Roku opens up more advertising data for marketing mix modeling to better compete with traditional TV

Roku already provides impression-level data for marketing mix models that break down campaigns at the impression level by designated market areas, i.e. groupings of U.S. cities and ZIP codes like New York, Honolulu and Montgomery-Selma that TV and radio advertisers historically refer to when measuring campaigns. Now the company is expanding that data set.

“What we’re trying to provide is a little bit more information on creative type, daypart, more granular view of DMA so that you can get even down to the ZIP code to provide a little bit more granularity into those models for each one of those measurement partners,” said Asaf Davidov, head of ad measurement at Roku.

The data will be limited to ads sold by Roku, such as The Roku Channel and the portion of inventory in other companies’ Roku apps that the CTV platform owner is able to sell, including inventory bought through Roku’s OneView buying platform.

Roku expanding the data provided for marketing mix modeling seems to address an issue with this form of analysis: It’s built around traditional media channels like television. As Digiday has reported, some advertisers are so beholden to their historical marketing mix models that it is limiting the flow of ad dollars moving from traditional TV to streaming and digital video. “They’re more of a blunt instrument,” an agency executive said of MMM.

Nonetheless, what those models are showing is that advertisers are surrendering some of TV’s efficiency as they move money to streaming, said a second agency executive. Advertisers may be exchanging TV’s broad reach for streaming’s more targeted options, but the streaming’s higher ad prices result in advertisers reaching fewer people for more money, which is an issue for brand advertisers aiming for broad audiences.

Marketing mix models “are not stupid. They’re catching up fast and saying we took a really efficient dollar [in TV] and made it inefficient [in streaming],” said the second agency executive.

Moreover, marketing mix modeling remains an important tool and even seems to be growing in importance. Not only is TV’s measurement system undergoing an overhaul, but digital measurement systems must reckon with the looming demise of the third-party cookie and mobile app tracking crackdowns by Apple and Google. These changes have spurred “a little bit of a resurgence of that MMM measurement,” said Davidov.

Additionally, the collision of TV and streaming advertising may lead to streaming eventually overtaking traditional TV’s share of ad dollars, but the traditional TV ad buying model seems to be winning out, as ad buyers discussed during the Future of TV Week Town Hall on April 19. That will likely only reinforce the role of marketing mix modeling in advertisers’ media planning processes.

“Marketing mix models will catch up, but the way we plan and buy TV today, that’s not changing,” said the first agency executive.

Opening up more data to MMM providers could at least help to catalyze the flow of money to streaming by giving advertisers and agencies a better read on their streaming ads’ performance relative to TV and other media channels so they may identify how to shore up any inefficiencies.

When it comes to data on creative types, Roku will break out information by video ad and display ad, such as the banner ads appearing on the CTV platform’s home screen. The video-specific creative data will also include the duration of an ad so advertisers will be able to gauge the impact of traditional 15- and 30-second spots as well as the newer six-second placement.

“Creative types are super important,” said the first agency executive. “Nobody talks about the role of creative in all of these conversations on effectiveness. We know that 70% of the impact [an ad has] is from the creative and the context it appears in. So to the extent, they’re providing this information, that’s a great thing.”