Add media, people, agency and tools, divide by new customers won in the same period. The page also shows the media-only figure and the gap between them.
Both terms have to cover the same period and the same people. Costs from January divided by customers won across January and February produces a number that means nothing, and it is the most common way this calculation goes wrong.
The two get used interchangeably in meetings and they measure different things. Cost per acquisition divides media spend by conversions. It answers a channel question: is this campaign buying results at a sensible price. Customer acquisition cost divides everything it took to win the customer by the customers won. It answers a business question: can we afford to grow this way.
In the worked example above the two are $22.99 and $66.21. Same month, same 435 customers, same company. The difference is $43.22 per customer, and it is entirely made of costs that are real, recurring, and invisible in the ad platform.
Use CPA to decide which campaign gets the next thousand dollars. Use CAC to decide whether the next thousand dollars should be spent at all.
Always in. Media spend across every paid channel. Salaries and payroll costs for everyone whose job is acquiring customers, including the fraction of a founder’s time that goes to it. Agency retainers and freelancer invoices. Software that exists to acquire customers: the ad management layer, landing page builder, analytics, call tracking.
Usually in. Creative production, whether that is a photographer, an editor or a stock subscription. Sales commission on new business, though not on renewals. Discount codes and first-order incentives, which are an acquisition cost dressed as a price cut.
Out. Costs of serving customers you already have: support, success, retention campaigns, and the software those run on. Rent and general overheads. Product development. These belong in other calculations and putting them here inflates CAC until it stops being comparable to anything.
The boundary that causes the most argument is a marketer who does both acquisition and retention. Split their cost by the share of their time, estimate it honestly, and write the assumption down so the next person can see what you did.
Blended CAC divides total acquisition cost by every new customer, including the ones who arrived through organic search, referral or word of mouth. It is a legitimate business metric. It is also the easiest number on the dashboard to accidentally flatter yourself with.
| Organic customers added | Total new customers | Blended CAC |
|---|---|---|
| 0 | 435 | $66.21 |
| 50 | 485 | $59.38 |
| 100 | 535 | $53.83 |
| 200 | 635 | $45.35 |
| 300 | 735 | $39.18 |
| 400 | 835 | $34.49 |
This is why the paid figure and the blended figure both belong on the report. The blended number tells you what the business pays for growth. The paid number tells you whether the paid channel is still working. When they diverge, the paid channel is usually the one that moved.
A CAC on its own is neither good nor bad. It only becomes one of those next to what the customer returns and how long they take to return it.
The ratio most often quoted is three to one, lifetime value against acquisition cost. It is a convention from venture-backed software, not a law, and applying it to a business with different economics produces confident nonsense. A grocery delivery service with monthly repeat purchases and a hardware manufacturer selling once every seven years cannot share a target.
The more useful companion is time. A CAC of $66 against a lifetime gross profit of $200 is a 3:1 ratio and looks healthy, but if the $200 arrives over four years while the $66 leaves your account this month, growth consumes cash faster than it produces it. That is a solvent business going bankrupt, and the ratio never warned anyone.
Check the payback period alongside the ratio. And use lifetime gross profit rather than lifetime revenue in the numerator, or you are comparing what the customer pays you against what the customer costs you, which is not a comparison at all.