CPA (cost per acquisition) is the average amount of ad money it takes to produce one sale. Spend €1,000 on ads, get 25 orders, and each order cost you €40 to acquire. It's a metric — a number — not a dimension (a category you slice numbers by).
Think of it as the price tag on a customer. Every product in your shop has a cost price; CPA is the cost price of the person buying it. And just like with products, the question is never "is €40 a lot?" — it's "a lot compared to what?"
One acronym, three names
CPA is the rare metric with three official expansions, and people swap between them mid-sentence without warning. Here's the decoder:
- Cost per Action — the broadest version: the cost of any tracked action, whether that's a lead form, a phone call, or an add-to-cart.
- Cost per Acquisition — usually means acquiring something that matters commercially: a sale, or a new customer.
- Cost per Conversion — the literal platform metric: what Google Ads and friends actually report (cost ÷ conversions, whatever you've defined a conversion to be).
In everyday use they're interchangeable, and in e-commerce all three usually boil down to the same question: what did one purchase cost me? The practical advice: define CPA in your reporting against the conversion that genuinely matters — for a shop, that's purchases — and the three names stop causing trouble.
The formula (with actual numbers)
CPA = Total Cost ÷ Total Conversions
€1,000 spent, 25 purchases → 1,000 ÷ 25 = €40 CPA.
And here's the genuinely useful party trick — CPA splits cleanly into two levers:
CPA = CPC ÷ Conversion Rate
A €0.80 click and a 2% conversion rate gives you a €40 CPA. Which means there are exactly two ways to improve it: pay less for clicks, or convert more of them. Every CPA tactic ever written is one of those two in a costume.
Why should I care about CPA?
Because it's the number that decides whether your advertising is a business or a hobby. If a sale brings you €30 of gross profit and costs €40 to acquire, you are paying customers €10 each for the pleasure of their company. Generous, but not a strategy.
It's also a diagnostic, if you read its trend together with conversion rate: a rising CPA with a flat conversion rate means clicks got more expensive — auction pressure, usually competitors or seasonality. A rising CPA with a falling conversion rate points inward: landing page, offer, or tracking. Same symptom, opposite directions to go looking.
What's a good CPA?
The only universally true answer: lower than your gross profit per order. A €40 CPA is excellent on a product where you earn €120 per order and ruinous on one where you earn €25. There is no good CPA in the abstract — only good CPA relative to margin.
For loose orientation: paid-search e-commerce purchases often land in the €20–50 range, B2C lead generation around €30–80, and B2B leads frequently €100+ (those customers are worth far more). Display and social prospecting run higher than search because the intent is lower. CEE markets, including Romania, generally enjoy lower click prices than Western Europe, which pulls CPAs down proportionally — one more reason your own numbers beat any global table, including this paragraph.
The classic mistakes
Reading CPA without margin. Covered above, but it's the one that actually bankrupts campaigns, so it earns a second mention. Know your gross profit per order. Tape it to the monitor if necessary.
Trusting the blended average. A healthy account-level CPA can hide one campaign burning budget at three times target while the others compensate. Averages are where bad campaigns go to hide.
Conversion definition drift. A pattern we see surprisingly often: someone adds a micro-conversion — say, newsletter signups — to the conversion column. CPA "improves" overnight, mild celebration ensues, and nothing real has changed; the metric is now counting cheaper things. If CPA improves dramatically and you didn't do anything, check what counts as a conversion before opening anything fizzy.
Panicking at this week's CPA. Conversions trickle in for days after the click, so the most recent period always looks worse before the data finishes arriving. Recent CPA is a rough draft, not a verdict.
How do I reduce my CPA?
Six levers, in rough order of effort-to-payoff:
- Cut waste first. Exclude search terms, placements, geos, and devices that spend without converting. Fastest wins available, zero creative work required.
- Improve conversion rate. Align the landing page with the ad's promise, speed up the site, simplify checkout. Doubling CVR halves CPA without touching a single bid — it's the same math we met in the formula.
- Feed target bidding properly. Target-CPA and target-ROAS strategies work when they have data — roughly 30+ conversions per month per campaign. Below that, the algorithm doesn't optimize so much as guess with confidence.
- Shift budget to proven segments. Remarketing, brand searches, high-intent queries. One honest caveat: scaling spend usually raises marginal CPA — the easy customers get bought first.
- Fix measurement. Proper conversion tracking, deduplication, consent setup. In our experience a surprising share of "CPA problems" turn out to be conversions that happened but were never counted — the campaign was fine; the tracking was lying.
- Improve ad relevance. Better CTR lifts Quality Score, which lowers CPC, which lowers CPA even if your conversion rate never moves. The auction's standing discount for being interesting.
Related metrics worth knowing: CPC and conversion rate (CPA's two parents), ROAS (the same economics viewed from the revenue side — and ROAS = AOV ÷ CPA, neatly), and AOV (which decides how much CPA you can afford).
Key Idea: There is no such thing as a good CPA — only a CPA that's lower than the profit on the order it bought.
This week's homework: write down your average gross profit per order. Then open your ad account and find every campaign whose CPA exceeds it. Pause the worst offender or fix its targeting. One number, one comparison, real money.
If you'd rather see CPA next to margin and conversion rate on one screen — instead of reconstructing it in a spreadsheet each month — that's the sort of thing we build at airdan.ai.
FAQ
What is a good CPA for e-commerce? One that's below your gross profit per order. As loose orientation, paid-search e-commerce CPAs often run €20–50, but margin and market matter far more than any benchmark.
What's the difference between cost per action and cost per acquisition? Cost per action covers any tracked action (forms, calls, add-to-carts); cost per acquisition usually means a sale or new customer. In day-to-day e-commerce use, both usually mean cost per purchase.
How do I lower my CPA? Two levers exist: pay less per click (better relevance, negative keywords, sane bid targets) or convert more clicks (better landing pages, faster site, simpler checkout). Everything else is a variation.
What's the difference between CPA and ROAS? CPA measures cost per order; ROAS measures revenue per euro of ad spend. They're two views of the same economics — in fact, ROAS = AOV ÷ CPA.