What is ROAS? The formula, what a good ROAS is, and what it leaves out

ROAS = revenue credited to ads ÷ ad spend

$4,000 ÷ $1,000 = 4. You'll see the same number written as 4, 4x, 4:1 or 400%. All mean $4 of credited revenue for every $1 of spend.

The formula is simple. What counts as "credited revenue" is not, and each ad platform counts it its own way:

• Meta counts purchases its pixel or Conversions API reports, within the ad set's attribution setting. The standard setting is 7 days after a click and 1 day after a view.
• Google Ads reports it as conversion value ÷ cost. Its default click window is 30 days.
• Amazon Ads counts sales within 7 days of a click for Sponsored Products and 14 days for Sponsored Brands.

So the same sale can raise ROAS in two reports at once, and a sale none of them saw raises none. The guide to attribution models covers why.

ROAS counts revenue, but you keep only your margin. Break-even ROAS is the ROAS at which the margin on credited sales exactly pays for the ads:

break-even ROAS = 1 ÷ margin before ads

Amazon's own guide uses gross margin (price minus cost of goods). That works where product cost is the main cost. On Amazon, referral and FBA fees take a big share too, so use the margin left after fees. With a made-up product:

• Price $30, product cost $7.50. Gross margin 75%, so break-even ROAS looks like 1.33.
• Take out $9.50 of referral and FBA fees and $1 of other costs, and the margin before ads is 40%. The real break-even ROAS is 2.5.

Lower margins push break-even up quickly:

The break-even ROAS calculator works it out from your price, costs and fees.

Amazon puts it plainly: there's no general threshold for a good or bad ROAS, and a good ROAS is one that's profitable for you. A ROAS of 4 is a loss for a product with a 20% margin (break-even 5) and a solid profit for one with a 40% margin (break-even 2.5).

Start from your break-even and set targets per campaign. Running below break-even on purpose can make sense when new customers reorder, because later orders can pay back the first one. The max allowable CAC calculator shows how much a new customer is worth paying for.

These get mixed up, but each answers a different question:

• ROAS: credited revenue per $1 of ad spend. It ignores product costs and fees.
• ROI: net profit per $1 spent, (revenue − cost) ÷ cost. For the ads above, with a 40% margin before ads, $4,000 of sales keeps $1,600. Minus $1,000 of ads leaves $600, an ROI of 60% on the ad spend.
• ACoS: ad spend ÷ credited sales, which is 1 ÷ ROAS. A ROAS of 4 is a ACoS of 25%. Amazon reports ACoS; most other platforms report ROAS.
• Blended ROAS (sometimes called MER): all revenue ÷ all ad spend, across every channel. It avoids arguments about credit but can't say which ads work. The TACoS and blended ROAS calculator works it out.

Reported ROAS counts sales the ads got credit for. Some of those would have happened anyway, and some sales the ads caused never get credited. Incremental ROAS (iROAS) counts only what the ads caused:

iROAS = sales the ads caused ÷ ad spend

See what is iROAS for the full definition, formula and how to measure it on Meta and Amazon.

A made-up retargeting campaign spends $2,000 and is credited with $10,000, a reported ROAS of 5. Most of its audience had already visited and many would have bought anyway. A holdout test finds the ads caused $3,600 of sales: an iROAS of 1.8, below the 2.5 break-even. The campaign that looked best loses money.

It can go the other way. Meta ads that send shoppers to search Amazon are often under-credited, because neither Meta nor Amazon sees the whole path. See Facebook ads for Amazon.

In large experiments at Facebook, researchers found that common non-experimental methods often misjudged what ads did, sometimes by a lot. The reliable answer comes from a test with a holdout group. Read how lift is calculated, and turn a result into numbers with the incremental ROAS calculator.

ROAS is an average across all your spend. Decisions about budget are about the next dollar, and returns usually shrink as you spend more.

A made-up brand spends $10,000 for $30,000 of revenue, a ROAS of 3. It raises spend to $15,000 and revenue rises to $39,000. The average ROAS is still 2.6, which looks fine. But the extra $5,000 only brought in $9,000: a marginal ROAS of 1.8, below the 2.5 break-even.

The marginal ROAS calculator estimates it from your own spend and revenue history.

• Treating ROAS as profit. A ROAS above 1 only means revenue beat ad spend. Compare it with break-even ROAS.
• Comparing ROAS across platforms. Different windows and counting rules make the numbers incompatible.
• Adding up every platform's revenue. Platforms often claim the same sale, so the total can exceed what you sold.
• Scaling on average ROAS. The average can stay healthy while the last dollars lose money.
• Never checking with a test. Reported ROAS can be far from what ads cause, in either direction.