ROAS vs CPAChoosing the number you steer by

Two ways of asking the same question#

Both metrics ask whether ad spend is efficient. CPA asks it in cost terms: how much did each conversion cost? ROAS asks it in revenue terms: how much revenue came back for each rupee spent? They are two views of the same relationship between spend and outcome, and choosing between them is mostly a question of what your outcomes look like.

When CPA is the right lens#

If every conversion is worth roughly the same to you, counting them is enough. A software company whose trial sign-ups convert at similar rates, a service business whose enquiries have similar value, or a subscription with a single price point can steer sensibly by cost per acquisition.

CPA's weakness appears when conversions are not equal. A lead from a large company and a lead from a student both count as one. If the bidding system is told to minimise CPA, it will find the cheapest leads, which are often the least valuable. Importing lead quality or sales outcomes, as covered in offline conversion tracking, is the fix.

When ROAS is the right lens#

If conversion values vary, revenue is the better measure. An ecommerce store selling items at very different prices needs bidding that favours larger baskets. ROAS lets the system and the reader see that one campaign producing fewer, larger orders may be better than another producing many small ones.

ROAS's weakness is that revenue is not profit. A campaign with a high ROAS on a low-margin product may lose money, while a lower ROAS on a high-margin product may be very profitable. Unless you adjust for margin, ROAS can steer budget toward the wrong products.

Finding your break-even#

A ROAS or CPA figure means nothing without a reference point. The reference point is break-even: the level at which ad spend exactly consumes the margin it generates. Below break-even ROAS, or above break-even CPA, each sale loses money before any other costs.

For ROAS, break-even is one divided by your gross margin as a fraction. At a 40 per cent margin, break-even ROAS is 2.5: you need ₹2.50 of revenue per rupee of spend just to cover the ad cost from margin. For CPA, break-even is the margin earned from the conversion. These are illustrations of the arithmetic; plug in your own margins.

Margin-adjusted and profit-based approaches#

A growing number of advertisers feed profit rather than revenue into their ad platforms. Instead of passing the order value as the conversion value, they pass the gross profit, or a value adjusted for margin by product. The platform then optimises for profit-weighted ROAS, which aligns its incentives more closely with the business.

This takes some engineering, because margins must be available at the point the conversion is recorded, and some businesses are reluctant to share margin data with platforms. A simpler step is to group products by margin band and set different ROAS targets for each group. Our ecommerce unit economics guide covers the inputs.

The lifetime value question#

Both CPA and ROAS are usually measured on the first conversion. For businesses with strong repeat purchase, that understates the value of a new customer, and steering by first-order break-even may cause you to underinvest in acquisition. Equally, assuming lifetime value you have not yet proven can lead to overspending.

A practical compromise is to set acquisition targets based on the value a customer reliably generates within a period you can finance, such as the first few months, using your own cohort data. Separate new-customer and returning-customer conversions where platforms allow it, so you know which you are paying for. See cohort analysis.

Which should you steer by?#

A worked illustration#

Illustration with simple numbers: two campaigns each spend ₹1,00,000. Campaign A produces 100 orders averaging ₹2,000, so ₹2,00,000 revenue: a CPA of ₹1,000 and a ROAS of 2. Campaign B produces 50 orders averaging ₹5,000, so ₹2,50,000 revenue: a CPA of ₹2,000 and a ROAS of 2.5.

Judged by CPA, Campaign A looks twice as good. Judged by ROAS, Campaign B looks better. Which is right depends on margin. If A's products carry a 50 per cent margin and B's carry 30 per cent, A generates ₹1,00,000 of margin and B ₹75,000. Both just break even or lose money once ad spend is deducted, and A loses less. Add a different margin assumption and the answer flips again.

The lesson is not that one metric is better. It is that neither can be read without the margin beside it. Put margin into your reports, or better still into the conversion values the platforms optimise against.

Reporting both, steering by one#

Most businesses benefit from reporting both metrics while steering by one. A retailer bidding on ROAS should still watch CPA, because a rising cost per order can signal that the system is chasing a few large baskets at the expense of volume. A lead business bidding on CPA should still estimate revenue per lead, because a falling CPA paired with falling deal values is not an improvement. One number drives the bidding; the other keeps it honest.

Both metrics share one flaw#

CPA and ROAS as reported by platforms count conversions the platform attributes to its ads. Some of those would have happened anyway, especially for brand searches and retargeting. A campaign can show an excellent ROAS while adding little. The more mature your spend, the more important it becomes to measure incrementality as well as efficiency. Our guide to incrementality testing explains how, and smart bidding shows how these metrics become bidding targets.

Published by Fabulous.Media, a network of specialist marketing agencies. Updated 9 October 2026. Platform features change often; check current official documentation before acting on platform-specific detail.