What Is Marginal ROAS in Ecommerce (And Why Average ROAS Is Lying to You)
by Trivas.ai
|
7 min read
Oct 06, 2026
Your Average ROAS Looks Fine. Your Next Dollar of Spend Might Not Be.
Blended ROAS is sitting at 3.5x. Dashboard's green across the board. So you push budget up 20% heading into a big push, expecting more of the same.
Instead, profit shrinks.
This happens constantly, and it's not a tracking bug. It's a math problem. Your average ROAS blends every dollar you've ever spent into one number, and that number can look completely healthy while the newest dollars you're adding are barely breaking even. Average ROAS tells you what already happened. It says nothing about what the next dollar will return, which is the only question that matters when you're deciding whether to scale.
That gap between "what happened" and "what will happen next" is exactly what marginal ROAS is built to answer. This piece covers what is marginal ROAS in ecommerce terms, the formula behind it, how it diverges from average ROAS, and how to actually put it to work in a budget decision instead of just nodding along with the concept.
Quick Refresher: What Average ROAS Actually Measures
Standard ROAS is simple: total revenue divided by total ad spend, over whatever period you're looking at. Spend $10,000, generate $35,000 in attributed revenue, and your ROAS is 3.5x.
That number is a blend. It covers every dollar you spent in that window, mashed into a single average, looking backward at a period that's already closed. It doesn't distinguish between the first dollar you spent, which probably went to your warmest audience and converted easily, and the thousandth dollar, which went somewhere far less efficient.
That's exactly why average ROAS is a poor tool for the one decision people keep using it for: should I spend more? A good average can sit on top of a terrible marginal trend for weeks before it catches up. By the time the blended number drops enough to notice, you've already overspent into the unprofitable zone. If you're trying to answer "what is marginal ROAS ecommerce teams should be watching instead," this is the gap it fills: it looks at the edge of your spend, not the middle of it.
What Is Marginal ROAS, Exactly
Marginal ROAS is the return generated by the next incremental unit of ad spend, not the return on everything you've spent so far. Think of it as the ROAS on just the last $1,000 you added to a campaign, isolated from the dollars that came before it.
The formula:
Marginal ROAS = Change in Revenue / Change in Ad Spend
You need two spend levels to calculate it, not one period of totals.
Here's a concrete version. Spend goes from $10,000 to $12,000 in a week. Revenue goes from $35,000 to $38,000. Your average ROAS still reads a comfortable 3.1x ($38,000 / $12,000). But look at just the increment: you added $2,000 in spend and got $3,000 in additional revenue back. That's a marginal ROAS of 1.5x on the new money.
The $10,000 baseline is still performing great. The $2,000 you layered on top isn't. Average ROAS hides that split completely. Marginal ROAS is the only one of the two that tells you whether scaling further is a good idea.
Why Marginal ROAS Drops Even When Average ROAS Looks Healthy
This isn't random. It's diminishing returns, and it happens in a predictable order.
Your first dollars of ad spend reach your warmest, cheapest-to-convert audience segments: retargeting pools, lookalikes built off your best customers, branded search. As you add budget, the algorithm has to reach further, into colder segments that take more impressions and more cost to convert. Each additional dollar is chasing a worse prospect than the one before it.
Auction dynamics make this worse. On Meta and Google specifically, pushing more budget into the same campaign or audience bids you into a smaller remaining pool of available impressions, and CPMs climb as a direct result. You're not just reaching colder people, you're paying more to reach them.
The part that actually burns budgets: average ROAS lags this shift by design, because it's still diluted by all those efficient early dollars. A campaign can be bleeding on its marginal spend for two or three reporting cycles before the blended number moves enough to set off alarms. By then the damage is in your margin, not your dashboard.
Marginal ROAS vs Average ROAS: Side-by-Side
Both numbers are legitimate. They just answer different questions, and mixing them up is where budget decisions go wrong.
[@portabletext/react] Unknown block type "table", specify a component for it in the `components.types` prop
Average ROAS belongs in the monthly recap. Marginal ROAS belongs in the Monday morning meeting where you decide whether to raise a budget cap.
How to Calculate and Track Marginal ROAS Without a Data Science Team
You don't need a modeling team to get a usable read on this. The manual version: compare week-over-week spend and revenue, or run a basic test-vs-control split, and look at the delta per channel or campaign. Spend up $3,000 this week versus last, revenue up $4,500, that's a 1.5x marginal read on the increment. Do it every week per campaign and you'll start to see the curve bend.
Where it falls apart is scale. Once you're running this across dozens of campaigns and five or six platforms (Amazon, Meta, Google, TikTok), each with a different attribution window and its own definition of "revenue," the manual version turns into a spreadsheet nobody trusts. Attribution windows don't line up, so a delta that looks marginal might actually be lag from last week's spend showing up late.
This is the actual argument for a unified data layer. When spend and revenue from every channel get normalized against the same clock, on something like a Redshift-based reporting setup, you can build a real marginal ROAS curve per channel instead of eyeballing blended numbers and guessing where the bend is. If you want to see how the underlying metrics get defined and reconciled across channels before you start comparing them, the data dictionary is a reasonable place to check your assumptions.
Using Marginal ROAS to Make Real Budget Decisions
Once you can see marginal ROAS per channel, the decision rule is straightforward: stop scaling a channel once its marginal ROAS drops below your breakeven ROAS, the point where it's no longer covering margin plus fixed costs. Not your average ROAS target. Your breakeven.
Say Meta prospecting is showing a marginal ROAS of 1.2x on the next $5,000, barely above water. Meanwhile your Google Shopping campaign is sitting at a marginal ROAS of 4x and clearly has room before it saturates. The move isn't to cut Meta entirely, it's to redirect that next $5,000 to the channel still returning real margin. That's the whole decision, every week, repeated across channels.
The harder version of this is doing it before you spend, not after. Forecasting and simulation tools can model what a channel's marginal ROAS curve is likely to look like at higher spend levels, using the response curve you've already got, so you're not finding out the ceiling exists by reading last week's invoice. If budget planning is part of your role, forecasting and simulation tooling built specifically for this is worth a look, especially if you're the one answerable for next quarter's spend efficiency, which is exactly the problem performance marketers run into every planning cycle.
Start Measuring the Metric That Actually Guides Spend Decisions
Average ROAS tells you what already happened. Marginal ROAS tells you what to do next. Both are useful, but only one of them should be driving your next budget increase.
If you haven't checked your current blended number recently, the ROAS calculator is a quick way to sanity-check where you stand before layering marginal analysis on top of it.
And if you're tired of stitching together spend and revenue data from five platforms just to get a rough sense of where marginal ROAS is heading, that's worth a second look too. Unifying that data is most of the actual work. Once it's in one place, the marginal ROAS curve stops being a once-a-quarter spreadsheet exercise and starts being something you can glance at on a Tuesday.
Content author and contributor at Trivas.ai, sharing insights on e-commerce analytics, business intelligence, and data-driven strategies to help businesses grow.
Continue Reading
explore more insights
Ecommerce Analytics You Can Trust: Trivas.ai and SOC2 Compliance
3 min read
Trivas Demo: What to Expect and How to Prepare
3 min read
Pruning Plan: Retiring 'Common Challenges and Solutions in Ecommerce' and Merging It Into a Pillar Page
.d53b12e5.png&w=3840&q=75 "Trivas.ai")
