ROAS to ACoS (and Back): The Formula, a Worked Example, and Where It Breaks

ROAS to ACoS: the conversion

ACoS and ROAS describe the same relationship between spend and sales, just flipped. ACoS divides spend by revenue; ROAS divides revenue by spend. Once you have one, you already have the other — no extra data required.

The two formulas:

• ACoS = Ad Spend ÷ Ad Revenue (shown as a percentage)
• ROAS = Ad Revenue ÷ Ad Spend (shown as a multiple, like 5.0x)

To convert directly: ROAS = 1 ÷ ACoS, and ACoS = 1 ÷ ROAS. Spend $100, generate $500 in attributed sales, and your ACoS is 20% ($100 ÷ $500). Flip it and your ROAS is 5.0 ($500 ÷ $100, or 1 ÷ 0.20). Same campaign, same margin picture — just described from two different angles.

A worked example with real numbers

Round numbers make the formula look simple. Real Amazon data rarely is. Say a campaign spent $2,340 last week and the search term report shows $16,850 in attributed sales. ACoS = 2,340 ÷ 16,850 = 13.9%. ROAS = 16,850 ÷ 2,340 = 7.2x. Both numbers describe the identical result — one campaign, one week, one truth, expressed two ways.

Scale that up and the conversion still holds. Across 30 advertisers we manage, the portfolio delivered 6.04x return on ad spend in July 2026 — measured across the whole book, not pulled from the single best-performing line item. Converted the other way, that's an ACoS of roughly 16.6% (1 ÷ 6.04). Neither figure is more true than the other; it's the same performance read two ways, which is exactly why Amazon's reporting shows both and why most PPC tools let you toggle between them.

Why Amazon gives you both metrics

ACoS came first. It's Amazon's native language for sponsored ads, built for sellers who think in margin: if ACoS is higher than margin, that sale lost money. ROAS is the metric the wider ad industry uses — Google, Meta, DSP platforms — and Amazon has leaned into it more as Sponsored Display and DSP reporting have grown, because a multiple compares more easily across channels that don't all sell physical goods at a fixed margin.

Neither is more correct. Retail-margin sellers running Sponsored Products often think in ACoS because it maps straight to breakeven. Brands running DSP alongside sponsored ads tend to think in ROAS because it lines up with how the rest of their media plan reports. If your team reports one and your agency reports the other, don't ask anyone to switch — convert with the formula above and stop treating them as two different results.

The mistake that makes the conversion look wrong

The formula never breaks. What breaks is comparing an ACoS and a ROAS built from different inputs and assuming the math is lying. Three ways this happens most often:

• Blended vs. attributed revenue. TACoS uses total sales; ACoS and ROAS use only ad-attributed sales. Mixing them into one number produces a figure that won't convert cleanly either direction.
• Mismatched attribution windows. A 7-day click window and a 14-day click window on the same campaign won't convert against each other, because the revenue side changed — not the arithmetic.
• Averaging an average. A blended figure across an entire book can hide very different numbers underneath. Our own portfolio ran a $1.42 blended cost-per-click across 78.4 million impressions at a $4.00 CPM last July — the $0.41 CPC some articles quote for this category is online-video only, not the whole book. Quote the slice, not the average, when someone asks you to defend a number.

We've made this mistake ourselves — reporting a DSP display ACoS next to a Sponsored Products ACoS as if they were the same currency, when one included view-through credit and the other didn't. Both convert to ROAS fine on paper. They just don't mean the same thing.

Your ACoS is too high — or your ROAS won't move. Now what?

Converting the metric doesn't fix the campaign. If ACoS sits above your breakeven (or ROAS below target), the fix is on the spend or targeting side, not the arithmetic. Check, in order: is the attribution window matched to how you're judging it; is match type leaking spend to irrelevant search terms; is the campaign still fighting for a keyword it can't win economically.

Sometimes the real problem is that ACoS and ROAS are the wrong question entirely. Both are last-click, single-channel numbers. Neither can tell you whether a sale would have happened anyway — whether the ad caused the purchase or just showed up in front of someone already buying. That's an incrementality question, and last-click attribution was never built to answer it. Holdout tests and matched-control comparisons can. If ROAS looks fine but sales aren't actually growing, that's usually where to look next — not at the formula.

Where reMKTR fits

reMKTR runs Amazon DSP as a managed service and reconciles it in Amazon Marketing Cloud, so DSP and sponsored ads stop double-counting the same sale before anyone converts a number between ACoS and ROAS. Across 30 advertisers we ran last July, the book carried a blended cost per acquisition of $5.49 across 57,137 attributed purchases, with 20.1% of those going to a shopper new to the brand — detail a single ACoS or ROAS figure can't show on its own. Whether or not that setup is right for your brand, the conversion above is the same one we use internally, and it's yours regardless.