Break-even ROAS and MER for Amazon brands: the formulas, and where they mislead

Break-even ROAS is 1 divided by your contribution margin before ad spend, written as a decimal: a sale that keeps 40% of its revenue can spend 40 cents per sales dollar on ads, so its break-even ROAS is 2.5. If you sell on Amazon and Shopify, you need a separate break-even for each channel, because the fees are different.

Most guides to ROAS and MER assume one Shopify store with one margin, where every sale lands in the store. A brand that sells mostly on Amazon breaks all three of those assumptions. Below are the formulas, a worked example with illustrative numbers, and the traps that make a losing campaign look fine.

ROAS, ACoS, TACoS and MER in plain words

These four ratios all divide sales by spend, or spend by sales. They differ in whose sales and whose spend.

• ROAS, return on ad spend, is sales divided by ad spend, as the ad platform counts sales. Meta defines purchase ROAS as purchase conversion value divided by amount spent, using purchases it credits to your ads.
• ACoS, advertising cost of sales, is Amazon’s version turned upside down: ad spend divided by ad sales. Amazon’s own guide gives both formulas and calls one the inverse of the other. An ACoS of 25% is a ROAS of 4.
• TACoS, total advertising cost of sales, is Amazon ad spend divided by all Amazon sales, including sales no ad touched.
• MER, marketing efficiency ratio, is all revenue divided by all marketing spend. It ignores what each platform claims. Some teams call it blended ROAS.

ROAS and ACoS count credited sales, not caused sales. A shopper who would have bought anyway still counts if they clicked an ad first. That makes them good for comparing one campaign with another on the same platform, and weak for deciding whether the whole budget pays.

The break-even ROAS formula

Break-even ROAS = 1 ÷ contribution margin before ad spend

Start from net revenue: what the buyer pays, after discounts and expected returns. Contribution margin before ad spend is what is left after product cost, fulfillment, storage, payment processing and marketplace fees. Our contribution margin guide calls this CM2 and builds it line by line.

On Amazon, the same limit is a break-even ACoS, and it equals the margin itself. Amazon’s guide puts it simply: to make a profit, ACoS “needs to be lower than your profit margin.”

One adjustment matters. Ad platforms credit sales at something close to the price paid, before returns. If your margin is measured on net revenue after returns, raise your break-even ROAS to match. Divide it by net revenue as a share of those credited sales. Otherwise every campaign looks slightly better than it is.

One product, two channels, two break-evens

Here is one body lotion sold on a Shopify store and on Amazon through FBA (Fulfillment by Amazon, where Amazon stores and ships the stock). All figures are illustrative numbers per unit, not any brand’s results.

Where the fee lines come from:

• Referral fee. Amazon charges a share of the sale price. In Beauty, Health and Personal Care it is 15% for items over $10. Here that is 15% of the $22.25 the buyer pays.
• Payment processing. Shopify Payments charges 2.9% plus 30 cents per online card sale on the Basic plan. On $22.25 charged that is about $0.95.
• Returns allowance. The expected cost of returns, taken off when the sale happens, as in the contribution margin guide.
• Fulfillment. Both figures are examples, not quotes. Check your own FBA fee in Amazon’s Revenue Calculator and your own warehouse invoice.

The two channels bring in the same net revenue per unit. Amazon’s 15% cut is most of the difference. It moves break-even ROAS from 2.10 to 2.75.

What one blended margin does

Now say Amazon is 75% of this brand’s revenue. Weight the two margins by revenue and you get a blended margin of 39.2%. That gives a blended break-even ROAS of 2.55, and the team applies it to every campaign.

An Amazon campaign runs at a ROAS of 2.60, an ACoS of 38.5%. Against 2.55 it passes. Against Amazon’s real break-even of 2.75 it fails.

• Every $1,000 of spend brings $2,600 of sales.
• At Amazon’s 36.4% margin, those sales leave about $945.
• The campaign loses about $55 per $1,000 spent, while the dashboard shows it earning.

The error runs the other way on the store. A Shopify campaign at ROAS 2.30 looks like a loser against 2.55. At the store’s 47.7% margin it earns about $98 per $1,000. A blended target cuts the winner and keeps the loser.

Gross margin gives an even worse answer

Some teams skip fees and use gross margin, which is net revenue minus product cost. Our contribution margin guide calls this CM1. For the lotion on Amazon that is 77.3%, for a break-even ROAS of 1.29. A campaign at ROAS 2.0 would look like a clear winner. At the real 36.4% margin, it loses about $270 per $1,000 spent.

