AI Is Enabling the Future of Advertising: Behavioral, Contextual, and Intent Signals

For years, digital advertising operated on a shared compromise. Advertisers wanted precision. Publishers wanted monetization. What both got instead was approximation.

The industry built increasingly sophisticated ways to predict what users might want, using demographics, contextual signals, browsing behavior, and machine learning models trained on historical data. It scaled into a massive ecosystem that helped power much of the modern internet.

For advertisers, that meant reaching highly relevant audiences at unprecedented scale. For publishers, it created sustainable monetization models. The system worked remarkably well.

But even the most sophisticated inference-based systems were working with an approximation of intent, not a direct expression of it. AI introduces something new: continuous, evolving expressions of intent.

The first two layers of advertising: behavioral and contextual signals

Modern digital advertising has largely been built on two categories of signals.

Behavioral signals capture what users have done. Searches, clicks, purchases, app usage, and browsing activity all contribute to a profile of likely interests and preferences. Over time, these signals build surprisingly accurate models of what a user might want next.

Contextual signals capture where users are. The content being consumed, the application being used, and the topic being explored all provide valuable information about what might be relevant in a given moment

Together, these signals created one of the largest and most sophisticated advertising ecosystems in history. Search sharpened it further by introducing declared intent. When a user typed a query, they revealed what they were actively looking for. For advertisers, this created one of the most effective performance channels ever built. For publishers and platforms, it unlocked a monetization model that scaled with user value.

But even search has limits. A query captures a moment, not a journey. It tells you what was typed, not the goal behind it, the constraint shaping it, or how the need might evolve.

The third layer: deep intent

Conversational AI moves beyond this.

Instead of compressing intent into a few keywords, users express it naturally, adding context, refining requirements, and revealing what they are actually trying to achieve. A query like "running shoes" becomes "I'm training for a half marathon and need something supportive under $150."

The difference is not just more information. It is higher quality intent. And unlike search, the interaction doesn't end there. A user may continue exploring options, refining criteria, comparing alternatives, and moving closer to a decision. Each interaction adds context. Each response reveals more about what they are actually trying to accomplish.

Intent is no longer captured at a single moment. It develops throughout the conversation creating a far richer signal than a keyword or pageview has ever provided.

The opportunity is not to replace behavioral or contextual targeting. It is to enhance it. Behavioral signals explain what users have done. Contextual signals explain where users are. Intent signals explain what users are trying to accomplish. Together, they create a far richer understanding of user needs than any individual signal alone.

For advertisers, the signal becomes clearer, closer to decision-making, and more actionable.

For publishers, it transforms the value of their inventory. Every interaction becomes a high-signal moment, not just a pageview but a step in a decision process.

Why the existing stack breaks

This shift creates a challenge for the current advertising infrastructure.

The existing stack was built around a specific assumption: that inventory is predictable. Content exists, slots are defined, and ads are placed around them. A publisher knows in advance where ads will appear. A DSP bids on audiences against those placements. Measurement is tied to impressions and clicks on fixed units.

Conversational AI violates those assumptions. There are no fixed pages. No stable layouts. No predefined ad slots. A conversation about planning a hiking trip has no banner position, no pre-roll, no sidebar. The interaction is the entire surface.

And while behavioral and contextual signals remain valuable, existing advertising systems were never designed to process intent signals at this depth and scale. Behavioral and contextual signals remain as valuable as ever, but they were built to operate within a specific structure: page URLs, placement IDs, defined inventory. A user asking about paint colors, then comparing contractor quotes, then researching permit requirements generates a stream of high-intent signals with no structural home for a traditional ad unit. It's not that the existing stack fails, it's that the inputs conversational AI produces are a format it was never designed to receive.

Publishers are sitting on high-intent interactions but lack the infrastructure to monetize them effectively. Advertisers want access to these moments but cannot reach them through existing systems.

The result is a growing gap between the value being created in conversational environments and the infrastructure available to monetize it.

Ads become part of the experience

What emerges instead is a different model. In conversational environments, advertising doesn't need to sit alongside the experience. It can become part of it. Search introduced this idea by embedding advertising directly within a user's flow of intent. Conversational AI extends it further.

Recommendations, suggestions, and sponsored experiences can appear naturally within the interaction itself, aligned with the user's goals and the context of the conversation. The result is advertising that arrives as part of the conversation rather than around it: useful at the moment it appears, rather than tolerated until it disappears.

That's a different creative brief entirely. And it produces a different kind of advertising — one that users are more likely to engage with precisely because it arrives when it's relevant, not simply when a slot needs filling.

A new layer of infrastructure

Making this work at scale is not a formatting problem. It's an infrastructure one.

The existing advertising stack was built for a different set of inputs, and it does that job well. What's needed now is a layer that sits alongside it, purpose-built for conversational environments: systems that can interpret natural language interactions, extract structured signals without exposing sensitive data, evaluate competing demand sources in real time, and translate the outputs into a form that existing bidding systems and measurement tools can work with.

This is what Velocity is building. By combining intent signals with behavioral and contextual data, Velocity is building the layer that makes this possible, combining intent signals with behavioral and contextual data to connect advertisers with users at the moment intent becomes commercially meaningful, and giving publishers the infrastructure to capture that value without compromising the experience their users came for.

The three layers of advertising — behavioral, contextual, and intent — have always pointed toward the same goal: understanding what users want and connecting them with relevant demand at the right moment. For the first time, the signal quality is there to do it properly. The work now is building the infrastructure to match.

Velocity is the infrastructure layer connecting all three signal types. Learn more at velocity.io