CAC vs CPA: One Is a Bid, the Other Is a Verdict
CPA is a number you set. CAC is the number the business reports back. The gap has a size, and that size decides the target CPA you can afford.
Your ad accounts say you are paying $34.78 to acquire a customer. Your finance sheet says $60.00. Nobody has made an error. The two numbers are measuring different things, in different places, with different denominators, and the distance between them is not noise. It has a size, it is stable enough to measure, and it is the number that should be setting your bids.
Most of what is written about CAC vs CPA stops at the definition. Acquisition means a paying customer, action means whatever you told the pixel to count, and there the article ends. That is true and it is useless, because it tells you nothing about what to type into the target field on Monday morning. The useful question is narrower: if my business can afford a customer at $72.24, what target CPA do I set so that the platform delivers one at $72.24 rather than at double that?
The definition everyone agrees on settles nothing
CPA is a platform number. It is media spend divided by the conversions that one platform is prepared to claim, inside its own attribution window, counted by its own pixel. You can set it. You can type a target into a campaign and the bidding system will chase it.
CAC is a business number. It is everything you spent to acquire customers divided by the customers you genuinely acquired, counted once each, in the ledger. You cannot set it. It arrives at the end of the month as a verdict on decisions you already made, which is why a target CAC field does not exist in any ad platform and never will.
So the comparison is not between two metrics. It is between an input and an output. The whole practical problem is that people take a number they computed on the output side, the affordable CAC, and type it into the input side, the target CPA, as though the two scales were the same. They are not, and the ratio between them is usually somewhere between 1.4 and 2.
What sits between the two numbers
Take one month of a business that sells at an average order value of $86.00 with a 42 percent contribution margin, so every order clears $36.12 after the cost of goods and the variable cost of fulfilling it. Paid media that month was $40,000.
Meta reported 700 conversions. Google reported 450. Summed, that is 1,150 conversions, and $40,000 divided by 1,150 gives a blended platform CPA of $34.78. The ledger tells a different story: 1,000 distinct people placed a first order. The 150 missing ones are not missing at all. They are buyers who touched both platforms and got counted twice, once by each.
Then there are the costs that never appear in an ad account. A $4,000 creative and agency retainer. $16,000 of salary and tooling that exists to acquire customers. Total acquisition cost for the month is $60,000, against 1,000 real new customers, so CAC is $60.00.
| Number | Calculation | Value | Can you set it? |
|---|---|---|---|
| Summed platform CPA | $40,000 / 1,150 claimed conversions | $34.78 | Yes, it is a bid target |
| Media cost per real customer | $40,000 / 1,000 new customers | $40.00 | No, it is computed after |
| Fully loaded CAC | $60,000 / 1,000 new customers | $60.00 | No, the month reports it |
Two errors, and they multiply
The gap between $34.78 and $60.00 is a single ratio, 1.725, and the reason it is worth naming is that it splits cleanly into two independent parts that most teams fix one at a time.
The first part is a counting error. The platforms between them claimed 1,150 customers where 1,000 existed, so every platform-reported conversion is worth 1,000 / 1,150 of a real customer. That is a factor of 1.15 on the way up.
The second part is a cost error. Total acquisition cost of $60,000 against media of $40,000 is a factor of 1.50. Nothing about the customers changed, only what you are prepared to call a cost of acquiring them.
| Component | Source | Factor |
|---|---|---|
| Duplicate conversions | 1,150 claimed / 1,000 real | 1.15 |
| Costs outside the ad account | $60,000 total / $40,000 media | 1.50 |
| Loading factor | 1.15 x 1.50 | 1.725 |
They multiply. That is the part people get wrong. A 15 percent counting correction and a 50 percent cost correction feel like they should stack to 65 percent, and a team that reasons that way lands on 1.65 and thinks the job is done. The true factor is 1.725. On a $34.78 platform CPA that understates the real cost by about $2.61 per customer, which at this volume is roughly $2,600 a month walking out of the business through an arithmetic habit.
The target CPA you can afford to bid
Now the number that pays for the rest of the article. Break-even here is the contribution from the orders a customer places inside the window you are willing to wait. Two orders at $36.12 is $72.24, so $72.24 is the most this business can pay to acquire a customer before the acquisition stops funding itself. Work out your own window with the CAC payback period rather than borrowing that assumption.
The obvious move is to put $72.24 into the target field. It is wrong by a wide margin. A platform that hits a $72.24 target CPA is buying customers at $72.24 x 1.725, which is $124.61, and $124.61 against a break-even of $72.24 is $52.37 of destroyed contribution on every customer. The campaign will look like it is performing. The bidding system will report that it hit target. The month will still lose money.
Divide instead. Break-even CAC of $72.24 divided by a loading factor of 1.725 gives $41.88, and $41.88 is the target you type in. Check it the other way: $41.88 x 1.725 = $72.24. The platform reports $41.88, the business experiences $72.24, and $72.24 is exactly what it can afford. The CPA calculator and the CAC calculator will each give you one side of that; the division between them is the step nobody automates.
When the factor moves without warning
A loading factor is not a constant, and both halves of it drift for reasons that have nothing to do with performance. Add a third platform and the duplicate count rises, because a buyer who sees three ads can now be claimed three times; the counting factor climbs even though the same people are buying. Hire a second acquisition marketer and the cost factor climbs while media spend sits still. Cut media spend in a slow quarter and the cost factor climbs hardest of all, because the salaries do not shrink with the budget, which is the case where a target CPA that was safe in March quietly stops being safe in July.
It drifts downward too. Deduplicate properly across platforms and the counting factor falls toward 1.0. Grow media spend against a fixed team and the cost factor falls. Either way the target you typed last quarter is no longer the target you would compute today, and no alert anywhere will tell you.
Recompute it monthly. It is two divisions against numbers you already have: claimed conversions over real new customers, and total acquisition cost over media spend. If the product of those two has moved more than about a tenth, your target CPA is stale and the ROAS you are reporting on is built on the stale version.
What to do with this
Open last month. Count the conversions your platforms claimed, add them up across every account, and put that number over the count of genuinely new customers in your order table. Then put your total acquisition cost over your media spend. Multiply the two results. That product is your loading factor, and until you have it, any target CPA in any of your campaigns is a guess wearing a decimal point.
Then take your affordable CAC, divide it by that factor, and set the result as the target. One division, once a month. It is the difference between a platform that reports success and a business that has it.
Benchmarks in this piece are illustrative. The figures come from one worked example chosen because every intermediate reproduces the printed result, not from a survey, and your own loading factor is the only one that should set your bids.
Related reading: the CAC formula and what a good ROAS actually means and cpa formula.