What is CAC?

A quarter's numbers: A$120,000 of ad spend, A$18,000 in agency fees, A$12,000 in creative production, A$6,000 in tooling and A$84,000 in sales salaries, against 400 new customers. Ad spend alone gives A$300, which is what the ad platform reports. The full cost is A$240,000, so CAC is 240,000 ÷ 400 = A$600 — double the figure on the dashboard, and the one the business actually pays.

Customer acquisition cost divides everything spent on winning customers by the number of customers won. It is the figure the business pays, as distinct from the figure the ad platform reports, and in most companies there is a large gap between the two.

The ad platform sees media spend and conversions it believes it caused. It does not see the agency retainer, the freelance editor who cut the videos, the subscription tools, the salaries of the people who made the campaigns, or the salespeople who called the leads. In a business where a human closes the sale, those items routinely exceed the media spend. A dashboard reporting a A$300 cost per acquisition can sit inside a business with a A$600 customer acquisition cost, and only the second number can be compared with what a customer is worth.

This is also why the metric is contested inside companies in a way that cost per acquisition is not. Cost per acquisition is a measurement; customer acquisition cost is partly a definition, and the definition is chosen.

There is no universal standard, which means the standard has to be written down locally and applied consistently. A defensible default:

In: media spend; agency, contractor and freelance fees; creative production; marketing and sales salaries with on-costs; the software used to run acquisition; discounts, credits and incentives used to win the first sale; and any affiliate or partner commission on new customers.

Out: anything that would exist if acquisition stopped tomorrow — product engineering, general and administrative overhead, and the cost of serving existing customers. Retention and customer-success spending sits outside, though the line between winning a customer and keeping one is genuinely blurry in subscription businesses.

The arguments about where to draw the line are less important than drawing it once. A figure computed one way in January and another way in June is not a trend, and most disputes about whether acquisition is getting more expensive turn out to be disputes about the numerator.

New customers. Not orders, not conversions, not sign-ups, not renewals.

This sounds obvious and is the most common place the figure gets quietly wrong. An ecommerce business dividing total acquisition cost by total orders is including repeat purchases from customers it already had, which understates the cost of acquiring a new one — often dramatically, because the repeat base grows faster than the new-customer base in any healthy business.

Timing is the second trap. Spend happens in the month it happens; customers arrive when the sales cycle finishes. With a ninety-day cycle, dividing this month's cost by this month's customers compares one thing to a different thing. The clean fix is to align the windows — cost incurred in a period against customers who entered the funnel in that period — which requires tracking that connects the two and is worth building once.

Customer acquisition cost is almost never read alone. It is read as the denominator of two other figures, and it is those figures that carry the decision.

The first is the ratio between the value of a customer over their lifetime and the cost of acquiring one. A widely cited rule of thumb holds that a healthy business sits at three to one or better: a customer worth less than three times their acquisition cost leaves too little for everything else the business has to pay for. The ratio is a heuristic rather than a law, and the right level depends on gross margin, on how fast the business wants to grow, and on whether growth is funded by profit or by investors.

The second is payback period: how many months of contribution it takes to earn the acquisition cost back. This is the cash-flow question, and it is the one that constrains growth in practice. A business that takes eighteen months to recover its acquisition cost has to fund those eighteen months from somewhere, however attractive the lifetime ratio looks on a spreadsheet.

Both views are necessary and each is misleading on its own.

The blended figure — all acquisition cost divided by all new customers — has one enormous advantage: it cannot be double-counted. Attribution systems routinely credit the same customer to several channels, so the sum of per-channel customers exceeds the number of actual customers and every per-channel figure looks better than reality. The blended number is immune to this, which makes it the honest measure of whether the business as a whole is getting cheaper or more expensive at acquiring customers.

Its disadvantage is that it cannot guide a budget. It says nothing about which channel to feed.

The per-channel figure guides budgets and inherits every weakness of the attribution behind it. Read together, with the blended figure as the control, they are hard to fool. Read separately, each supports a bad decision: the blended one leads to no action, and the per-channel one leads to piling money into whichever channel is best at claiming credit.

The media half responds to the usual work — creative, targeting, exclusions, landing pages, bidding against a ceiling rather than an average.

The non-media half is larger than most teams expect and less examined. Close rate is the biggest lever in it: a sales process that converts a quarter of qualified leads instead of a fifth lowers the acquisition cost by a fifth with no change in advertising at all. Speed of follow-up, the qualification the form does, the handover between marketing and sales, and how much discount is needed to close all sit here.

Mix is the third lever. Channels differ enormously in their acquisition cost and in how far they can scale before it climbs, and the blended figure moves with the mix even when no individual channel changes.

And the last one is the customer being acquired. Segments differ in what they cost and what they are worth, and a business that knows which segment repays its acquisition cost fastest can aim at it deliberately rather than accept whatever the auction delivers.

Cost per acquisition is what the ad platform charged for a conversion; customer acquisition cost is what the business paid to win a customer. The gap between them is everything the platform cannot see — agency fees, creative production, tooling, sales salaries, discounts used to close — and in businesses with a sales team the gap is routinely a factor of two or more. The second difference is the denominator: cost per acquisition counts whatever event the account configured, which is frequently a lead rather than a customer, while customer acquisition cost counts customers. Read cost per acquisition to manage campaigns week to week, because it moves fast and points at the thing you can change. Read customer acquisition cost to decide whether the channel should exist at all, because it is the only one of the two that can be compared with what a customer is worth.

• Fix the numerator in writing before anyone argues about the number. Whether agency fees, creative production, tooling and sales salaries are inside it is a decision, and every comparison afterwards depends on having made it once.
• Count new customers in the denominator, not orders. Repeat purchases from existing customers belong nowhere near an acquisition cost, and including them is the most common way the figure gets quietly understated.
• Match the period of the cost to the period of the customer. Spend in a month that closes in the following two produces a flattering figure in the first month and a punishing one in the third unless the windows are aligned.
• Report it by channel and blended at the same time. The per-channel figure tells you where to move money; the blended figure tells you whether the business is getting cheaper or more expensive at acquiring customers.
• Work the close rate and the onboarding cost, not only the media. In businesses with a sales team the non-media half of the numerator is usually the larger half and almost always the less examined one.