Avoid these four common ROAS traps

Retail media buyers love and hate return on ad spend (ROAS)—which misses incrementality.
Join our newsletter to stay up to date on product updates
By subscribing you agree to with our Privacy Policy and provide consent to receive updates
You're in!
Great things are coming your way.
Oops! Something went wrong while submitting the form.

While the calls for a more robust efficiency metric than return on ad spend (ROAS) have been rallying for years, retail media buyers are stuck. ROAS is everywhere. And it has its upsides—it’s easily understood and readily available at low levels of granularity and near-real-time frequency, making it very attractive for day-to-day campaign optimization. While most brands have access to stronger measurement signals via marketing mix models or lift tests, neither of those can deliver results at ROAS’s granularity and frequency. On the other hand, ROAS has no sense of incrementality (would the sale occur without advertising?), and, as we’ve shown in previous research, actually tends to not correlate with incremental sales.

This isn’t the first time easy access to last-touch-based attribution has lured marketers, and there’s a lot to learn from the early days of digital marketing. This post highlights  some of the common “ROAS Traps” that can lure in media buyers, and offers some ways to escape them.

Four common ROAS traps

Cheap pay-per-click (PPC) ad units with high click-through rate (CTR)

If a particular placement is generating a CTR that seems too good to be true, it probably is. That high CTR is likely an artifact of a high base propensity to purchase, or at the very least click, in the audience you are targeting. While these “high-intent” audiences or keywords appear attractive, it's likely that these users would convert organically at a similarly high rate and advertising isn’t doing much to increase that conversion rate. This dynamic played out in the early days of digital advertising as high CTRs and conversion rates among ‘intent based audiences’ coming from a brand’s own retargeting pools or third-party data providers were often high, but there was almost no difference when compared to the base conversion rate of these audiences without advertising.  

Branded keywords

This trap comes with a caveat. While branded keywords are often considered non-incremental sales, and Incremental data does suggest that branded keywords can be over-credited by ROAS, this is not true for all brands or strategies. Where branded keywords tend to be less incremental is for mature brands that have high penetration or high brand awareness. A large base of existing buyers and strong brand equity often gives these shoppers a high base propensity to purchase from the brand already. Where branded keywords can be used to drive incrementality for these shoppers is by using this as an opportunity for basket-building, driving cross-sell and upsell. If a consumer is searching for [brand] + [product a], show them a multi-pack version of the product or a new product innovation that has features that might be attractive to a regular user of the product. Stay tuned for a later article on when else branded keywords are incremental.


This trap is another one that echoes the early days of digital. A consumer clicks on a product, maybe even adds it to cart, and then they are haunted with ads for weeks with complete disregard for any attempt at frequency capping. Cheap open web ad inventory and strong intent signals may seem like a great recipe, but again, is this audience likely to convert before the introduction of advertising? The answer is most assuredly yes.

Overlap with promotions

This is a place where last touch–based attribution and ROAS often take credit for another marketing lever. If a brand is also running promotions on that stock keeping unit (SKU), the elevated conversion rate likely inflates ROAS. This is not to say that advertising cannot effectively be used along with promotions (offsite is a great way to drive traffic to these promotions), but onsite media needs to be thoughtfully used to ensure it isn’t taking credit for another lever.

Three overlooked low-ROAS ad types with strong incrementality

It wouldn’t be fair to just point out problem areas without offering up at least a few solutions. To that end, there are a couple ad types which are often overlooked because they have superficially low ROAS, but in reality, often have strong incrementality.

Upper Funnel/Offsite

This advertising isn't designed to drive immediate sales, but prime the consumer for future marketing activity. To that end, there is often a longer lag between spend in these strategies and sales impact. Given the last-touch nature of ROAS, that gives other advertising that sits closer to the point-of-sale ample time and space to touch that consumer and take credit for the sale.

Sponsored Brands

Compared to Sponsored Products ads, this ad type is generally more expensive, with a lower CTR. The problem here is that click-based measurement undervalues the impressions of this ad unit. While most consumers will scroll down and select a single product, the massive above-the-fold position of these ad units has visual impact. At Incremental, we’ve seen it especially effective in conquesting, which leads to the last example.


This can become expensive, especially in highly competitive categories, and often has low CTRs because these ads are overcoming a consumer’s existing propensity to purchase a different brand. (After all, they are searching for a competitor’s brand.) However, the cost tends to be overset by high incrementality, and this ad type is often one of the more efficient ways of delivering incremental sales, despite its apparently low ROAS.

While ROAS is still a common metric, as better alternatives like iROI or iROAS with an inherent incrementality signal become more broadly available, brands will have better tools available to find those ROAS traps and avoid them.


Featured Stories