ROAS Looks Good, But Is It Driving Growth?
▼ Summary
– High reported ROAS can be misleading, as it often captures conversions that would have occurred through organic or direct channels anyway.
– True performance is measured by incremental lift, which assesses the causal impact of ads by comparing exposed users to a control group.
– Marginal ROAS is crucial for scaling decisions, as it reveals the efficiency of each additional dollar spent, which often declines beyond a certain budget level.
– Automated ad platforms can efficiently target users already likely to convert, potentially harvesting existing demand rather than creating new growth.
– Practical incrementality testing methods include geo-split tests and platform holdout groups to move beyond attribution and guide smarter budget allocation.
A paid search campaign can deliver impressive conversion numbers and a strong return on ad spend, making the strategy appear flawless. However, these surface-level metrics often mask a critical issue: many of those conversions might have happened through organic or direct channels regardless of the ads. To move beyond vanity metrics and understand true performance, marketers must analyze incremental lift and marginal ROAS.
Consider a famous experiment where eBay paused its branded paid search ads for a test group. The result was negligible revenue loss, as organic search absorbed nearly all the demand. This case highlights a common pitfall; platforms report attributed return, crediting ads for conversions they influenced, not the causal lift they actually created. With automated campaigns like Performance Max, algorithms excel at finding the easiest path to a sale, often becoming the most expensive touchpoint for a customer already intent on purchasing. Without incrementality testing, automation can simply amplify non-incremental signals, such as brand search capturing existing demand or retargeting ads reaching users moments from a purchase.
Incrementality testing measures the true value a campaign generates by comparing exposed users to a control group. It answers whether marketing created something extra that wouldn’t have occurred otherwise. A channel can boast a fantastic platform-reported ROAS while contributing little incremental growth because it is merely harvesting demand, not creating it. This makes incrementality a far more useful lens for budget decisions than attribution data alone.
Yet, proving a channel is incremental isn’t the final step. To allocate budget effectively, you need marginal ROAS. This metric evaluates the return on the next dollar spent, not the average return across all historical spend. Initial budget often performs efficiently, but each additional dollar can yield diminishing returns. For example, spending $10,000 to generate $50,000 revenue is a 500% ROAS. Adding $5,000 more might only bring in an extra $5,000. The average ROAS drops to 366%, but the marginal ROAS on that last $5,000 is just 100%, essentially trading a dollar for a dollar. This reveals wasted spend that a strong average ROAS would hide, showing a channel may be efficient at a base level but not scalable.
Implementing incrementality tests is accessible. Geo-split testing involves dividing markets into comparable regions, running ads in one and pausing them in another to measure the revenue difference. Search lift tests use platform tools to create holdout groups not shown ads, comparing their behavior to exposed users. These methods, along with controlled spend reductions, provide clearer insights than endless attribution debates.
The fundamental shift is moving from using measurement to report on past performance to using it to guide future investment. ROAS explains which channel gets credit, incrementality shows what actually moved the needle, and marginal ROAS dictates where the next dollar should go. By focusing on whether the next dollar spent is efficient, marketers can transition from justifying spend to strategically allocating capital for genuine growth.
(Source: Search Engine Land)
