There have been dozens of articles recently from eMarketer, retailers, SSPs claiming that Retail Media is the Future of CTV. Although there may be specific use cases where retail data might be applied to CTV, it’s more likely that these are very small, highly targeted opportunities for getting a buyer of a brand to switch up their behavior. The main use of CTV is finding reach for a brand’s message, not hitting a small number of shoppers with precision targeting. These articles get frothy about knowing who the consumer is, but before we get so excited because TV has never been so addressable, my first question is - do you need to spend this much?
The fidelity of retail data is so good that it can literally be used to predict the future of buying patterns. For example, in grocery, shoppers buy the same stuff every week, a great foundation for prediction models. Even outside of grocery, certain categories like candy work really well at drug chains like CVS (CVS Media Exchange (CMX)) and Walgreens for Nestlé , Mars and Mondelēz International. For those brands, retailer data can enable a buyer to understand not just has the shopper bought the category before but when they are going to buy next.
The core use case for Retail Media data is to drive sales within the retailer's ecosystem. It works best when fueling bottom-of-the-funnel activities like sponsored product ads or in-site placements that drive measurable returns. But the moment you extend Retail Media data to a broader goal—like combining it with CTV for upper-funnel awareness—you risk dilution. If you use high-cost, transaction-driven, item-level data to reach consumers on a general CTV platform introduces inefficiencies that will likely lead to reduced return on ad spend (ROAS).
So perhaps first-party retailer data isn’t for every campaign. Just as you couldn’t afford to book Taylor Swift to sing for your daughter’s birthday party, similarly, when you see the absurd price of combining Retail Media’s dataset, and CTV, the ad format that only runs on the most expensive media inventory, you’ll start to question what you’re trying to accomplish.
We know this challenge because we looked at using CTV for Retail Media campaigns at Quotient and saw what Roku and Kroger were testing in 2021. We tried to sell a Gatorade buyer on the benefits of having high quality video creative to enhance sales lift. We couldn’t increase sales lift because Retail Media is measured in ROAS.
Even with a fully automated creative build, the inventory cost of CTV was untenable when looked at against the most popular off-site inventory, mobile programmatic display. Andrew Lipsman talked about this in this week’s Media Ads Commerce post on the 6 pitfalls for RMNs. If mobile inventory was $3 CPM and retail data added $1.50, yielding a $1.00 ROAS over a $4.50 media cost, that's great. But if your inventory cost is $20.00 CPM for CTV, you'll see a negative ROAS. You might argue for including CTV at the start of the campaign to introduce sight, sound, and motion, hoping that shoppers recall your CTV creative when they encounter mobile banners later. While effective, this approach turns CTV into a minor tactical add-on rather than a significant shift in media spend. Moreover, if the buyer relies only on first-party retailer data, which is inherently limited, applying that data to CTV not only narrows available inventory but drives CPM even higher!
To highlight the misalignment between the promise of taking Retail Media + CTV at face value, we’d like to show that for scaled CTV campaigns, there is a spectrum of data sources available at different price points.
Retailer Data - First-Party: is item-level transaction data for one retailer, only available through their Retail Media Network, which usually covers a restricted geo like the US Walmart Connect Amazon Ads Roundel or a sub-region like Ahold Delhaize USA for East Coast, Albertsons Media Collective for West Coast. It would typically include a $1.50 CPM premium to use this data for off-site extension campaigns.
Consumer Savings App Data - Third-Party: Coupon, receipt aggregators, or cash-back providers like Fetch and iBotta can provide similar, item-level purchase data for less. It can be modeled to national scale, and it works across retailers, which helps when you consider you might not find enough shoppers to target if you were to rely on a single retailer. The benefit of an aggregator is that this type of data can limit incorrect inferences because if for example a single retailer like Kroger says the shopper is not a pet category buyer that could just be because they buy that heavy stuff on Amazon.
Retail Aggregator Data - Third Party: Data providers like NielsenIQ Circana Kantar Media MRI-Simmons GfK - An NIQ Company offer extensive scale in grocery, tech, and durables, along with item-level purchase data. These won’t be pure “retail media” since they rely on persona modeling instead of exact shopper-level granularity, they can still be highly effective for channels like CTV. This is similar to Meta’s approach, where broad consumer reach can outweigh the need for precise attribution.
Retail Aggregator (3P) + Ubiquitous Consumer Apps (1P) : That last insight has not been lost on the commerce media networks that Dr. Mark Grether has been incubating over the past 4 years, Uber PayPal. Those companies have enough consumer scale to be able to enliven aggregated data with their highly frequent, location-aware consumer graphs.
We get that we all want Retail Media and CTV to be the power couple. However, while retail data's precision is unmatched for bottom-funnel conversions, scaling this data into broader CTV campaigns is a stretch. That’s why we see analysts try to cover up the measurement problem by overstating the likelihood of consumers to use their remote controls to buy directly from CTV spots. CTV can, of course, benefit from quality data, and the right data product at the right price for the job exists, you just have to know where to look.
