4.2x ROAS. That was the headline number on the monthly report from the agency managing a confectionery brand's Amazon advertising — a brand that's now a Neato partner. The CMO was thrilled. The CFO was confused. Because despite a "4.2x return on ad spend," the Amazon P&L was underwater.
The ROAS wasn't fabricated. It was measuring the wrong thing.
Amazon advertising ROAS has become the most weaponized metric in eCommerce. Agencies use it to justify performance. Brands use it to justify budgets. Nobody asks whether the number actually connects to profit.
Amazon's advertising console reports ROAS as attributed ad revenue divided by ad spend. Every component of that equation is structurally inflated.
Attribution windows inflate ROAS. Amazon's default attribution window for Sponsored Products is 7 days. For Sponsored Brands, 14 days. For DSP, 14 days with a 14-day lookback. A customer clicks your ad Monday and buys Sunday — after seeing organic results, reading reviews, and comparing competitors — and your ad gets 100% credit.
The 7-day window captures approximately 30% more conversions than a 1-day window. Agencies know this. They optimize for click volume knowing the long attribution window catches downstream conversions that may have happened anyway.
Brand search cannibalization. This is the big one. A customer searches "Wiley Wallaby licorice" and they're going to buy Wiley Wallaby licorice. If your agency is running Sponsored Products ads on your own brand name — and they almost certainly are — those ads capture sales that would have happened organically. ROAS on branded terms looks spectacular (10x, 15x, 20x) because you're paying to intercept your own customers.
Incrementality studies consistently show 60–80% of branded search ad conversions would have occurred without the ad. If your agency reports a blended 4x ROAS and 40% of spend is on branded terms with zero incrementality, your actual incremental ROAS on non-branded spend might be 2x or less.
ACoS manipulation through campaign structure. Sophisticated agencies structure campaigns to produce flattering numbers. High-performing branded keywords in a separate "efficient" campaign. Exploratory non-branded spend buried in a "discovery" campaign that doesn't get scrutinized. Report the aggregate — efficient branded campaigns blend with inefficient growth campaigns into something that looks reasonable.
Revenue attribution without margin consideration. A 4x ROAS means $4 in revenue for every $1 in ad spend. But after Amazon's referral fee (15%), FBA fees ($3–5 per unit for CPG), COGS, and the ad spend itself — what's left?
Component | Amount |
|---|---|
Retail price | $25.00 |
Amazon referral fee (15%) | -$3.75 |
FBA fee | -$4.50 |
COGS | -$8.00 |
Ad spend (at 4x ROAS = 25% ACoS) | -$6.25 |
Net margin | $2.50 (10%) |
That 10% looks okay — until storage fees, returns (8–12% in CPG), and agency management fees enter the picture. Suddenly you're at 3–5% or negative.
What Honest Reporting Looks Like
Incrementality Reporting
Separate branded and non-branded performance completely. Then go further: run incrementality tests. Amazon offers Brand Lift studies for DSP. For Sponsored Products, run a pause test — turn off branded ads for two weeks and measure total revenue impact (not just ad-attributed revenue).
When we onboarded that confectionery brand at Neato, we ran this test in the first 30 days. Pausing branded Sponsored Products reduced total sales by 8%, not the 40% the previous agency's reporting implied. We reallocated 70% of that branded budget to non-branded conquest campaigns and grew total revenue 14% in the following quarter.
TACoS (Total Advertising Cost of Sale)
TACoS = total ad spend / total revenue (not just ad-attributed revenue). Far more honest because it shows what percentage of your entire business you're spending on ads.
Healthy TACoS for CPG on Amazon: 8–12%. Concerning: 15–20%. Alarm bells: 20%+.
If your agency doesn't report TACoS, ask yourself why.
Contribution Margin After Advertising
The only number that ultimately matters: what profit does your Amazon business generate after all costs including advertising? If ROAS goes up but contribution margin is flat or declining, advertising isn't working — it's getting more expensive.
New-to-Brand Metrics
Amazon reports "new-to-brand" in Sponsored Brands and DSP. This shows what percentage of ad-driven sales come from customers who haven't purchased your brand in 12 months. An agency driving 5x ROAS from existing customers is running a very different operation than one driving 3x ROAS with 60% new-to-brand.
Five Questions Your Agency Doesn't Want You to Ask
1. "What percentage of our spend goes to branded vs. non-branded keywords?" More than 25% branded? You're overpaying for customers you already own.
2. "What's our estimated incrementality on branded campaigns?" If they can't answer with data, they haven't measured it.
3. "Show me TACoS trended over 12 months." TACoS rising while revenue is flat = advertising becoming less efficient.
4. "What's our contribution margin per unit after advertising?" If they can't connect ad performance to unit economics, they're optimizing a number disconnected from your business objective.
5. "What would happen to total revenue if we cut ad spend by 30%?" The question agencies fear most. For many brands, the honest answer is: total revenue would drop less than 30%.
Building an Honest Practice
Action 1: Demand a full campaign audit. Branded vs. non-branded breakdowns. Automatic vs. manual campaign performance. Top-of-search vs. rest-of-search placement data. The granularity reveals where money actually works.
Action 2: Run an incrementality test. Pause branded advertising for 14 days. Measure total revenue impact. The gap between "ad revenue lost" and "total revenue lost" is your cannibalization rate. It's usually eye-opening.
Action 3: Shift your primary KPI from ROAS to contribution margin. Build a SKU-level P&L with all costs — COGS, referral fees, FBA fees, advertising, returns. Optimize against this, not ROAS in isolation.
At Neato, our performance marketing team manages advertising as part of the integrated 2P operation — ad efficiency is directly tied to our own profitability. When we overspend on advertising, it comes out of our margin, not yours. That structural alignment produces very different decisions than an agency model where spending more means earning more in management fees.
The Uncomfortable Truth
Amazon advertising is not optional for CPG brands. Organic rank depends on sales velocity, and advertising drives velocity. You have to play.
But the way most brands play — delegating to an agency, accepting inflated ROAS as proof, never questioning incrementality — is leaving money on the table. Or more accurately, handing it to Amazon.
Amazon's advertising revenue hit $56.2 billion in 2024. Every dollar came from a brand. The brands that win aren't spending the most. They're the ones who know exactly what each dollar buys — and what it doesn't.
Your ROAS number is technically accurate. But technical accuracy in the service of strategic blindness is its own kind of deception. The brands that figure this out don't just advertise better. They profit.
Stacey Silva is Senior Brand Manager at Neato, where she manages brand performance and advertising strategy for CPG brands on Amazon. Neato is a 2P eCommerce acceleration partner — we buy inventory, become seller of record, and grow brands on Amazon with certainty.
