How understanding users’ intent redefines context in digital advertising

As VP of global product commercialisation at RTB House, it’s my honour to be a bridge between the diverse needs of marketers from around the world and the products that RTB House designs, builds and brings to market.

In recent years, key trends have caused various ‘tried and true’ types of digital marketing to be, well, a bit more tried and a bit less true. Privacy rules have become stricter. Third-party demographic data has become less accurate, and the demand to prove the value of marketing across the funnel has amplified.

Marketers have a strong desire to reach both new and existing customers, but often struggle to get beyond behaviourally-driven tactics. Nowhere is this more true than in the realm of contextual targeting.

Why current approaches fall short

Most contextual targeting today operates on surprisingly basic principles. We scan web pages for keywords, categorise content into broad categories, and make assumptions about what users want based on where they are.

The fundamental issue is conceptual. We’ve been optimising for reach and scale rather than relevance and understanding. Behavioural targeting attempted to solve this by tracking user actions over time, but even before privacy regulations tightened, this approach had limitations. Historical behaviour doesn’t always predict current intent, and broad behavioural categories often miss the nuances of what users actually want in the moment.

The ‘deep learning’ breakthrough

Over the past few years, I’ve been involved in developing what we call IntentGPT – an AI system that analyses webpage content in ways that go far beyond keyword matching. The breakthrough came from applying ‘deep learning’ to understand not just what content says, but what it means in context.

The system examines content structure, language patterns and implicit meaning within text. More importantly, it processes this information alongside real-time user engagement patterns to identify genuine intent signals. Rather than categorising content into predetermined categories, it identifies specific aspects that indicate why users are engaging with particular web pages.

One of the most interesting technical challenges we solved was connecting product inventories with contextually relevant placements. The system identifies web pages where specific products would genuinely resonate with users, then enables targeted bidding for those particular ad opportunities. This creates a direct bridge between what advertisers want to sell and where users are most likely to be interested.

What we’ve learned about performance

The results from early implementations have been encouraging. We’re seeing engagement rates that are 44% higher on average compared to traditional contextual methods. But more interesting to me is what this improvement represents. It’s better at aligning what users are actually seeking and what advertisers are offering.

For measurement, we’ve found that engagement quality metrics matter more than traditional reach indicators. Time spent with advertisements, conversion intent signals and interaction depth provide better insights than click-through rates alone.

Implications for the industry

Based on my experience developing these systems, I believe we’re at an inflection point. The advertising industry can either continue trying to replicate old targeting methods within new privacy and technological constraints, or we can build fundamentally better approaches to understanding user intent.

For advertisers, this means rethinking campaign planning around content context and behaviour indicators rather than demographic assumptions. It also means adjusting measurement frameworks to emphasise engagement quality over quantity.

We’ve found this approach particularly effective for sectors like travel, fashion, classifieds and automotive – industries where user motivation significantly impacts conversion likelihood. Context matters enormously in these verticals.

The shift represents moving from assumption-based targeting to data-driven understanding of user intent. Companies implementing these methods can expect improved campaign performance and more efficient budget utilisation.

Looking forward

As someone who has listened closely to the needs of real-world marketers for several years, I’m optimistic about where the industry is heading. AI and deep learning now allow technology vendors to create true win-win propositions – those rare approaches that both provide higher levels of user privacy and higher levels of product performance.

For those interested in exploring these approaches further, I’d encourage looking into how contextual AI systems can be integrated into existing campaign workflows. The technical barriers are lower than many assume, and the performance improvements can be substantial.

Visit us at rtbhouse.com to learn more.

Jaysen Gillespie is VP of global product commercialisation at RTB House.