Sooner or later, everything old is new again. In the digital advertising industry, new guardrails around data privacy are giving new life to a 60-year-old measurement technique.
Marketing mix modeling (MMM) first emerged in the 1960s as a way to help marketers analyze how their activities impact business performance over time.¹ Today, MMM has found new relevance for marketers seeking privacy-first cross-platform measurement without the use of individual identifiers.
But as MMM builds momentum, many models fall short of accurately reflecting today’s more complex media environment, including powerful channels such as TV streaming. “A lot of MMM providers are trying to understand where TV streaming fits into that landscape,” says Michael Ryan, senior ad measurement lead at Roku.
These efforts are paying off. In fact, some models show that streaming video may be the most efficient way to drive both short-term and long-term sales impact.² Unfortunately, a few persistent misconceptions and pitfalls have held marketers back from gaining a full view of TV streaming‘s value in the media mix.
MMM projects how online and offline marketing activities known as the 4Ps—price, product, place, and promotion—affect sales or other key business objectives. MMM relies on aggregate historical campaign and channel data, rather than user-level data, driving renewed interest in MMM as third-party cookies and mobile ad IDs become increasingly limited.
Given its reliance on aggregate data, Anna Miller, Senior Manager, Ad Measurement, at Roku says,
“For individual platform or publisher investment decisions, marketers will want to use measurement capabilities that track actual consumer response from campaigns, such as sales lift or incrementality measurement.”
Sales lift is a better fit for advertisers looking to make investment decisions based on how effective a publisher is at driving sales. Unlike the broad strokes of MMM, sales lift tracks real time change in sales driven by actual campaign exposure by comparing sales for those who saw a campaign to sales of a like-for-like control group. This measurement isolates the specific impact of a publisher campaign and takes less time to produce results, making it the best fit for investment decisions based on sales performance.
In traditional MMM models, TV streaming may incorrectly appear to be less efficient compared to other channels. This is a challenge faced by all streaming providers. Miller says,
Advertisers buy TV streaming because it provides an addressable audience that can be targeted, optimized and exposed to custom ad types that are proven to resonate. These attributes are reflected in TV streaming’s higher inventory prices, but since they are not accounted for in most MMM, which historically reward inexpensive CPMs, TV streaming can appear inefficient. While CPM is a factor in MMM, it is just one of many that should be considered. Yet even today, some models only consider factors that are easily accessible, like impressions and CPMs.
“While that might have worked for linear when that was all the data you could get your hands on, because of the advent of TV streaming, there’s a lot of value that’s frequently not tracked [in MMM],” Miller says.
To better assess how TV streaming contributes to broad business outcomes, MMM needs to weigh a greater range of factors including viewability, daypart, genre, custom ad types such as custom banner units and branded experiences, targeting tactic, and video completion rate.
Many marketing mix models mistakenly include TV streaming in linear or digital categories when TV streaming does not have enough historical data or media weight to be modeled independently. This leads to misattribution as key components of these models (such as saturation point, frequency, half-life, and penetration) vary dramatically by marketing channel.
“These models tend to be two-year historical regression-based models and you need sufficient media weight to accurately identify and group impact per channel,” Miller says.
Classifying TV streaming in either linear or digital categories can lead to misrepresentation. TV streaming should be grouped with other streaming providers for a channel-level read whenever possible.
“If someone wants TV streaming to be accurately reflected in their model, they should ensure their campaign size and flighting meets modeling provider minimums for accurate reporting, which tends to start around 50 million impressions minimum,” Miller says.
Similar campaigns across channels can lead to high rates of correlation in a model, making it impossible to distinguish the impact of each channel. If a read on TV streaming is a requirement, ensure TV streaming as a channel has sufficient media weight and distinct flighting to properly separate it from other media in the model.
As marketers use MMM for channel-level budget allocations, they need models that can more accurately reflect the business questions they want to answer — as well as the complexities of today’s media environment.
To better account for TV streaming in MMM, advertisers need to ensure there is sufficient weight with distinct flights, an independent TV streaming classification, and factor in attributes not normally reflected in models such as custom ad type and targeting tactic.
MMM shouldn’t be considered in a vacuum. Ryan says,
Although CTV may look like an underperformer in some marketing mix models, it can generate a 30 percent stronger ROI compared to other types of ad spend, according to Analytic Partners, the leading provider of commercial analytics, which is one of four vendors in Roku’s Marketing Mix Modeling Partner Program. TV streaming has exploded in recent years, but the difficulty in capturing its true impact may actually be holding back adoption.
Ryan advises marketers to get more advanced in their conversations with MMM partners and ensure that they’re passing the most granular data possible. Roku can provide granular, real time reporting for any advertiser model if requested, easing the burden of data collection. This is aimed to help marketers more accurately determine the true value of all of their placements, including more premium ad experiences.
Ultimately, marketers need living and evolving models that adapt to their changing needs and the media landscape.
Learn more about Roku’s Measurement Program here and subscribe to our newsletter to receive regular updates from Roku Advertising.
¹ IAB; The Essential Guide to MMM and MTA