In the agentic era, is your traffic driving revenue or risk?

A few years ago, Forrester analysts Sandy Carielli and Joanna O’Connell made a smart argument*: Security and marketing leaders would need to come closer together to fight fraud.

They were not wrong; they were early.

Marketing fraud was already thriving in the empty spaces between teams. The researchers found that security teams were not engaged in ad and marketing fraud discussions about 90% of the time, while the chief information security officer (CISO) was often not involved at all.

The problem was never the logic. It was the operating model. Security and marketing were — and still are — measured on different things, funded differently and asked to tolerate very different kinds of friction. As one source in the report put it, “The CISO is about risk reduction. The marketer doesn’t want to block a paying customer.”

The researchers’ answer was thoughtful: shared KPIs, clearer accountability and a more formal operating model across marketing, security, fraud and customer experience.

In most organizations, though, that convergence never fully happened. The incentives remained too different and the cadences too separate. Fraud mattered, but it was still possible to treat it as someone else’s problem often enough to preserve the org chart (and meeting calendars).

Agentic traffic changes that.

The internet is moving from a world where AI mostly reads the web to one where it increasingly acts on it. Human analyzed more than 1 quadrillion interactions in 2025 and found that automated traffic grew about eight times faster than human traffic, AI-driven traffic grew 187% over the course of the year and agentic AI traffic grew 7,851% year over year.

More than 95% of that AI-driven traffic was concentrated in just three industries: retail and e-commerce, streaming and media, and travel and hospitality.

The agentic shift

Those numbers matter not just because they are big, but because of where the activity is happening.

In 2025, 77% of observed agentic activity landed on product and search pages. But nearly 9% hit account pages, 5% touched authentication flows and 2.3% reached checkout. This is a fundamental change. The same machine-driven interactions that shape early-buyer-stage moments like discovery, recommendation and conversion are now touching login, identity, session management and checkout/transaction flows too. And early 2026 figures show that those higher-stakes use cases are growing quickly.

That is why the old separation between “growth” and “security” starts to break down. Marketing can no longer treat automated traffic as a back-end security issue, and security can no longer treat it as a traffic-quality problem for someone else to sort out. The decision to block, allow, verify, govern or optimize synthetic traffic now directly affects both revenue and risk.

This is also why the emerging debate around AI traffic is often framed too crudely. Many companies still behave as if the choice is binary: block automated traffic or let it in. But that is the wrong frame for the more complicated reality we inhabit today. Across the interactions analyzed in Human’s report, only half a percentage point separates the rate of benign automation from the rate of malicious automation. In other words, the difference between a valuable machine-driven interaction and a harmful one is often narrow, contextual and economically significant.

Not all automation is created equal

A shopping agent comparing products on behalf of a real consumer is not the same as an abusive scraper extracting data at scale. An automated login interaction may be a legitimate assistant operating on behalf of a user or the first step in a coordinated fraud attempt. A publisher deciding how to handle AI access is not just making a content-protection decision; they are making a distribution and monetization decision too.

In the agentic era, blocking and accepting automated traffic as a whole is no longer a technical setting. It is a business policy.

What organizations need now, then, is not simply a call for security and marketing to “work together better.” They need a shared trust layer: a common way to see who or what is acting on digital properties, understand intent and apply controls that reflect business priorities rather than blanket fear. The question shifts to whether that interaction is trustworthy, governable and aligned with business goals, not just if the interaction was automated or not.

That trust layer matters because agentic traffic is not just another category of automation. It is a new operating reality. It sits across the customer journey. It touches the surfaces marketing wants to optimize and the surfaces security needs to protect. It forces a harder, more consequential kind of decision-making, because now the same interaction can represent new demand, new efficiency, new abuse or all three at once.

The companies that win here will not be the ones with the loudest agentic strategy decks, they’ll be the ones that build the ability to distinguish which machine-driven interactions create value, which create risk and which require governance instead of reflexive blocking. Carielli and O’Connell saw early that marketing and security were converging around fraud. The agentic era is the moment when that insight stops being aspirational and becomes operational.

In the next phase of the internet, trust is no longer just about control. It is critical revenue infrastructure.

*Forrester Research, Inc., Now Is the Time for Security and Marketing to Team Up to Fight Marketing Fraud, by Sandy Carielli, Joanna O’Connell, Amy DeMartine, Emily Collins, Isabelle Raposo and Jen Barton, Feb. 3, 2022.

This op-ed represents the views and opinions of the author and not of The Current, a division of The Trade Desk, or The Trade Desk. The appearance of the op-ed on The Current does not constitute an endorsement by The Current or The Trade Desk.