Return on Ad Spend (ROAS)

How ROAS is calculated

The formula is simple: attributed revenue ÷ ad spend. The complexity hides in the word “attributed.” Spend is a fact the ad platform bills you for. Revenue attributed to that spend depends on three measurement choices: which tool records the conversion, which attribution model assigns it to a channel, and which attribution window bounds the claim. Change any of the three and the same campaign produces a different ROAS — which is why Google Ads, Meta and your analytics rarely agree on the number.

Why incomplete data understates ROAS

The asymmetry is the problem. The denominator (spend) is always complete. The numerator (measured revenue) is complete only if the analytics tool observed every conversion. In the EU, cookie-based analytics does not come close: consent rejection costs 15–60% of visitors depending on sector, brand strength and traffic mix, ad blockers remove more, and browser restrictions cut cookie lifetimes. In the compounded worst case, a tool like GA4 sees as little as 13% of real EU traffic; on a real Shopify store measured over 48 days, GA4 did not record 29% of visits. Either way, a ROAS computed on that fragment divides full spend by partial revenue.

The distortion is not evenly distributed either. Consent rejection and blocker usage vary by market, device and audience, so some campaigns lose more measured conversions than others. A prospecting campaign reaching privacy-conscious German desktop users can look far worse than a retargeting campaign reaching returning mobile buyers, even when their true returns are similar. Budget then flows toward the campaigns that are easiest to measure, not the ones that perform best. Complete data — revenue attribution computed on observed orders without consent gaps rather than on the consenting minority — removes that bias from the numerator.

What ROAS does not tell you

ROAS is a revenue ratio, not a profit ratio. It ignores margin, returns, shipping and the cost of goods — a ROAS of 4 on a 20%-margin product loses money. It is also blind to incrementality: it credits revenue that may have arrived anyway, which is why branded search campaigns post spectacular ratios. And because it measures a single transaction, it says nothing about customer lifetime value — a campaign with mediocre first-order ROAS can still be your best acquisition channel if those customers reorder. Treat ROAS as a comparative efficiency signal between campaigns measured the same way, not as a verdict on profitability.