Why ROAS Is Hard to Measure: Attribution and Real Profit

Updated November 2026.

Return on ad spend is conversion value divided by ad cost. Google Ads defines the metric that way, but a dashboard depends on the conversion values and attribution settings supplied to it. See the Google Ads definition. The formula can be correct while the decision drawn from it is wrong.

1. Confirm what counts as a conversion

Separate completed sales from add-to-cart events, form submissions, and assigned lead values. If a lead is worth an estimated $200 in the ad account, a reported 4:1 ROAS does not mean $4 of cash receipts for every $1 spent. Google allows advertisers to assign different values to different conversion actions; document the rule and check it against actual sales. See Google’s conversion-value guidance.

2. Compare attribution settings

One channel may receive the last-click credit while another model distributes credit across earlier interactions. GA4 describes data-driven, paid-and-organic last-click, and Google-paid-channel last-click models. Review GA4’s model definitions. When reports disagree, inspect their models, windows, time zones, and included conversions before adding their totals together. Two platforms can each claim a role in the same sale.

3. Reconcile with orders and refunds

Compare attributed purchase value with the order system for the same period, allowing for reporting delays. Remove cancelled orders and returns where the platform can receive those adjustments. Check tax and shipping treatment. A promotion may lift reported revenue while reducing margin. These are accounting and tracking checks; they do not require inventing an “industry average.”

4. Add the costs the dashboard omits

Ad spend is only one acquisition cost. Creative production, agency fees, discounts, payment processing, delivery, and customer service can change profitability. A campaign can post high ROAS but lose money after fulfillment. Conversely, a lower first-order ROAS might be acceptable when documented repeat purchases and margins justify it. Use actual cohort data before assuming future value.

5. Ask whether ads caused the sales

Attribution assigns credit under a model; it does not by itself prove that a sale would not have happened without the ad. Branded search and remarketing often reach people already familiar with the business. When scale justifies it, a controlled holdout or other incrementality test can estimate the additional effect. For smaller accounts, be cautious about causal language and watch total sales and contribution as budgets change.

Use ROAS as a diagnostic metric with a written definition, then make spending decisions with margin and customer economics. For the benchmark question, see our guide to average ROAS by industry.