TV Advertisers Need Better Audience Segments

On TV And Video” is a column exploring opportunities and challenges in advanced TV and video.

Today’s column is written by Tom Weiss, chief technology officer and chief data scientist at Dativa.

With online segments overlaid on linear schedules, many are hoping for a golden age of TV advertising, where brands can break out beyond the demographics used for linear audiences, such as “truck intenders” and “parents with children.”

However, little time has been spent considering where these segments come from and whether the quality is good enough for TV.

If marketers are running pay-per-click campaigns online, the quality of their audiences is less relevant because they are only paying for people who click. The presence of riff-raff on their lists has almost no impact. But when the same segments are used to map target audiences to TV advertising, the errant members of the list drive the media plan equally as much as the list’s valid members.

On TV, the effective cost of the campaign is directly influenced by segment quality. Hence, media planners need to focus on the purity and accuracy of the segments they use. Just mapping what they have applied to digital campaigns can be highly problematic.

The value of online segments is derived from both the quality of the segment and the reach. Gradations of quality are acceptable if ROI can be proven.

In TV, you do not need segment reach as TV already gives you reach. You do need the highest-quality segments. As an industry, we have spent much time validating viewing data from set-top-boxes and smart TVs, but we now need to take a much closer look at the audience segments that we are matching to these viewing data sets.

Most online segments come from online activities, with a probabilistic measure of a consumer’s attributes. For example, we might have a set of cookies from people who have recently purchased diapers online. With a high probability, we can assert that these cookies belong to parents with children. With a lower but still high confidence, we can argue that they are likely to be mothers of children.

To build reach online, data brokers will take the segment that starts off with diaper buyers, market it as “parents with children” and then ratchet up the lookalike model to get as many people with similar behavioral characteristics. That lookalike model might be conservative (narrower) or aggressive (broader), depending on the product and the economics of conversion.

The broker then sells that segment with a good reach among the target segment online. By the time modeling has been done, the segment – “parents with children” – might be 20 times the size of the original seed of recent diaper buyers.

If the same segment is used on TV, the buyer matches the cookies to TV viewing to pull the linear spots that index highly for that segment. However, because the broker has diluted the segment to give reach, buyers will get a much less tightly focused set of spots than if they were just looking at the primary data.

To be effective on TV, segments need to have the most distinctive characteristics to bring out the most strongly differentiated programming in the indexing model. “Recent diaper buyers” are perfect for a Pampers campaign but of limited use for other parenting campaigns.

TV has become the lifeblood of brand advertising precisely because it provides reach. Audience-based, data-driven TV enables advertisers to focus that reach on more targeted audiences. However, to do this well, media buyers need to start asking questions about segment provenance and drive transparency in the data sets they are purchasing.

First, buyers need to know precisely how the segment was sourced. Does it come from cookies or actual PII-tracked activity? What is the action that caused the user to be included in the segment? And how many different data sets have been combined to make it?

Second, buyers need to know if a lookalike model was used to expand the size of the segment. If the source data set is too small to be matched against TV, then chances are that adding a lookalike will make the situation worse.

By holding brokers and data providers to account, buyers can be sure that the data they are getting is the right size and shape to fuse to TV.  Data needs to be high-quality on both sides of the process – viewing data and audience data.

Follow Dativa (@Dativa4Data) and AdExchanger (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!