Product families differ too

Now add a travel-size lotion that sells for $9 on Amazon. Items at $10 or less pay an 8% referral fee in that category. But the FBA fee takes a much bigger share of a $9 price. Say the unit costs $2.40 to make and land, the FBA fee is $3.00 and storage is $0.18. After the $0.72 referral fee, the travel size keeps $2.70, or 30%, before ads. Its break-even ROAS is 3.33, against 2.75 for the full size.

Put both sizes in one campaign and read it against one target, and the small size quietly loses money. Group listings into product families and set a break-even for each family on each channel.

MER when Meta sells on Amazon

MER avoids the tracking arguments. It still goes wrong in a specific way for Amazon brands.

Take one closed month, with illustrative numbers and the same margins as the lotion. The brand’s Meta ads mostly send shoppers to its Amazon listings.

Three ways to read the same month:

1. Store-only MER. Store revenue divided by Meta and Google spend is $88,000 ÷ $100,000, or 0.88. It looks like marketing loses money on every dollar. But most of the Meta spend sold on Amazon. Meta’s pixel, its tracking code on your store, cannot see an Amazon checkout, so the store view never sees those sales.
2. Whole-business MER. All revenue divided by all spend is $418,000 ÷ $145,000, or 2.88. Held against the store’s break-even of 2.10, that looks comfortable. But about four fifths of that revenue is Amazon revenue, which carries Amazon’s fees.
3. MER on margin. Contribution before ads is $120,000 from Amazon at 36.4%, plus $42,000 from the store at 47.7%. That is $162,000. Divide by $145,000 of spend and you get 1.12.

MER on margin has a simple break-even: 1.0. At 1.12, each marketing dollar brings back $1.12 of contribution. The month leaves $17,000 toward salaries, rent and profit. That is a thin month, not a comfortable one. Neither revenue-based MER showed it.

If this brand is enrolled in Amazon’s Brand Referral Bonus, the picture improves a little. Amazon says brands can earn a bonus averaging 10% of qualifying sales that come from outside traffic, paid as a credit against referral fees. Count that credit in the Amazon margin on those sales, once the credit actually arrives.

For how to measure what Meta sells on Amazon in the first place, see do Meta ads drive Amazon sales.

The metrics side by side

Where it goes wrong

These are the traps we see in real multi-channel data.

Judging last week’s ads on sales still being counted

Amazon keeps crediting sales to an ad click for days after the click. Amazon Attribution, which tracks outside traffic to Amazon, uses a 14-day window. So last week always looks worse than the week before it, even when nothing changed. Compare weeks at the same age, or wait until the window has closed. More in Amazon ad sales numbers are not final.

Adding up every platform’s credited sales

Meta, Google and Amazon each credit sales to their own ads. Two platforms can claim the same order. Add their credited sales together and you often get more than the business actually sold. That is one reason MER uses total revenue instead.

Missing spend

MER is only as honest as its denominator. Agency fees paid as a share of spend, creator payments and affiliate commissions are all marketing that scales with sales. Leave them out and MER looks better than it is.

An open month next to a closed one

On September 15, half of September’s revenue is in, and most of its refunds and fees have not posted. Put that next to a closed August and the ratios move for reasons that have nothing to do with marketing. Use closed months for MER, and label anything current as “so far.”

How to set it up

1. Build contribution margin before ads for each channel. Use settled fees: the Amazon settlement report, Amazon’s statement of what it paid you, and Shopify payouts. Start from the contribution margin guide.
2. Split it by product family. Map every Amazon listing and store product to a family. Set a break-even ROAS and ACoS for each family on each channel.
3. Tag every campaign by destination. Amazon or store, whatever platform sold the ad. Judge each campaign against the break-even of the channel where its sales land.
4. Wait for sales to finish counting. Read a week only once its attribution window has closed. Or compare two weeks at the same age.
5. Report MER monthly, on all revenue and all spend. Include Amazon ads, agency fees and creator pay. Use closed months.
6. Put MER on margin next to it. Contribution before ads divided by all marketing spend. Above 1.0, marketing pays for itself. Below it, it does not.
7. Give each number its job. ROAS ranks campaigns inside one platform. Break-even ROAS sets the floor for each campaign. MER on margin decides whether the total budget should go up or down.

Break-even is a floor, not a goal. A campaign right at break-even adds nothing toward fixed costs. It can still be worth running if its new customers buy again, but that is a lifetime value decision, made on purpose.

Synthesis is built for this kind of question. Whether an ad sends people to Amazon or to the store is written down once, and every answer uses it. Every number carries its revenue basis, its period and caveats such as “Meta’s pixel cannot see Amazon checkouts.” ROAS is computed fresh each time, never stored as a fixed fact.