Customer Acquisition Cost: The Formula, and the Four Things People Leave Out
Sales and marketing spend divided by new customers. The arithmetic takes one line; the disagreements take the rest of this page, and they move the number by more than any of the arithmetic does.
Customer acquisition cost is the easiest metric in this whole category to calculate and one of the easiest to get wrong, because nothing about the formula tells you what belongs in it. I have watched two people compute CAC for the same company in the same week and come out 40% apart, both correct by their own definition, neither having written that definition down. Here is the formula, then the four decisions that actually determine the answer.
The formula, stated once
For a chosen period:
CAC = total sales and marketing spend / new customers acquiredSpend $40,000 in a quarter, acquire 200 new customers, CAC is $200. That is the whole calculation. Every argument worth having is about the two inputs.
Decision one: what goes in the numerator
The narrow version is ad spend. The honest version is everything you spend to make acquisition happen: ad spend, the salaries of the people doing sales and marketing, agency fees, the tools that only exist to support that work, content production, commissions on new business.
The gap is not small. A two-person marketing team on $75,000 each, running $30,000 a quarter in ads, has a numerator of roughly $67,500 a quarter with salaries and $30,000 without. Same company, same customers, CAC differs by 2.25x.
Decision two: who counts as a new customer
Harder than it sounds once you look at a real export. A returning customer who lapsed for a year and came back — new or not? A free trial that converted this month but signed up four months ago — which period owns them? An account that bought three times in the period — one customer or three orders?
The last one catches people constantly, because an order export has one row per order and counting rows gives you orders, not customers. You need distinct customers, and you need a customer identifier that is actually stable — email addresses change case and get typo'd, and Jane@co.com and jane@co.com will happily count twice.
Decision three: blended or paid
This is the one that most changes what the number is good for.
- Blended CAC divides all spend by all new customers, including everyone who arrived through word of mouth, organic search or a referral. It answers: what does it cost this business to grow.
- Paid CAC divides paid spend by only the customers that paid spend acquired. It answers: is the money we are spending working.
Blended is always the lower number, because organic customers dilute the denominator without adding to the numerator. Presenting blended CAC while discussing ad performance is the most common way this metric gets quietly abused, and it is rarely deliberate — it is usually just whichever number the spreadsheet already had.
Decision four: which period owns the cost
Money spent in January acquires customers in February and March. Divide January's spend by January's customers and you have compared a cause to the wrong effect. It matters most where the sales cycle is long or the spend is lumpy — a campaign that launched three weeks before quarter end will make that quarter look expensive and the next one look brilliant.
There is no clean fix in a spreadsheet, only an honest choice: either lag the spend by roughly your sales cycle, or use periods long enough that the smearing does not dominate. Quarterly is usually enough. Weekly CAC on a 60-day sales cycle is noise with a decimal point.
The most useful thing I ever did with CAC was stop reporting it as one number. We reported paid CAC by channel, and blended CAC for the business, on the same page, with both definitions printed underneath in small type.
It was not more accurate. It was harder to argue with. The argument in those meetings had never actually been about acquisition cost — it was about whose number was the real one, and printing both definitions ended it permanently. A metric that everyone computes differently is not a metric, it is a topic.
If you take one thing from this page, take that: write the definition next to the number. It costs a line and it saves the meeting.
CAC on its own means nothing
A CAC of $400 is excellent or ruinous depending entirely on what a customer is worth. The standard pairing is the LTV:CAC ratio — lifetime value divided by acquisition cost.
| LTV:CAC | Usual reading |
|---|---|
| below 1:1 | You lose money on every customer acquired. |
| around 3:1 | Commonly treated as healthy. |
| above 5:1 | Often underspending rather than winning. |
Treat 3:1 as a convention rather than a law — it comes from venture-backed SaaS and travels badly to businesses with different margins. The related number, CAC payback period, is often more useful for a small business: how many months of gross profit it takes to earn back what you spent acquiring the customer. That one is about cash, and cash is what actually constrains you.
Doing it from your exports
The inputs usually live in three different files, which is most of why this is annoying rather than hard.
- Spend. Ad platform exports for paid, the P&L or payroll for the loaded version. Sum to one figure per period.
- New customers. From the order or deal export, take distinct customer identifiers whose first order falls in the period. In Excel that is a helper column with
=MINIFS(order_date, customer_id, [@customer_id]), then count the rows where the first order date lands inside the period. Normalise the identifier first —=LOWER(TRIM(email))— or case differences will inflate your customer count and deflate CAC. - Divide, and write both inputs next to the result so the next person can see what you chose.
Doing it without rebuilding it every month
The calculation is not the work. Rebuilding it every month across three exports, and remembering which definition you used last time, is the work.
This is what Quiriz is for: the definition of CAC — what is in the numerator, who counts as new, blended or paid, which period — gets written down once in Company Context, and then the same question returns the same number next month, from the same three files. Not because the model is clever, but because it is no longer re-deciding what you meant. We wrote about why that matters in why AI gives you a different number every time.
One definition, every month
Write down what your business means by CAC once, then ask for it in plain English against this month's exports. Free to start.
Try Quiriz free →