For something that has been evolving for several decades, marketing mix modeling is still very much a journey in progress. To Spark Foundry’s Lisa Giacosa, the final destination should represent a hybrid of short- and long-term ROI, brand equity and de-duplicated reach and frequency metrics.

Overall, Giacosa says in this interview with Beet.TV at the CIMM Cross-Platform Video Measurement and Data Summit, “I think we are still solving problems that we probably should have already cracked. But as an industry there are challenges with getting all of that data to live in the same space, with walled gardens over here and set-top box data being over there.”

Giacosa, who is EVP, Global Managing Director, Head, Data, Technology, Analytics & Insights, acknowledges the need to build scale and reach across platforms that are still somewhat siloed. “We’re getting some of the data signals, but it will be so much more powerful when we bring all of those together,” she says.

Having participated in the CIMM session titled TV Attribution Takes Center Stage, Giacosa thinks “there’s no doubt that TV drives other media as well. Certainly in the models that we do at Spark Foundry,” which include POEM—Paid, Owned and Earned Media mix models.

“What that does is it looks at not only the TV attribution and the direct plays but also the indirect plays it has on other media,” she explains. “When something happens in TV, what was the impact on paid social, search, in-store or website?” Bringing those elements together helps to avoid marking “dark decisions on what to reinvest in or take out of the mix.”

Spark Foundry, which is part of Publicis Media, also incorporates brand equity data for a baseline not only of sales performance but in “how we can get to be more fast moving in terms of how we look at brand equity as well. I’m very fearful oftentimes that people follow ROI off of a cliff” because the focus typically is short-term results.

Advertisers also need to think about content and “those longer-term, brand-building investments that we make to then look at the long-term return on ROI as well by building that hybrid of a model that looks at the ROI plus brand equity” and, finally, de-duplicated reach and frequency.

Asked about the subject of identity graphs, she prefers to see a move toward a “meta graph” given, among other things, the impact that the General Data Protection Regulation has had in Europe. “What we’re already starting to see in Vermont, in L.A. and other places within the United States is they’re following that path too.”

Giacosa predicts we will see “GDPR-esque within the United States fairly soon.”