What is a good CPA for Facebook ads?
There is no universal good CPA, because a good CPA is one below what a customer is worth to you, and that depends entirely on your order value and margin. Compute your break-even CPA as average order value multiplied by gross margin percentage; any CPA below that is profitable on the first order, and your target should sit enough below it to pay for overhead and profit. A $30 CPA is excellent for a $200 order at 60% margin and ruinous for a $40 order at 40% margin.
Last updated 2026-08-11
The break-even calculation
Break-even CPA equals average order value times gross margin. If your AOV is $80 and your gross margin after product and fulfilment costs is 50%, each order generates $40 of contribution, so $40 is the most you can pay for a customer before the first order loses money. Everything starts from this number. Compute it with your real blended AOV including shipping revenue, and your real margin including payment processing, shipping cost and packaging, not the flattering version. Most disappointing what-is-a-good-CPA conversations turn out to be margin conversations in disguise, because the account was benchmarked against a number the unit economics never supported.
From break-even to target
Break-even is a ceiling, not a target. Running at break-even means acquisition is free but contributes nothing to overhead, salaries or profit, so set the target below it by the contribution you require per order. A simple structure: decide what fraction of contribution margin you are willing to spend on acquisition, and the remainder is what each order banks. A business that wants half its contribution as profit on a $40-contribution order targets a $20 CPA. This framing also makes disagreements productive, because a debate about whether $28 is acceptable becomes a debate about required profit per order, which is a business decision rather than a benchmarking one.
Repeat purchase changes the ceiling
If customers reliably reorder, first-order break-even is too conservative, because the acquisition buys a stream of orders rather than one. The adjustment is paying up to some fraction of expected contribution over a defined window, such as sixty or ninety days, rather than the first order alone. Two disciplines keep this honest: use a payback window you can actually measure and fund from cash flow, and use observed repeat behaviour from your own data rather than an optimistic lifetime value. Subscription businesses live entirely in this framing. One-off purchase businesses should mostly ignore it and hold the first-order line.
Why comparing CPAs across businesses misleads
Two stores in the same vertical can have wildly different viable CPAs because of AOV, margin structure, repeat rate and offer strength, so a competitor's rumoured CPA tells you almost nothing about what yours should be. Even inside one account, CPA varies structurally: prospecting costs more than retargeting, new customers cost more than returning ones, and different optimisation events are not comparable at all. The useful comparisons are internal: this month against your trailing average, one campaign type against the same type, and every number against your own break-even. A benchmark that does not know your margin cannot tell you whether you are winning.
Using target CPA operationally
Once the target exists, it becomes the yardstick for every mechanical decision. Kill rules key off it, for example cutting a creative that reaches a multiple of target CPA with no conversions. Test budgets derive from it, since each creative needs a multiple of CPA in spend to be judged. Scaling decisions compare a winner's CPA to target with room to degrade, because CPA typically drifts up as spend expands into colder pockets of the audience. Write the number down where the whole team sees it, and recompute it whenever pricing, product costs or shipping economics change, because a stale target quietly misgrades every ad in the account.
When CPA looks bad but is not, and the reverse
CPA judged too early in a learning phase reads high because delivery is still exploratory, and CPA on a tiny event count is noise in either direction; both are timing problems, not performance problems. In the other direction, a flattering CPA can hide trouble: optimising to a cheap event upstream of purchase produces beautiful CPAs and no revenue, and heavy discounting buys good-looking acquisition at margins that break the break-even math you started with. Whenever CPA and profitability disagree, trust the profit calculation, and check whether the attribution window or the optimisation event, rather than the ads, is what changed.
Any specific dollar figure quoted as a universally good Facebook CPA is meaningless without your AOV and margin attached. This page deliberately gives you the calculation instead of a number, because the number is yours to compute.