel audience, a healthy-looking blended CAC can hide the fact that acquisition has stopped working entirely.

Most commerce platforms can tag first-time versus returning orders. Do that, then track new-customer CAC weekly. It is the earliest warning system you have.

The margin question nobody asks

All three numbers are ratios of spend to customers, and none of them knows what a customer is worth. A $60 CAC is excellent on a $400 first order at 60 percent margin and terrible on a $70 first order at 30 percent. Contribution margin after CAC, per new customer, is the number that decides whether you can spend more.

If you can only track one number, track contribution margin per new customer. It is the only one that tells you whether scaling makes you money.

How to test what is real

Attribution modelling can only redistribute the credit you already have. To learn what a channel genuinely adds, you have to remove it and watch:

• Geo holdouts. Turn a channel off in matched regions for two to four weeks and measure the difference in total orders.
• Scale-back tests. Cut spend 50 percent in one region and compare against a control.
• Post-purchase surveys. One question at checkout — "where did you first hear about us?" — is imprecise per response and remarkably useful in aggregate.

These tests are inconvenient, and they occasionally produce answers people do not want. That is exactly what makes them worth running.

A workable reporting stack

If you want help building this, it is what our measurement engagements deliver. Use platform CAC for in-platform decisions. Use blended and new-customer CAC for budget decisions. Run an incrementality test on your two largest channels twice a year. Report contribution margin per new customer to the people who decide how much you get to spend.

Four numbers, clearly defined, beats one number everyone quietly interprets differently.