MADRID — For years, contextual advertising has largely been a game of matching keywords on a page to the topic of an ad. But as the technology improves, so does the ambition to move beyond simple semantics – into realm of human nuance.
The next frontier involves training systems to understand not just what a piece of content is about, but also the likely interest, emotion, or purchase intent of the person consuming it. This approach aims to differentiate between a reader casually researching a new processor and one actively looking for the best back-to-school laptop deal, even if both are reading about technology.
This shift toward what Grego Martínez Luque executive calls “neurocontextual” targeting relies on modeling advertising signals to function more like a human brain, interpreting the subtle cues that define a user’s mindset, said the SVP, product, Seedtag, in this video interview with Beet.TV.
From network analysis to neuroscience
“When I joined Seedtag, we were doing what I call today, the basic of contextual,” he said. “We were trying to compete with others in who could understand much better each one of the URLs. At the time, we didn’t have the technologies we have today.”
The company’s technology has since progressed from simple URL analysis to network-level analysis and now to modeling context through embeddings, a form of AI representation. “We discovered that with new technologies like embeddings, we could train system in something way more complex, which is understanding the context like a brain,” Luque said. “This is where neuroscience meets contextual.”
“We are able to get all signals and we are able to model it out… in order to understand not only what the page is talking about, but also the interest that someone is having when they are reading it, emotion that you feel when you are reading or watching a content or even the intention of what are you thinking about doing just after,” he added.
For one toy company, this meant targeting the emotion of nostalgia by identifying and grouping online content about ‘90s shows and cartoons.
The agent will see you now
Criteo’s CRO Brian Gleason joined Seedtag as CEO in August.
To make these complex targeting capabilities more accessible, Seedtag is developing an AI agent that acts as a conversational interface for building campaigns. Luque believes that one of the most difficult aspects of contextual advertising is not the technology itself, but the process of creating and explaining a custom audience for a brand.
“Instead of getting into a platform and deciding between thousands of different audiences, we changed the approach and now you talk with an agent,” Luque said. “You tell the agent what are your purpose, what’s the brief you have for your campaign, what are the KPIs you’re looking for… you throw it to the agent and the agent is going to start working for you. It’s going to draft a custom audience for you.”
This approach is intended to streamline the creation of bespoke audiences, a process Luque said is not easy to do on the fly. By simplifying campaign setup, the company aims to scale its custom solutions, a strategy supported by recent moves like its June 2024 acquisition of the sell-side ad platform Beachfront to bolster its CTV capabilities. A user can then refine the agent’s suggestions before activating the audience.
