Return on advertising spend, or ROAS, measures the revenue generated for every dollar spent on advertising. The formula is straightforward: ROAS equals sales attributed to ads divided by the cost of those ads. While useful for comparing campaigns and allocating budgets, ROAS has limitations when attribution models misallocate credit.
The Attribution Problem
Mike Murphy, vice president of marketing at Incremental, explains that ROAS becomes misleading when it assigns too much or too little value to an ad, or overlooks organic purchase intent. For example, an ecommerce company spending $10,000 on retail media might report a 5:1 ROAS using a last-touch attribution model. This model credits the final ad interaction before a sale, regardless of whether that ad influenced the purchase.
That approach creates a significant gap. ROAS may attribute revenue to organic sales that would have occurred without the ad, or to sales driven by earlier touchpoints. It also fails to track sales influenced by ads in ways that are not directly measurable. This is especially problematic in retail media, where buyers often have high intent to purchase. A shopper might click an ad and then buy the same product organically, or see an ad on one device and complete the purchase on another.
Measuring Incrementality vs. Attribution
The core issue lies in distinguishing attribution from incrementality. Attribution assigns credit to the last ad touchpoint, while incrementality asks whether the ad actually generated additional revenue. In the $10,000 retail media example, the ad might not always be visible. When it appears alongside an organic product listing, it only gets credit for incremental sales, lowering the ROAS to 3:1. The organic results likely drove sales without the ad, highlighting the difference between attributed and incremental performance.
Related Post: Digital receipts reshape retail advertising
Conversely, an ad may produce revenue that ROAS does not track. If a shopper sees an ad on Amazon but purchases later on the merchant’s website, or switches devices during the buying process, the ad’s contribution may go undetected. Google Ads uses conversion modeling to account for such cross-device and privacy-restricted scenarios, noting that without modeling, reported conversions represent only a fraction of actual performance. This leads to underreported ROAS when attribution systems fail to recognize a sale’s true cause.
Marketers can test for accuracy by comparing attributed sales with incremental ones. For smaller budgets, a holdout test—pausing ads for a product group while continuing spend on others—provides directional insights. Larger advertisers with retail media partnerships may access randomized or geographic tests unavailable in self-service platforms. The goal is to ensure that increased ad spending correlates with higher sales and profits, while reduced spending leads to the opposite.
Beyond ROAS: Broader Metrics
While ROAS offers a quick snapshot, other metrics provide deeper context. The marketing efficiency ratio compares total marketing spend to total revenue. Contribution margin, calculated as gross profit minus variable costs, reveals profitability. Customer acquisition cost and lifetime value help determine what to spend to gain a first-time buyer. Yet, as Murphy emphasizes, the ultimate measure remains the bottom line.
