Cost per acquisition (CPA) is what you pay in ad spend, on average, for one conversion: a purchase, signup, booking or other action you count as an "acquisition."
You calculate it by dividing total ad spend by the number of conversions.
If you spent $1,000 and got 25 sales, your CPA is $40.
Below: the formula, how CPA differs from CAC and CPL, worked examples using real campaigns from our database, how to set a target CPA from your margin, and what actually lowered CPA in the results we've collected.
The cost per acquisition formula
The formula for cost per acquisition is:
CPA = total ad spend ÷ number of conversions
Use the same date range and the same conversion definition for both numbers.
If "acquisition" means a purchase, count purchases, not add-to-carts.
A few things to keep in mind when you calculate cost per acquisition:
- It's an average. Some conversions cost far more than others. Google says this directly about its Target CPA bidding: "some conversions may cost more than your target and some may cost less", but the average is what it aims for.
- Attribution changes the number. Meta, Google and TikTok each count conversions with their own attribution windows. The same campaign can show different CPAs in Ads Manager and in your store backend.
- Ad spend only, usually. Most people calculate CPA from media spend alone. Creative production, agency fees and tools go into CAC instead (more on that next).
The platforms don't all call it CPA. Google Ads reports "cost per conversion" (Cost / conv.), and Meta Ads Manager reports "cost per result," where the result is whatever objective you picked.
Both are the same formula. Make sure the conversion column you read is the one you care about, since an account can track several actions at once.
CPA also moves with the metrics upstream of it.
CPA equals cost per click divided by conversion rate.
A $2 click with a 4% conversion rate gives a $50 CPA.
Halve the click cost or double the conversion rate and CPA falls to $25. When CPA rises, checking which of the two changed tells you whether the problem is the ad and audience (clicks got pricier) or the landing page and offer (fewer visitors buy).
If you want to skip the math, our CPA calculator does it for you and shows how CPA moves when spend or conversions change.
CPA vs CAC vs CPL
These three get mixed up a lot. They answer different questions.
| Metric | Formula | What it tells you |
|---|---|---|
| CPA (cost per acquisition) | Ad spend ÷ conversions | How efficiently a campaign or ad buys one conversion |
| CAC (customer acquisition cost) | All sales and marketing costs ÷ new customers | What it really costs the business to win one new customer |
| CPL (cost per lead) | Ad spend ÷ leads | How much you pay for a form fill, signup or booked call, before anyone buys |
CPA is a campaign metric. CAC is a business metric: it includes salaries, agency fees, creative and tools, and it usually counts only new customers, while CPA often counts repeat buyers too.
CPL is CPA where the "acquisition" is a lead.
It is often much cheaper than CPA because a lead is not a sale yet.
In one webinar funnel in our database, @SamPiliero paid about $9 per lead on TikTok vs about $2 per lead on Meta.
That gap only matters if both lead pools convert to buyers at similar rates, which is why lead-gen advertisers should track CPL and the CPA of the final sale.
A quick illustration of all three: a brand spends $10,000 on ads, gets 500 leads and 100 sales, and also pays $5,000 in agency and creative fees.
CPL is $20, CPA is $100, and CAC (if all 100 buyers are new) is $150. Each number is correct; they just answer different questions.
Worked examples from real entries
Here is the formula applied to real campaigns credited in our archive. Numbers come from the advertisers' own posts.
Example 1: spend and purchases known. @SamPiliero ran creative packs across 64 ad sets and got 1,238 purchases on $42.5k of spend.
$42.5k ÷ 1,238 works out to about $34.33 CPA, which matches the "≈$34 CPA" in the entry.
Example 2: working backward from CPA. @runtraffic reported a $32.98 cost per purchase on $253k in a static-only Meta campaign.
Rearranging the formula, conversions = spend ÷ CPA, so $253,000 ÷ $32.98 ≈ 7,670 purchases.
That's an estimate, since the spend figure is rounded.
Example 3: CPA and ROAS together. @AshvinMelwani shared an Obvi GLP-1 angle test with a $58.96 CPA and 1.06 ROAS across 123 ads.
Because ROAS is revenue ÷ spend, revenue per purchase is roughly CPA × ROAS: $58.96 × 1.06 ≈ $62.50. In other words, the ads roughly paid for themselves on the first order.
Whether that's good depends on margin and repeat purchases, which is the next section.
How to set a target CPA from your margin
A target cost per acquisition should come from your economics, not from an industry average. Start with your break-even CPA:
Break-even CPA = average order value × gross margin %
Say your average order is $80 and your gross margin after product cost, shipping and payment fees is 60%.
Each order leaves $48 to spend on acquiring it.
Pay more than $48 per purchase and you lose money on the first order.
Then set your target below break-even to leave room for profit and overhead.
With the same numbers, a target of $35-40 leaves $8-13 per order.
You can adjust the target in two directions:
- Go higher if customers reliably buy again. If the average customer orders three times in a year, you can afford a first-order CPA above $48, as long as you have the cash to wait for those repeat orders.
- Go lower if returns or refunds are high, or if CPA in the ad platform looks better than what your backend shows.
The same logic works as a ROAS target: break-even ROAS = 1 ÷ margin. Our break-even ROAS calculator runs both.
Benchmarks are useful as a sanity check, not a target.
For example, WordStream's 2025 Google Ads benchmark report put the average Google Ads cost per lead at $70.11, up from $66.69 in 2024, across more than 16,000 campaigns.
A business with a $40 margin per sale can't use that number for anything.
Look up the average CPA for your platform and vertical, then trust your own margin math over it.
If you use Target CPA bidding in Google Ads, the target you enter is the desired average cost per conversion.
Start near your recent actual CPA rather than your ideal one.
A target far below what the account achieves now usually limits volume.
What lowered CPA in the database
We tag every result by the lever that moved it.
Across the CPA results we've collected, a few patterns show up.
Browse the full set on what lowered CPA, or see the breakdown in our research on what moves CPA.
1. The offer, not the ad. @SamPiliero cut a product's price by €10 with the same ads and budget, and cost per purchase fell 45%.
In a vendor-reported TikTok for Business story, Gill's Point S, a local tire shop, ran a holiday offer and CPA went from $15.18 to $3.14.
If CPA is stuck, test the offer before you rewrite every ad.
2. New angles for a tired account. @SamPiliero launched 20 new angles in a fatigued account and CPA dropped from $101.29 to $60.50 (−40.27%).
New angles mean new reasons to buy, not just new edits of the same message.
3. More, better creative. With a flat budget, @SamPiliero brought cost per order from €84.38 to €58.63 (−30.5%) by improving ad quality and volume.
A chart from NP Digital shared on LinkedIn found campaigns with 20+ ads had a 6.8% lower CPA than those with 1-2 ads, a smaller effect than single-account stories suggest.
4. Format and placement fit. In vendor-reported platform stories, Prose saw a 23% lower CPA after making vertical, Reels-first ads instead of reusing standard video, and AutoDoc reported 44% more conversions at a 15% lower CPA with YouTube video action campaigns.
A caution on the biggest numbers: platform and tool case studies are published by the vendor, and they pick their best wins.
Before-after results from one account also can't rule out seasonality.
Treat any single result as an idea to test, not a prediction.
If your CPA is above break-even and you've run out of ideas, an experienced media buyer can often find the leak faster than another round of tests.