CPA vs ROAS comes down to what each measures: CPA is spend ÷ conversions, the cost of one result; ROAS is conversion value ÷ spend, the revenue each dollar brings back. They are linked by ROAS = value per conversion ÷ CPA, so a higher CPA is good news whenever value per conversion rises faster. Steer by CPA when conversions are worth about the same, by ROAS when they aren’t.
Key takeaways
- CPA is the cost of one conversion (lower is better); ROAS is conversion value per unit of spend (higher is better).
- ROAS = value per conversion ÷ CPA, so ROAS improves whenever value per conversion grows faster than CPA.
- Use CPA or cost per qualified lead when conversions are worth about the same, and ROAS when order values vary.
- Break-even ROAS = 1 ÷ margin, and break-even CPA = value per conversion × margin.
- ROAS is only as reliable as the value behind it: attribution, margins, returning customers and low volume all distort it.
What is the difference between CPA and ROAS?
CPA tells you what one conversion costs; ROAS tells you how much conversion value each unit of spend brings back. CPA ignores what a conversion is worth, and ROAS ignores how many conversions you got.
CPA = spend ÷ conversions
ROAS = conversion value ÷ spend
value per conversion = conversion value ÷ conversions
ROAS = value per conversion ÷ CPA
The last line follows from the others: divide value per conversion by CPA and conversions cancel out, leaving conversion value ÷ spend. That one identity explains every disagreement between the two metrics.
ROAS is written as a ratio (3.0), a percentage (300%) or 3:1, and all three mean $3 of conversion value for every $1 spent. In Google Ads, it is the Conv. value / cost column.
How can a higher CPA be good news?
A higher CPA is good news when value per conversion rises by a larger percentage than CPA, because ROAS then goes up: each conversion costs more but brings in even more.
Example: a campaign spends $10,000 in each of two months.
| Metric | Month 1 | Month 2 | Change |
|---|---|---|---|
| Conversions | 250 | 200 | −20% |
| CPA | $40 | $50 | +25% |
| Value per conversion | $120 | $225 | +87.5% |
| Conversion value | $30,000 | $45,000 | +50% |
| ROAS | 3.0 | 4.5 | +50% |
Month 1: ROAS = $120 ÷ $40 = 3.0
Month 2: ROAS = $225 ÷ $50 = 4.5
Change: 1.875 ÷ 1.25 = 1.50, so ROAS +50%
A CPA report calls this a 25% deterioration. In fact the same $10,000 brought in $15,000 more revenue. You see this pattern when bidding, audiences or product mix shift toward bigger orders: fewer conversions, each worth much more.
Margin makes it clearer. With a 40% gross margin, gross profit after ad spend went from $30,000 × 0.40 − $10,000 = $2,000 to $45,000 × 0.40 − $10,000 = $8,000.
The reverse happens too: CPA falls, the report looks great, and ROAS drops because the campaign found cheaper, lower-value conversions.
Should you optimize for CPA or ROAS?
Optimize for CPA when every conversion is worth roughly the same, and for ROAS when conversion values vary. The real question is whether your conversions are interchangeable.
Leads of similar value: CPA
If every lead has about the same chance of becoming a customer of about the same size, CPA is the honest metric. Better still, track cost per qualified lead: if sales rejects half the form fills from a campaign, its real cost per useful lead is double what the platform shows.
Ecommerce and variable baskets: ROAS
When one order is $30 and the next is $300, CPA counts them as equal and pushes you toward the cheap ones. ROAS weights each conversion by what it brought in. For most online stores, ROAS should steer and CPA should be a secondary check.
Leads of very different value: assign values
If a quote request for a large project is worth far more than a newsletter signup, counting both as one conversion makes CPA meaningless. Give each conversion action a value based on what it tends to be worth, then steer by ROAS.
expected value per lead = close rate × average deal value
Example: 10% × $2,000 = $200 per lead
With values in place, value-based bidding in Google Ads (Maximize conversion value, optionally with a target ROAS) can optimize for them. Revisit the values as close rates change.
What ROAS do you need to break even?
Break-even ROAS is 1 divided by your margin: at that ROAS, the gross profit from sales exactly pays for the ads. Use the margin left after product cost and other variable costs, such as shipping and payment fees.
break-even ROAS = 1 ÷ margin
Example: 1 ÷ 0.40 = 2.5
break-even CPA = value per conversion × margin
Example: $120 × 0.40 = $48
In the earlier example, both months clear the 2.5 break-even. Per conversion, month 1 left $48 − $40 = $8 of gross profit after ad cost (250 × $8 = $2,000), and month 2 left $225 × 0.40 − $50 = $40 (200 × $40 = $8,000): the same totals as above.
