A Falling Amazon ROAS Is Not a Verdict. It Is Arithmetic.

Amazon ROAS measures the mix of your ad spend before it measures how well that spend performs. Shift budget from branded to non-branded and competitor targeting, and ROAS falls by construction, even when execution improves and more of your buyers are genuinely new. The decline is the arithmetic of the decision. The real question is whether it is the good kind or the bad kind.

• Branded campaigns harvest demand you already own, which is why their ROAS is high and their contribution to growth is low.
• Moving budget toward acquisition lowers blended ROAS mechanically, with no change in execution.
• A good ROAS on Amazon is set by your margin, not by a category benchmark.
• A good decline comes with a rising new-to-brand share, a rising non-branded share of spend and stable or growing total revenue.
• If nobody agreed in advance which metric would be allowed to move, the reallocation usually gets reversed.

Amazon ROAS Measures Your Spend Mix Before It Measures Performance

Amazon ROAS is attributed sales divided by ad spend, and the two halves of that ratio depend first on where the money goes. A campaign on your own brand name reaches shoppers who were already looking for you. They convert well, so attributed sales per dollar are high. A campaign on generic category terms reaches shoppers who have not chosen yet. They convert less often, so attributed sales per dollar are lower.

Neither number says much about execution on its own. A branded campaign with a 10:1 ROAS can contribute almost nothing to growth, because a large share of those orders would have come through organic results anyway. A non-branded campaign at 3:1 can be the only part of the account that brings in customers who did not know the brand.

That is why ROAS behaves as a mix metric. It moves whenever the composition of spend changes, independently of how well anything is run.

Moving Budget to Non-Branded Lowers Amazon ROAS by Construction

The effect is easy to demonstrate with round, illustrative figures.

• Before: $100,000 a month, split $60,000 on branded terms at 10:1 and $40,000 on non-branded terms at 3:1. Attributed sales: $720,000. Blended ROAS: 7.2.
• After: $20,000 moves from branded to non-branded, with both segments performing exactly as before. Branded now returns $400,000 and non-branded $180,000. Attributed sales: $580,000. Blended ROAS: 5.8.

That is a 19% fall in reported ROAS with no bid, keyword or creative getting worse. The reported drop also overstates the real loss: part of the $200,000 of branded sales that stopped being attributed to ads keeps arriving through organic listings, because those shoppers were searching for the brand anyway.

The dashboard shows a decline in efficiency. The business may have gained customers. Only other metrics can tell which is true.

What Is a Good ROAS on Amazon? The One Your Margin Allows

There is no universal answer to what is a good roas on amazon, and any category benchmark hides the only variable that matters: your margin.

The useful reference is break-even ROAS, equal to 1 divided by contribution margin before advertising. A product with a 25% margin breaks even at a ROAS of 4. Below it, each attributed sale loses money on a first-order basis. A product with a 50% margin breaks even at 2. The same ROAS of 3 is profitable on the second product and loss-making on the first.

Two consequences follow. A blended target applied across a catalogue with different margins pushes spend toward the wrong products. And a ROAS below break-even can still be a sound decision on acquisition campaigns, if the new customers come back. That judgement depends on repeat purchase, not on the ad report.

ACoS vs ROAS vs TACoS: Which One to Judge a Reallocation On

ACoS and ROAS are the same measurement expressed two ways. ACoS is ad spend divided by attributed sales, so it is the inverse of ROAS: a ROAS of 5 is an ACoS of 20%. On the acos vs roas question, choosing one changes the reading convention, not the information.

Both share the same limit. They only count sales that Amazon attributes to an ad, so both move with the spend mix.

TACoS, total advertising cost of sale, divides ad spend by total revenue, organic included. For judging a reallocation, TACoS is the more honest of the three. ROAS reads campaigns, while TACoS Amazon-wide reads the business, because it captures what the other two miss: whether total sales grow when ad spend moves. A reallocation that lowers ROAS while TACoS stays flat and revenue rises is doing its job.

Three Checks Separate a Good Amazon ROAS Decline From a Bad One

A good decline and a bad one look identical in the ROAS column. They separate on 3 other figures, all available in standard Amazon reporting.

1. New-to-brand share of attributed orders. If the reallocation is working, the share of orders from customers who had not bought the brand in the previous year rises. If it stays flat, the new spend is reaching the same shoppers by another route.
2. Non-branded share of spend. It should rise in line with the decision. If it does not, the ROAS drop has another cause and needs a different diagnosis.
3. Total revenue, organic included. It should hold or grow over the following weeks. Falling total revenue alongside a falling ROAS is not a mix effect.

A good decline shows all 3 moving the right way. A bad decline shows a flat new-to-brand share and flat or falling revenue. That is inefficiency, and it should be treated as such.

The Bad Decline Looks Exactly the Same on the Dashboard

A falling ROAS is often exactly what it looks like. Rising cost per click in a contested category, a listing that lost its rating, a stockout on a hero product or a competitor undercutting on price all lower ROAS without any change in the spend mix.

The 3 checks exist to reach that conclusion as readily as the reassuring one. If the new-to-brand share has not moved and revenue is flat, the honest reading is that the account is paying more for the same customers. Explaining that decline as a strategic mix shift would be a permission slip, and a leadership team will read it as one.

Amazon Advertising ROAS Needs a Decision Rule, Not a Target

Amazon ROAS reflects the spend mix before performance, and a reallocation toward acquisition lowers it by construction. A good decline is confirmed by a rising new-to-brand share, a rising non-branded share of spend and stable or growing total revenue. A bad one shows none of the three.

The fix sits upstream of reporting. Before any reallocation, agree which metric is allowed to move, by how much, and which one must not. Without that agreement, amazon advertising roas becomes the argument two months in, and budget drifts back to branded terms whatever the growth did.

Setting that rule means reading attributed and total sales, margin and new-customer data together at product level. That is the consolidated view Primelis Market is built to produce across a portfolio. It supports more than 100 brands and has contributed to €1.2 billion in revenue generated on Amazon. The useful first step is a short one: write down the expected ROAS range for your next reallocation before it starts.