By Brian Mandelbaum, CEO and Co-Founder of Attain
Despite the tumult and headaches cookie deprecation has caused over the past few years, it’s spurred a much needed marketing measurement renaissance. Vanity metrics like clicks are increasingly seen for what they are: a loose proxy of success. The influx of first-party purchase data gives marketers the opportunity to actually evaluate outcomes based on real sales.
Yet with this new measurement landscape comes widespread conflicts of interest. The majority of platforms with first-party purchase data also sell media, and use that same data to measure performance. It’s in their best interest to report back on positive incrementality, attribution, and ROAS in order to encourage advertisers to invest more into their platforms. Additionally, if the platform sits further down the funnel or incentivizes purchases, better sales outcomes are almost guaranteed. This raises an important question: can advertisers trust the accuracy of attribution and incrementality metrics reported by platforms that directly benefit from positive outcomes? Can measurement be trusted if it is not media agnostic?
More Options, More Caveats
The 800 pound gorilla of the moment, retail media networks (RMNs), are designed to capture consumers at the very bottom of the funnel. While the increasing availability of RMN off-site targeting leaves more room for brand marketing and awareness campaigns, platforms often use retargeting to reach customers who have already browsed for the product. Perhaps most importantly, RMNs do not have visibility into the full retail landscape. For example, if an impression for yogurt from one RMN led to a purchase at another retailer, that incremental sale does not get counted. Without insight into all possible points of purchase, true incrementality measurement by RMNs is theoretically impossible.
While RMNs have experienced meteoric revenue growth, expected to hit $165 billion in 2025, their overall percentage of digital ad spend is plateauing at 22%, and is even expected to decrease slightly next year. This can be seen as a sign of maturity, but it could also be signaling a growth ceiling due to lack of legitimate measurement.
Another novel source of first-party purchase data are rewards apps that offer discounts and coupons. These apps drive their own type of “incremental” sale, with the goal of introducing new products to consumers and converting them into loyalists. However, using incentivized rewards data for measuring cross-channel campaigns can significantly bias results. Additionally, these apps skew heavily CPG, cutting out a huge swath of brands residing in other verticals – QSR, apparel, insurance, and telecom, to name a few. This leaves the majority of brands with few solutions for closed-loop measurement outside of the major walled gardens, who are known to over-attribute and lack incrementality solutions.
An old school solution to eliminating bias is to work with legacy sales lift providers who have data sharing agreements with retailers and a more holistic view of the retail landscape. Yet this data sharing relationship is becoming increasingly tenuous due to the rise of RMNs and the increased value of first-party data, making retailers much less willing to share their data. These partners also offer solutions that are almost exclusively CPG oriented, and reporting is notoriously expensive, slow and not conducive for real-time optimization.
Bidding with Bias
Bias itself is now baked into many optimization strategies. Many marketers choose to rely on ROAS as their holy grail, and for good reason. They want to ensure ad dollars are efficient and performant. However, by setting walled garden and RMN optimization strategies to prioritize ROAS above all else, bidding algorithms default to finding consumers who are already at the bottom of the funnel. This does not drive incremental purchases, but rather captures consumers who are already well on their way to making a purchase. This can lead to a fundamental mismatch of what’s happening in the real world, such as organic sales growth, compared to what’s being reported by platforms and retailers.
While a myriad of purchase-based solutions to our cookieless problems have accumulated over the last couple of years, will marketers stand for the newest iteration of “grading your own homework?” While measurement now hinges off of what matters most – sales – we can’t ignore the fact that most platforms with sales data have inherent conflicts of interest. Data sets that hold the most promise in solving this dilemma are unbiased observers of purchase behavior, and better yet, cross-channel, cross-vertical, and cross-retailer.
