Seedtag has rolled out its latest product innovation – AI Intention Models – aimed at helping advertisers more effectively reach high-intent audiences across the open web, without relying on personal data. This new development enhances Seedtag’s contextual AI engine, Liz, with real-time intent recognition to bridge the gap between casual browsing and purchase readiness.
Rather than relying solely on keyword or category-based targeting, the AI Intention Models analyse sentiment, content depth and engagement signals, and differentiate between mid-intent and high-intent to determine the likelihood of a user taking action and For example, a user reading a general article on about the pros and cons of buying vs. leasing a car would be scored differently to someone viewing a detailed car review – enabling advertisers to fine-tune placements based on actual intent, not just topic relevance.
Designed to reduce mid-funnel inefficiencies and deliver measurable outcomes, the models are trained on intent-labelled datasets and continuously adapt based on brand-specific KPIs.
“The launch of our high-intention models is a pivotal step in enhancing activation effectiveness,” said Fabio Arnau, VP of Brandformance at Seedtag. “We optimise both reach and profitability by analysing diverse datasets, including search engine positioning, helping brands engage decision-ready users and discover new audiences.”
A recent campaign for Nissan in Spain illustrated the model’s effectiveness. Aimed at promoting the Nissan Qashqai to users actively evaluating new car options, the campaign achieved a 68 per cent lower Cost per Quality Visit (CPQV) and a 35 per cent lower Cost per Lead (CPL) than the target, while tripling the number of qualified visits. “These results confirm that contextual AI is key to driving brand performance,” added José Manuel Muries, Cluster Director, Nissan United.
The launch further underscores Seedtag’s focus on delivering product innovation through AI – a growing priority for marketers in the Middle East looking to balance privacy, performance and contextual relevance.
