Customer acquisition cost
Every ad platform reports a version of this, and the versions sum to more customers than exist.
Also called CAC, Cost per acquisition, Blended CAC.
What it is
Customer acquisition cost is total acquisition spend divided by the number of new customers that spend brought in, over the same period. It exists to answer one question: what does the next customer cost to win. Both halves are contested — the numerator because nobody agrees what counts as acquisition spend, and the denominator because the systems reporting it are counting orders, and counting some of them twice.
acquisition spend ÷ new customers acquired
- acquisition spend
- Media, plus the people, agencies and tools that ran it, plus any discount given specifically to win a first order
- new customers acquired
- Distinct customers whose first-ever order falls in the period — not orders, and not conversions as counted by a platform
The part the formula leaves out
The denominator is not orders. Spend over orders and spend over customers are different questions, and the gap between them is the repeat business you already had. In the worked project below, 4,318 orders trace back to an advertising click but only 3,412 of them are somebody's first order. Dividing by the larger number produces $92.26 and dividing by the smaller produces $116.76 — a 27% difference, in the direction that makes the channel look affordable. If a business is healthy, its repeat rate rises and this error grows every quarter.
The numerator is not media spend either. The conventional figure is what was paid to the platforms, because that is the number sitting in an export. The costs that do not appear there are the salary of whoever ran the campaigns, the agency retainer, the tooling, and the discount code that existed only to convert a first order. A business that runs acquisition with two full-time people and a 15% welcome code can be reporting a CAC that is half its real one, and the half it is missing is the half that does not scale down when spend does.
Then attribution, which is where the arithmetic stops being arithmetic. Each platform counts a conversion inside its own window, so a customer who saw two ads is claimed by both, and the totals add up to more customers than the business has. There is no correction that recovers the truth from the exports alone, because none of them records what the customer actually saw. What is recoverable is the reconciled total: distinct customers traced to any click. That is a real number and each platform's own claim is not, which is why a blended CAC computed against the order ledger is worth more than three precise-looking channel figures that overlap.
How it is usually computed wrongly
01
Adding up the platforms' own cost-per-acquisition figures
They share a numerator and double the denominator. Each platform divides its spend by conversions it claims, and the same customer is claimed by more than one, so the combined figure describes a business with more customers than this one has.
Compute one blended figure: total spend over distinct customers traced to any click, joined against your own order ledger. Report the per-platform claims separately and label them as claims.
02
Dividing by orders rather than by first-time customers
Repeat orders cost nothing to acquire. Including them inflates the denominator and understates CAC, and the error grows exactly as retention improves — so the metric gets more wrong the better the business does.
Filter to customers whose first-ever order falls in the period, which needs a customer identifier stable across orders. If you only have order-level data, say the figure is cost per order and do not call it CAC.
03
Counting media spend only
Salaries, agency fees, tooling and first-order discounts are all costs of acquiring the customer, and all of them sit outside the ad platform's export. Leaving them out is not conservative — it produces a CAC that a payback calculation cannot survive contact with.
State the numerator's contents on the figure. A media-only CAC is a legitimate number if it is labelled as one; the failure is quoting it against a payback target computed on fully loaded costs.
04
Comparing CAC to average order value
An order's revenue is not what it is worth to you. Set against AOV, a channel selling heavily discounted, heavily returned product looks affordable right up to the point the returns land.
Compare CAC to contribution per customer over a stated horizon, net of returns, fees and cost of goods. In the worked project the cheapest channel on claimed conversions turns out to lose money once returns are counted.
What your file needs
- Ad spend by channel and month, reconciling to each platform's own invoice total
- Orders with a customer identifier stable enough to distinguish a first order from a repeat
- A click id, first-touch reference or landing page on the order — or values and timestamps close enough to join on
Anything missing is reported as unavailable rather than substituted with something weaker computed on worse evidence.
Compute it on your own file
No account needed to start. You only pay when you like what you see.