Bayer — pharmaceuticals, consumer health and crop science — is testing models that flag customers about to leave a brand, find people who aren’t buying in a category yet, and assign different values to individual audiences and bids. In small-scale tests, the predictive audience work has returned four times the effectiveness of standard demand-side platform targeting, said Maria Givens, VP and head of media and digital platforms. She is not promising it repeats. “I consider it a scientific experiment, so we’re constantly iterating,” she said. “I’m not saying that 4X is a number we will hit every time.”
The attrition piece connects a partner’s sales intelligence to audience intelligence in real time. Once a customer has left, they land back in the new-to-brand bucket and cost more to convert; reaching them with specific messaging beforehand is meant to keep growth net-new rather than replacement. Bayer has also tested contextual AI modeling in connected TV down to the program level, and customized bidding algorithms to value individuals rather than cohorts. Economic uncertainty is pushing it toward free ad-supported streaming as viewers drop paid services.
Targeting is outrunning creative. “Media is only as effective as creative and vice versa,” Givens said, and the closed-loop system linking the two does not exist yet at Bayer. Her stated limit is personalizing purposefully rather than personalizing because personalization is available.
Generative engine optimization looks different in healthcare. What drives citations in beauty does not carry over; medical journals and medical sites appear to be among the most influential sources models draw on. That pulls medical affairs and science teams into media strategy — establishing authority with journals, then putting that material on Bayer’s site so the reference is crawlable. Givens does not see paid media as the lever. Buying an ad on an influential source, she said, “doesn’t feel authentic.”
Most hands-on-keyboard work is internal; Omnicom handles linear television buying and the upfront. On agentic buying, Givens sees no build-or-borrow choice yet — “I don’t think there is an ‘or’ right now. It is all ‘and.’” What she does not have is proof. She wants an A/B test running agentic and traditional buying head to head, and says she has not seen many public results comparing the two. “The devil is in the details.”
Read more at Ad Age.