Break-even is a floor for ROAS and a ceiling for CPA, not a target. Set your target CPA below the ceiling by the profit you need per conversion. If repeat purchases matter, you can accept more on the first order, but only with real repeat-purchase data.
How do CPA and ROAS compare?
Side by side, each metric covers the other’s blind spot.
| Aspect | CPA | ROAS |
|---|---|---|
| Formula | Spend ÷ conversions | Conversion value ÷ spend |
| Better when | Lower | Higher |
| Best for | Leads or sign-ups of similar value | Ecommerce, variable order values, valued leads |
| Blind spot | What each conversion is worth | How many conversions, and the margin behind the revenue |
| Typical trap | Cutting campaigns that bring fewer, bigger orders | Trusting revenue the ads didn’t cause, or revenue with thin margins |
| Needs | A reliable conversion count | A reliable conversion value |
When is ROAS misleading?
ROAS misleads when the value behind it is wrong or doesn’t turn into profit. Five traps deserve a check:
- Attributed is not incremental. Platform ROAS counts revenue from people who clicked or saw an ad, including some who would have bought anyway. Brand search and remarketing often look strongest partly for that reason.
- Margins differ between products. A ROAS of 4.0 on a 20% margin loses money (break-even 5.0), while 3.0 on a 50% margin is profitable (break-even 2.0).
- New and returning customers are mixed. Revenue from existing customers inflates ROAS. If growth is the goal, look at new-customer revenue on its own.
- Low volume swings. Example: a campaign spends $1,000 and gets 8 orders of $150, a ROAS of 1.2. One extra $1,200 order doubles ROAS to 2.4.
- Platform value vs GA4 or backend. Each platform uses its own attribution and credits its own ads, and value may include tax or shipping in one place but not another. Compare the platform figure with GA4 and your backend, and make business decisions on the backend.
How should you read CPA and ROAS changes together?
Read them as a pair: together they tell you whether a change is about cost, value or both.
| CPA | ROAS | What it usually means |
|---|---|---|
| Down | Up | Clear improvement: cheaper conversions and more revenue per dollar |
| Up | Up | Fewer, more valuable conversions; usually good, but check it isn’t one large order |
| Down | Down | Cheaper but lower-value conversions; check value per conversion and product mix |
| Up | Down | A real problem: value didn’t keep up with cost |
Only the last row needs a diagnosis, and it starts with the CPA: split the CPA increase into CPC, conversion rate and campaign mix to see what moved.
Borealis applies the same logic to Google Ads with a rule called value dominance: when a campaign’s ROAS improved, a simultaneous CPA increase is not reported as a problem, and the ROAS gain is what gets reported. See how it works in Google Ads monitoring.
Frequently asked questions
What is a good ROAS?
A good ROAS is one above your break-even point, and that depends on your margin, not on an industry number. Break-even ROAS is 1 ÷ margin: 2.5 at a 40% margin, 4.0 at 25%, 5.0 at 20%. Add the profit you need on top. A ROAS of 3.0 can be excellent for one store and a loss for another.
Is ROAS the same as ROI?
No. ROAS divides revenue by ad spend; ROI divides profit by cost. Example: with a ROAS of 4.0 and a 30% margin, each $1 of ads brings $4 of revenue and $1.20 of gross profit, so profit after ad spend is $0.20, an ROI of 20% on the ad spend. A high ROAS can still hide a thin or negative ROI.
Can you use ROAS for lead generation campaigns?
Yes, once leads carry a value. Estimate each lead type’s value as close rate × average deal value, for example 10% × $2,000 = $200, and send it with the conversion. ROAS then shows which campaigns bring valuable leads, not just many leads. Without values, stick to CPA or cost per qualified lead.
Why is ROAS different in Google Ads and GA4?
Because they count differently. Each uses its own attribution model and windows, GA4 splits credit across all your channels while Google Ads credits its own ads, and Google Ads dates a conversion by the click while GA4 dates it by when it happened. The value can differ too, for example with or without tax and shipping. Expect a gap, and watch for the gap suddenly changing.
Why did my ROAS go up while conversions went down?
Because value per conversion rose faster than conversions fell. Example: on the same spend, 250 conversions worth $120 each ($30,000) becoming 200 worth $225 each ($45,000) means 20% fewer conversions and 50% more ROAS. It usually reflects bigger orders or a better product mix. Check that one or two unusually large orders aren’t behind it before adding budget.