The calculator above turns spend and conversions into a CPA, then checks it against what a conversion is worth to you. Here's what the number means, how it chains back to CPC and conversion rate, and where it goes wrong.
What CPA is and how to calculate it
A subscription box company came to us proud of a $38 cost per acquisition. Their agency had been reporting it for two years and it had been drifting down, which everyone took as good news. Their first-box gross profit was $31. They'd been paying $7 more than each new customer was worth on day one, and because the average subscriber stayed 2.4 months, the real payback happened, barely, in month two. Nobody had done the math. They'd been looking at a number without a line to compare it to.
CPA is cost per acquisition, sometimes cost per action. Google calculates it as the total cost of conversions divided by the number of conversions. Meta reports the same thing as cost per result, where the result is whatever you told the campaign to optimize for. Sale, lead, signup, install, call. The formula doesn't care what the action is, which is exactly why the number on its own tells you so little.
The formula. CPA = ad spend / conversions. $4,000 in spend for 80 sales is a $50 CPA. Whether $50 is good depends entirely on what a sale is worth, which is the second half of the calculator.
The number that makes CPA useful, your maximum affordable CPA
Every business has a ceiling on what it can pay for a customer. On a single order it's gross profit per conversion, which is average order value times gross margin. A $120 order at 40% margin carries $48 of gross profit, so $48 is the break-even CPA. Pay $47 and you made a dollar. Pay $50 and you paid the ad platform for the privilege of shipping a box.
| Order value | Margin | Break-even CPA | CPA for 25% profit on spend |
|---|---|---|---|
| $40 | 50% | $20.00 | $16.00 |
| $80 | 30% | $24.00 | $19.20 |
| $120 | 40% | $48.00 | $38.40 |
| $250 | 35% | $87.50 | $70.00 |
| $600 | 60% | $360.00 | $288.00 |
| $2,000 | 80% | $1,600.00 | $1,280.00 |
The last column is where a Target CPA should come from. Decide what return you need on the money at risk, then set the target below break-even by that much. Google's Target CPA bidding will work toward whatever number you give it, and it has no idea what your margin is. Feed it last quarter's average and you've told the algorithm to keep doing what it was doing.
Lead generation works the same way with one extra step. A lead is worth close rate times average deal value times margin. If 10% of leads close, the average deal is $3,000, and margin is 50%, a lead is worth $150 and that's the ceiling. Most lead-gen accounts we've reviewed never calculated it and were guessing at a target CPA from what felt reasonable.
Above this line you're funding growth out of pocket. That can be a deliberate choice if repeat purchases pay it back. It's a disaster when nobody knew the line was there.
How CPA chains back to CPC and conversion rate
CPA is CPC divided by conversion rate. That's the whole relationship, and it means there are exactly two ways to move it.
| Scenario | CPC | Conversion rate | CPA |
|---|---|---|---|
| Starting point | $2.00 | 4.0% | $50.00 |
| Cut CPC 20% with bids | $1.60 | 4.0% | $40.00 |
| Raise CVR 20% with the landing page | $2.00 | 4.8% | $41.67 |
| Both | $1.60 | 4.8% | $33.33 |
| Cut CPC 40% by widening targeting, CVR halves | $1.20 | 2.0% | $60.00 |
Look at the last row. That's the most common way we see CPA get worse while the account looks healthier. Cheaper clicks from broader targeting, worse conversion rate, higher CPA. The dashboard shows CPC falling and click volume rising, both of which look like wins until you check the only number that matters.
The conversion rate lever is usually bigger and cheaper than the bid lever. Bids move CPC by tens of percent at best, and the platform fights you the whole way. A landing page rewrite, a shorter form, or a clearer offer can move conversion rate by half or more. Quality Score improves at the same time, which lowers CPC for free.
Four ways CPA lies to you
Not every conversion is a customer
A conversion is whatever you told the platform to count. Add-to-carts, form fills, page views with a two-minute dwell, all of them can be conversions. We've opened accounts with a $6 CPA that turned out to be $6 per newsletter signup, on a campaign that hadn't produced a sale in a month. Check what the conversion action is before you react to the number.
Broken tracking looks like bad performance
When a tag breaks or a consent banner blocks it, conversions drop and CPA doubles overnight. The ads didn't change. The counting did. In our experience a sudden CPA spike with no change in spend or CPC is a tracking problem far more often than a market problem. Check the tag before touching the bids.
Small numbers swing wildly
With 5 conversions in a week, one extra sale moves CPA by 20%. Judging a campaign on that is coin-flipping. We hold every CPA decision to at least 30 conversions in the window, and prefer more. Google's own guidance on conversion volume for automated bidding exists for the same reason.
Fees live outside the number
Platform CPA is media cost only. Add a percentage-of-spend management fee, a landing page tool, a call tracking bill, and the real cost per customer climbs. Put every fee in the spend field above. If the verdict flips from green to red, the campaign was never profitable and the reporting was hiding it.
CPA or ROAS, which one to run on
CPA works when every conversion is worth about the same. Leads, trials, signups, and single-product stores fit. ROAS works when order values vary, because a $30 order and a $300 order shouldn't cost the same to win. A lot of accounts should use both, CPA on the lead campaigns and ROAS on the ecommerce ones. The ROAS calculator runs the same break-even logic from the revenue side.
Orders vary in size? Break-even looks different from the revenue side.
Open the ROAS calculatorHow to lower CPA, in the order that works
- Fix tracking first. Every optimization downstream runs on the conversion count. If it's wrong, everything is wrong.
- Raise conversion rate. Landing page, offer, form length, page speed. This is the biggest lever and the one most agencies skip because it isn't inside the ad platform.
- Cut what doesn't convert. Sort search terms, placements, audiences, and products by CPA. Pause everything above your ceiling with enough volume to be sure.
- Lower CPC through relevance. Tighter ad groups, ads that match the query, pages that match the ad. Quality Score does the rest.
- Then set the target. Break-even CPA times the margin of safety you want. Give the algorithm that number and enough conversions to learn from it.
The bottom line
CPA is only half a number. It needs the ceiling next to it, and the ceiling is order value times margin, or for leads, close rate times deal value times margin. Run the calculator with real spend including fees. If the verdict is red, the campaign has been buying customers at a loss no matter how much the CPA improved since last quarter.