Measurement
ROAS depends on attribution — a model deciding which revenue counts as "from" advertising. MER skips that decision entirely. That difference is why they move differently, and why neither one alone tells the whole story.
Return on ad spend takes revenue an attribution model has credited to advertising and divides it by what was spent to generate it. Because it relies on attribution, ROAS is sensitive to every choice baked into that attribution: which model, which window, which platform. Two stores with identical sales and spend can report meaningfully different ROAS if their attribution settings differ.
Marketing efficiency ratio is simpler by design: total store revenue for a period, divided by total marketing spend for that same period. No individual order is credited to any specific ad. It doesn't try to answer "which campaign caused this sale" — it answers "given what I spent on marketing overall, how much did the business bring in." That makes it far steadier than platform ROAS, since it isn't exposed to attribution-window changes, tracking gaps, or cross-platform double-counting.
ROAS is more useful at the campaign or ad-set level, where you need enough granularity to compare one creative or audience against another — MER doesn't break down that far. MER is more useful at the business level, for questions like "is our overall marketing spend sustainable relative to revenue," where attribution noise would otherwise make the trend hard to read. Neither replaces the other; they're suited to different altitudes of the same decision.
Neither ROAS nor MER proves incrementality — that the ad spend caused sales that wouldn't have happened anyway. A high MER can coexist with a business that would have sold nearly as much through organic and repeat demand alone. Both metrics measure efficiency against spend, not the counterfactual of what would have happened without it.
Connect your store and ad accounts to see both metrics calculated from the same underlying data.