How to evaluate advertising beyond ROAS

Return on ad spend (ROAS) divides attributed revenue by advertising cost. It is simple, available in every ad platform, and easy to compare. It is also easy to over-trust, because it leaves out several things that determine whether advertising is actually profitable.

1. Revenue is not profit

A ROAS of 4 means very different things for a product with a 70% gross margin and one with a 25% margin. Before comparing campaigns, convert revenue into contribution margin: revenue minus cost of goods, shipping, payment fees, discounts and expected returns. A break-even ROAS for each product line follows directly from that margin, and it is a far better target than a single account-wide number.

2. Platform-attributed revenue overlaps

Each advertising platform reports the conversions it can claim under its own attribution rules. When a customer sees a Meta ad and later clicks a Google search ad, both platforms may count the same order. Summing platform-reported revenue therefore often exceeds the revenue your store or CRM actually recorded.

A useful habit is to compare total platform-attributed revenue with actual revenue for the same period. The gap won't tell you which platform is right, but it shows how much overlap you are dealing with and stops double-counting from inflating your sense of performance.

3. Not every sale is new

Retargeting and brand-search campaigns often show the highest ROAS because they reach people who were already likely to buy. Some of that revenue would have arrived anyway. Splitting results by new versus returning customers shows how much of your spend is genuinely acquiring customers, and how much is collecting demand that already existed.

4. Incrementality is the real question

The question that matters is: what would have happened without this spend? Incrementality tests (geographic holdouts, platform conversion-lift studies, or carefully timed on/off tests) estimate it directly. They need planning and enough volume to be meaningful, and the results come with confidence intervals rather than certainties. Even an occasional, well-designed test can recalibrate how much weight you give to day-to-day attributed numbers.

5. Customers have a lifetime, not just a first order

For businesses with repeat purchases or subscriptions, the first order rarely tells the full story. Customer acquisition cost (CAC) compared against expected contribution margin over a defined period, and the payback period that implies, is a more durable lens than first-order ROAS. Be conservative with lifetime-value assumptions, especially for young businesses with limited purchase history.

Putting it together

A balanced weekly view might include:

• Spend, platform-attributed revenue and actual revenue, side by side
• Contribution margin after ad spend, by channel
• New-customer share and cost per new customer
• Payback period for recent acquisition cohorts
• Notes on any incrementality evidence and its limits

ROAS remains useful for spotting changes quickly inside a platform. It just shouldn't be the only number deciding where the budget goes.