Customer Acquisition Cost: How to Calculate It Fully Loaded Without Fooling Yourself

Customer acquisition cost, usually shortened to CAC, is the total money spent winning new customers in a period divided by the number of new customers won in that period. The word to hold on to is total. A fully loaded CAC counts people, software, discounts and agency fees alongside advertising, because all of it was spent to bring those customers in.

Two sources that investors and operators lean on say the same thing. Andreessen Horowitz's “16 Startup Metrics” argues that CAC should be the full cost of acquiring users, stated per user, and names two common errors: leaving out costs such as referral fees, credits or discounts, and calculating a blended figure that hides how paid marketing performs. David Skok's “Startup Killer” essay computes CAC from the entire cost of sales and marketing over a period, including salaries and other headcount expenses, divided by the customers acquired. He adds that for pure web businesses whose headcount does not grow with customer volume, it is also useful to look at CAC without the headcount costs.

Both of those are written from a software-startup view, and a plumber or a regional agency has different cost lines. The logic travels, though. The usual way CAC goes wrong is not complicated arithmetic. It is a numerator that quietly shrank to the cost of ads.

That shrinking is easy to do by accident. Google Ads describes conversions as whatever you choose to identify as valuable, such as a purchase, sign-up or phone call. A platform's cost per conversion therefore measures the cost of the action you configured, which is often a form fill, and only for the spend inside that platform. It is a useful operating metric. It is not the cost of a customer.

Decide the cost list once, write it down, and apply it every period. The table sets out the lines most businesses debate. Where a line is a judgment call, the useful discipline is consistency, so that a rising CAC means customers got dearer and not that someone changed the definition.

Every figure in this section is illustrative. The numbers were chosen to show the arithmetic and describe no real business, client or market. Imagine a subscription-style service that spent the amounts below in one month and won 40 new customers, 20 of whom came from paid channels and 20 from referrals, search and direct visits.

With the illustrative costs above, the same month produces six different figures depending on which numerator and denominator you choose. The spread is the point: a team that reports only the first row will believe it acquires customers for less than half of what they cost.

Customers earn, in this example, $400 a month in revenue at a 60 percent gross margin, which makes monthly gross profit per customer $240. Payback in months is then the CAC divided by $240. The payback arithmetic is ordinary division and is shown here for the example only.

Blended CAC divides all acquisition cost by all new customers, whichever route they took. Paid CAC divides the cost of paid marketing by the customers credited to paid marketing. Channel CAC does the same for one channel at a time. The a16z essay says blended CAC is not wrong but does not show how paid campaigns are performing or whether they are profitable, and that investors weight paid CAC more heavily for exactly that reason: it informs whether the company can scale up acquisition spend profitably.

There is a catch the definitions do not remove. Which customers count as paid depends on an attribution rule, and advertising can plant an idea that a person acts on later through search or a direct visit. The essay itself allows that paid acquisition may feed organic acquisition, but says you would need to demonstrate that effect before leaning on the blended figure, and that many investors like to see both. Our reading is that paid CAC can overstate the cost of paid marketing, and organic CAC can look better than it deserves because it harvests earlier ad exposure. Marketing Attribution Models Explained covers how the credit rules differ, and Attribution Counted the Conversion. Did the Advertising Cause It? shows how to test the causal question instead of arguing about credit.

A related trap appears when platform numbers and the CRM disagree. If the CRM shows 40 new customers and the ad platforms together report 65 conversions, the gap is usually explained by leads counted as conversions, the same person counted twice, and several platforms each claiming credit for one customer. Start from the CRM count, which can only count a customer once, and treat platform conversions as clues about which channel touched whom.

CAC payback period is the number of months of gross profit from a new customer needed to recover what it cost to acquire them. The formula is CAC divided by monthly gross profit per customer, where monthly gross profit is monthly revenue multiplied by gross margin. Use margin, not revenue. The a16z piece makes a parallel point about lifetime value: it favors a contribution-margin basis, after variable costs, because a revenue or gross-margin basis suggests a higher ceiling on acquisition spend. It also calls the contribution-margin LTV to CAC ratio a good measure for working out CAC payback. This article uses gross margin because it is the figure most small teams can actually pull; use contribution margin if you can.

Two refinements keep the number honest. First, customers leave. If a share of them cancel before the payback month, the cohort needs longer than the simple formula says, so calculate payback on a cohort that has had time to age. This Month’s Leads Have Not Had Time to Become This Month’s Revenue explains why a young cohort flatters every metric. Second, a business that sells one-off jobs has no monthly stream. There, compare CAC with the gross profit on the first job and ask whether repeat work covers the gap.

On benchmarks, treat what you read with the origin in mind. Skok's SaaS guidelines, drawn from software companies, say many of the best software businesses recover CAC in five to seven months and that profitability turns anemic once recovery stretches beyond twelve. He calls them guidelines. A cash-rich lender, a high-margin software firm and a local service business sit in different places, and no published threshold is a rule for yours. The better test is internal: how long can you afford to wait for cash to come back, and does the payback you measure sit inside that?

Pick the customer unit before the formula. A person who bought once with a steep coupon and a company that signed a twelve-month contract are not the same unit, and dividing spend by a count that mixes them produces a number no one can act on. The table shows where the numerator and the payback basis tend to shift. These are our working conventions, not standards.

Whatever the business, keep the unit constant across periods and say what it is in the report heading. When a business sells to both households and companies, calculate CAC for each, because the blended figure will describe neither. And when the first purchase is a loss leader, say so openly and report payback on the second purchase as well, so the strategy can be judged on what it was designed to do.

Most of the damage comes from a short list of habits. Each one is easy to fix once it has a name.

Work backward from the customer's gross profit. The most you can sensibly spend to win a customer is the gross profit they bring multiplied by the share of it you are willing to invest in acquisition. Lead Generation for Small Business: Which Two Channels to Pick, and How to Run Them for 90 Days walks through the same logic at the cost-per-lead level, and the two figures should agree: your allowable CAC divided by your close rate gives your allowable cost per lead.

CAC also cannot tell you everything. It says nothing about whether acquired customers stay, expand or complain. That is why the investor guidelines pair it with lifetime value, and why Skok's guideline that LTV should exceed CAC by more than three times is a software-company rule of thumb rather than a universal law. If your customers are worth less over time than your model assumes, a flattering payback hides a problem for another year.

Use the number to make three decisions. Whether to raise or cut a channel's budget. Whether the sales process is too expensive for the average deal size. And whether growth needs cash you do not have, since payback length is partly a cash-flow question. If a figure is not informing one of those decisions, it is a vanity metric with a formula.