The Ecomm Analyst

Standard ROAS is calculated on revenue, and every ad platform reports it that way. But revenue-based ROAS tells you nothing about whether an order made money, which is why I keep a profit-based version alongside it. The practical answer is to run both, using revenue ROAS for comparability and profit ROAS for decisions about scaling.

This is a different question from whether to use ROAS or MER. That one is about which denominator you trust. This one is about what goes in the numerator, and it changes which campaigns look like winners.

Why the numerator matters more than people expect

Revenue ROAS treats every dollar of revenue as identical. Profit ROAS does not, and on any store with a mixed catalog the two rankings diverge fast.

A campaign selling a $120 bundle at 55 percent margin and a campaign selling a $40 entry product at 25 percent margin can both post a 3.0x revenue ROAS. On profit, the first returns $1.65 of margin per ad dollar and the second returns $0.75. Same headline number, and one is more than twice as good. If you allocate on revenue ROAS you will fund both equally, and the mix will drift toward whichever product converts most easily rather than whichever product pays the bills.

This shows up hardest on stores running heavy discounting or a wide price ladder. I have seen a Meta account where the top three campaigns by revenue ROAS were the bottom three by contribution, because they were all pushing a loss-leading starter kit.

The case for keeping revenue ROAS

Revenue ROAS is not wrong, it is just incomplete. It has three genuine advantages worth preserving.

• It is what the ad platforms optimize toward, so it is the number their bidding algorithms actually respond to
• It is comparable across brands and over time, since it does not depend on your internal cost assumptions
• It updates in real time, while margin data usually lags behind COGS updates and returns settling

Profit ROAS is only as good as your cost data. If your COGS table is eight months stale or your returns allowance is a guess, a profit ROAS carries false precision. A wrong number with a decimal point is more dangerous than a rough number you know is rough.

How I run both

Revenue ROAS at the ad set and creative level, because that is where I am comparing like against like and where the platform’s own optimization lives. Profit ROAS at the campaign and channel level, because that is where budget allocation happens and where product mix actually varies.

Getting profit ROAS to a usable state means attributing orders at the SKU level rather than the order level, so margin can be applied per item. ThoughtMetric, which sponsors this blog, handles product-level attribution by SKU, which is what makes the margin overlay possible without exporting everything to a spreadsheet. It prices on pageviews starting at $99 per month with every feature at every tier, and offers a two week free trial.

The variant nobody agrees on

There is a third version some operators use, sometimes called POAS, where you divide contribution profit by ad spend and treat 1.0x as break-even. It is cleaner conceptually, since anything above 1.0 makes money and you no longer need a separate break-even calculation.

I like the logic and rarely use it, for a boring reason. Nobody else reads it the same way. Tell a media buyer the campaign is at 1.3 and they assume disaster before you explain the denominator. If your whole team has adopted it, it is the better metric. If you are the only one, you will spend more time explaining it than acting on it.

Whichever version you pick, the discipline that matters is comparing it to the right threshold. I went through how those thresholds interact with the broader question of which metric should govern the business in MER vs ROAS.

Getting the cost data honest first

Before switching anything to a profit basis, audit the inputs. Pull your COGS table and check the date it was last updated. Check whether landed cost includes freight and duty or just the unit price from the supplier. Check whether your returns rate is measured or estimated, and whether refunds are being deducted from attributed revenue at all.

On most stores I look at, at least one of those four is wrong, and usually it is the returns handling. Refunds settle weeks after the order, so a profit ROAS on a seven day window overstates performance on anything with a high return rate. Fix the inputs, then change the metric.

Common questions

Does ROAS include COGS?

Standard ROAS does not. It divides attributed revenue by ad spend, with no cost of goods subtracted. Profit-based variants subtract COGS and other variable costs from the numerator.

Is profit ROAS better than revenue ROAS?

It is better for allocation decisions across campaigns with different product mixes, and worse when your cost data is unreliable or stale. Most stores benefit from running both rather than replacing one with the other.

Should ROAS use revenue before or after discounts?

After discounts and after returns. Revenue that gets refunded or was never collected at list price is not revenue your ads earned, and including it inflates the figure most on exactly the promotions where you can least afford the error.