RTB wasn't built for CTV. Agents can be.
Real-time bidding (RTB) solved an important problem. It’s just not the problem we’re now asking it to solve.
When Google and the rest of the industry began building RTB, we were trying to automate display advertising. That was the design center. I was involved in those early conversations before OpenRTB became the industry standard. We were solving for banners on web pages.
Then something happened that almost every successful technology experiences: It became the hammer, and every new media channel became a nail.
Mobile, digital audio, digital out-of-home, connected TV: RTB encompassed them all.
Succeeding in one environment, though, doesn't make a technology the right architecture for every environment that follows. Nowhere is that more obvious today than with CTV.
CTV inherited infrastructure that wasn't designed for it
Television isn't just ‘display advertising on a larger screen.’ On TV, inventory is scarce. Ad pods create dependencies between decisions, and most content is premium. User sessions are longer, and measurement looks different.
Yet we largely forced CTV into infrastructure that was originally designed to auction banner impressions.
Over time, the industry compensated. Supply path optimization emerged because too many paths reached the same inventory. Curators arose because buyers couldn't navigate an increasingly complex marketplace. Private marketplaces multiplied because it became difficult to pass enough information through RTB. And request duplication exploded, because multiple participants competed for the same opportunity.
Each new kind of company solved problems. But the problems existed largely because the underlying architecture wasn't built for television in the first place.
Agents start with outcomes, not auctions
Agentic advertising approaches the problem differently. Instead of asking about the next impression, it asks what the advertiser wants to accomplish. An AI agent can understand campaign objectives, available inventory, historical performance, pacing, constraints, and optimisation goals simultaneously.
Instead of forcing humans to create hundreds of rules, PMPs, insertion orders, and manual workflows, the system can orchestrate those decisions around the desired outcome.
The pipes won't disappear overnight, and RTB will continue transporting transactions for years. Orchestration, though, will begin to move elsewhere. The optimisation layer will become agentic even as much of the execution keeps flowing through existing infrastructure.
I expect that transition to happen unevenly across channels. CTV will be the first major inventory category to transact agentically at scale, because that’s where RTB’s limitations are most apparent. From there, I expect digital out-of-home and digital audio to follow, then mobile and native. Display, ironically, may be the last to move because it's the environment RTB was actually built to serve.
This isn't replacing RTB tomorrow
One misconception we hear regularly is that agentic advertising requires companies to throw away their existing infrastructure. That's not how transitions happen.
Campaigns can be planned and optimised agentically while still executing through RTB where scale requires it. The difference is that fewer manual decisions will happen upstream.
Instead of people stitching together dozens of systems, software begins orchestrating those systems on their behalf. That transition is likely to be gradual.
Why CTV may benefit first
Every emerging channel exposes the assumptions built into existing infrastructure. CTV simply exposes more of them.
The more valuable the inventory, the more expensive complexity becomes. Each additional intermediary adds latency, margin, and operational overhead. Every manual workflow offers another opportunity for friction.
RTB also leaves CTV inventory on the table. When an impression doesn't fit the mechanics or economics of an auction, that inventory can go unfilled even though there may be an advertiser willing to buy it.
An agentic approach can change that. By starting with an advertiser's objectives and evaluating available supply against them, agents can make more of the CTV inventory pool addressable. And the opportunity doesn't have to stop with inventory already available programmatically. For the first time, the same agentic orchestration could extend into linear and broadcast television inventory that has historically sat outside RTB.
That's why I believe CTV will be the first major channel where agentic infrastructure demonstrates measurable advantages. Agentic will expand the inventory that can transact programmatically.
We don't need a better hammer
RTB deserves credit for fundamentally changing digital advertising. But technologies have design centers, and RTB's design center was display advertising.
The industry's challenge now is to recognize that newer channels, especially CTV, require a different architecture for planning, orchestration, and optimisation.
The future is agents making better decisions while existing infrastructure evolves underneath them. That may finally give CTV an infrastructure built for the medium.
[Editor's note: This is a contributed article from Adgentek. Streaming Media accepts vendor bylines based solely on their value to our readers.]