How to calculate cost per customer acquisition takes one sentence. Add up what you spent to win new customers in a period, and divide by the number of new customers you won in that period. The result is your customer acquisition cost, or CAC.
The division is the easy part. Two founders with identical businesses can report very different CAC figures, and both can be doing the sum correctly. They made different choices about what to count. This guide goes through those choices one at a time, for a consumer company: an app, an online store, a brand.
It covers the formula, the difference between paid and blended CAC, what belongs in the spend, the period trap, and why a good CAC cannot be given as a single number. There is a worked example with invented figures, and a free calculator if you would like the sums done for you.
The customer acquisition cost formula in words
The customer acquisition cost formula has two parts. On top is the money spent on winning new customers during a period. Underneath is the count of new customers won during the same period. Spend divided by customers.
Each part hides a decision. The top asks which costs count as acquisition. The bottom asks who counts as a new customer. And the words "the same period" hide a third decision, because spending and buying do not always happen in the same month.
So when someone asks how do you calculate cost of customer acquisition, the full answer is: decide three definitions, write them down, then divide. The definitions matter more than the arithmetic. A CAC built on unstated definitions cannot be compared with anything, including your own figure from last quarter.
Paid CAC and blended CAC answer different questions
There are two common versions of the figure, and mixing them up is the most frequent mistake.
Paid CAC takes advertising spend and divides it by the customers the advertising brought. It answers a narrow question: are the ads paying for themselves?
Blended CAC takes all acquisition spend and divides it by all new customers, wherever they came from. That includes people who found you through a friend, a search for your name, or your own social accounts. It answers the question the business lives on: what does a new customer cost us, on average?
| Paid CAC | Blended CAC |
|---|---|
| Ad spend divided by customers credited to ads | All acquisition spend divided by all new customers |
| Tells you whether a channel earns its budget | Tells you whether the company can afford its growth |
| Depends on the ad platforms' own count of who they converted | Depends only on your own records of spend and customers |
| Cannot see customers who arrived without an ad | Can hide a weak ad account behind strong word of mouth |
Neither is the real one. Calculate both, every period, and label them. If a board slide or an agency report says CAC with no label, ask which.
The third row of that table deserves a note. Paid CAC rests on a count you did not make. Researchers studied large advertising experiments at Facebook covering 1.6B ad impressions and compared the usual ways of measuring a campaign with randomized experiments on the same campaigns. The usual methods often failed to match what the experiments showed. The study was published in Marketing Science in 2019. Treat a platform's count of its own conversions as an estimate.
What to include in the spend
The narrowest version counts advertising only. The widest counts every cost that exists because you are trying to win customers. Most of the distance between two companies' CAC figures comes from where they drew this line.
These are the costs to decide on:
- Advertising spend on every platform.
- Fees to agencies, freelancers and creators who work on acquisition.
- The cost of making the content and the ads: shoots, editing, product given away for filming.
- Software used for marketing and sales.
- The pay of people on your team who work on acquisition, or a fair share of it if they do other things too.
- Discounts, free trials and first-order offers given to new customers. Margin you give away to win an order is acquisition spend, even though no invoice arrives for it.
- Referral rewards and affiliate commissions.
A CAC that counts only advertising is often called media-only. One that counts everything on the list is called fully loaded. Both are legitimate. The fully loaded figure is the one to use when you ask whether the business works, because salaries and discounts are paid with the same money as ads.
One cost is commonly left out by accident: the founder's own time. If the founder spends two days a week on content, that is acquisition work, and the fully loaded figure should carry a fair value for it.
What to leave out: costs of serving a customer you already have. Support, shipping the order, hosting, the product itself. Those belong to the margin on a sale and have no place in the cost of winning one.
Who counts as a new customer
The bottom of the formula looks simpler than it is. A customer is someone who paid. For a store, that is a person placing a first order. A second order from the same person is not a new customer, however good it is to see.
For an app, the definition takes more care. An install is not a customer. A free signup is not a customer either, unless your business earns from free users. If the app sells a subscription, the customer is the person who starts paying. Cost per install is a useful number for judging ads. It is a different number from CAC, and an app can have a wonderful cost per install on its way to running out of money.
A free trial adds a delay. People who start a trial this month become paying customers next month, or do not. Count them when they pay.
For a marketplace there are two kinds of customer, buyers and sellers. Calculate a CAC for each side separately. An average of the two describes nobody.
Whatever you choose, take the count from your own records: the order system, the payment provider, the subscription dashboard. Do not add up the customers each ad platform claims. Two platforms can each claim the same person, and the total will exceed the number of people who bought.
The period trap
The formula says the same period, and that is where a correct sum produces a wrong answer. Money is spent on one date. The customer it wins may arrive weeks later.
Take a launch. You spend heavily in the month before and during it. Some customers buy straight away. Many more buy the following month, after they have seen the product a few times. A monthly CAC makes the launch month look terrible and the next month look brilliant. Neither is true. The spend and the customers belong together, and the calendar split them.
The same thing happens with slower channels. Videos on your own accounts, a creator's post, a press mention: each can go on bringing customers long after the cost was booked.
There are three ways to handle it, from simplest to most exact.
- Use a longer period. A quarter smooths out most of the timing problem. For a company with a long gap between first contact and purchase, a quarter is the shortest period worth reporting.
- Use a rolling window. Each month, calculate CAC across the last three months together. The figure moves slowly and hides less.
- Match spend to the customers it produced. If you know your typical delay between first visit and purchase, compare each month's spend with the customers who arrived that much later. This is the most accurate and the most work.
Whichever you choose, do not judge a single month alone. And never compare a monthly figure with a yearly one. Seasons move both the spend and the buying, so only like periods compare.
A worked example with invented numbers
Now the period trap, with the same imaginary brand and figures that are just as invented.
If you would like to run your own figures through the same steps, the free calculator does the division for you. Bring your spend and your customer count for one period.
What is a good customer acquisition cost?
What is a good customer acquisition cost is the question every founder asks next. A CAC is good when the customer pays it back, with margin left over, within a time the business can afford to wait. That makes it a comparison between two numbers. The CAC is one. What the customer is worth is the other.
What a customer is worth is the margin they leave you over the time they stay, after the cost of the product and of serving them. Revenue is the wrong measure here. A customer who spends a lot on a product with thin margins can be worth less than one who spends little on a product with wide ones.
Three things set that worth: how much margin each purchase leaves, how often the customer buys again, and how long they stay. An app with a monthly subscription and a brand selling one item once are in different positions even if their CAC is identical.
The time it takes a customer to repay their own acquisition cost is called the payback period. It matters as much as the total, because a company has to fund the gap. A CAC that pays back in the first order is easy to live with. One that pays back over a year needs cash in the bank to cover that year.
This is why repeat purchase changes the answer so much. On one direct to consumer brand in our portfolio, the second-purchase rate rose 50% and the third-purchase rate rose 40%. The receipt records those two results. It does not record the brand's CAC, so I will not attach one. The point stands without it: when more customers come back, the same acquisition cost buys more, and a CAC that was too high can become affordable with no change to the ad account.
Why this guide gives no average cost per customer acquisition
Search for the average cost per customer acquisition and you will find tables of figures by industry. This guide publishes none. I could not source an average that applies across categories, and I would sooner leave the space empty than fill it with a number that looks official.
The trouble with those tables is the subject of this whole guide. A published average rarely says whether it is paid or blended, media-only or fully loaded, or what it counted as a customer. Averages of figures built on different definitions describe no real company. And even a clean average would mix a product that costs a few coins with one that costs a month's rent.
Industry CAC tables are popular because they answer a hard question quickly. Unfortunately it is somebody else's question.
The benchmark that matters is your own. Your CAC against what your customer is worth, and your CAC this quarter against the last, calculated the same way both times.
Five mistakes that make CAC look better than it is
Counting returning customers as new. If repeat orders slip into the bottom of the formula, CAC falls on paper. Use first-time buyers only.
Leaving out discounts. A first-order code is a cost of winning that order. Count it.
Reporting only the flattering version. Blended media-only CAC is almost always the lowest figure available. It has its uses, and it should never be the only one on the page.
Crediting ads with customers who were coming anyway. eBay tested this on itself. The company switched off search ads on its own brand name and measured what it lost. It found no short-term return from those ads. The study was published in Econometrica in 2015. eBay is a very well known name, so your result may differ, but the principle holds for any company: an ad shown to someone already on their way to you adds spend to the top of the formula and nobody new to the bottom.
Changing the definition between periods. If last quarter's figure was media-only and this quarter's is fully loaded, the comparison is meaningless. Write the three definitions on one page: what counts as spend, who counts as a customer, and which period. Keep that page with the figure.
How often to calculate it, and what to do with the result
Calculate CAC every month, and read it on a rolling three-month basis. Keep paid and blended next to each other, with the fully loaded figure beside them.
Then watch the gap between paid and blended. If blended sits well below paid, a good share of your customers arrive without an ad, through your own accounts, searches for your name or word of mouth. If the two are nearly the same, almost every customer is being bought, and the business stops growing on the day the budget does.
Once the figure is honest, the next question is how to bring it down. That is a separate guide, which ranks the options by how long each saving lasts.
Read how to reduce customer acquisition cost →
What our own CAC figure covers
Having argued that a CAC figure means little without its definition, I should apply that to ours. Across Underboss portfolio engagements, the average CAC reduction is 35%. Paid ad spend across the 2025 portfolio was $0.
That is an average of reductions across engagements. Individual results sat above and below it. It is not a CAC in money terms, it does not say which change moved the cost on any one engagement, and it is no forecast for your company. I would hold any figure you are shown, from any agency, to the same questions this guide asks of your own.
See how we work with consumer companies →
Key takeaways
- CAC is acquisition spend divided by new customers won in the same period. The definitions behind those words matter more than the division.
- Calculate paid CAC and blended CAC, and label each. Use the fully loaded figure to judge whether the business works.
- Count discounts, creator fees, tools and a fair share of salaries as spend. Count only first-time paying customers, from your own records.
- Match spend to the customers it produced. Use a quarter or a rolling three months, and never judge one month alone.
- A good CAC is one the customer pays back with margin to spare. There is no single good number, and this guide publishes no industry average.
If you have the figure and want to know what would lower it for your company, ask for the free teardown. Send a link to the product and its channels. Within three days you get a written read on what brings customers in today and what we would build first. No call, and the plan is yours whether or not we work together.