The IAB Standardized Signaling for CTV's New Ad Formats. The VAST Payload Is Still Unverified.

The IAB just finished defining how CTV's new formats get bought and sold

On June 5, 2026, IAB Tech Lab closed the public comment period on standardized signaling for its CTV Ad Portfolio. The changes are now merging into the GitHub specifications. They are not a new format spec. They are the missing trading vocabulary for formats the portfolio already named, and they touch three core standards at once: AdCOM, VAST, and the Native Ads specification.

This is the concrete deliverable behind a line in the Tech Lab agentic roadmap that most people skimmed past. The roadmap committed to completing the signals for programmatic trading in OpenRTB, VAST, and SIMID by Q2 2026. The CTV Ad Portfolio signaling is a large part of what that sentence meant. Q2 is now over, and the work shipped.

What the signaling actually adds

For years, connected TV advertising meant one programmatic shape: a linear video ad that interrupts the stream, served as a VAST tag with MediaFiles and tracking. The CTV Ad Portfolio names a set of formats that do not fit that shape. Pause ads appear when the viewer stops playback. Overlay ads sit in the lower third while content keeps running. Squeeze back ads shrink the program into an L or a frame to share the screen. Menu and screensaver ads live in the platform interface, outside video playback entirely.

Buyers and sellers already ran these as custom deals, trafficked by hand, because there was no standard way to say inside a bid which format was on offer. The new signaling adds that vocabulary. A seller can now declare, in a machine-readable field, that an impression is a pause ad rather than a mid-roll, and a buyer can bid on it as such. As IAB Tech Lab CEO Anthony Katsur put it, standard formats are only part of the equation, and you need a clear way to signal what is actually being bought and sold.

The formats getting a standard programmatic signal

• Pause ad: a static or video experience shown when the viewer pauses content with the remote.
• Overlay ad: a non-linear unit in the lower third or a corner while the program continues to play.
• Squeeze back ad: the program resizes into an L-shape, frame, or double box so an ad can share the screen.
• Screensaver ad: an OS or app-initiated experience shown during idle time.
• Menu ad: a unit placed in the home screen or navigation UI, outside video playback.
• In-scene insertion: branded elements composited into the content itself, closer to product placement than a spot.

This lands on the agentic buying stack immediately

Two agentic standards stacks are already in production, and both resolve a video buy to a VAST tag. IAB Tech Lab's AAMP wires OpenRTB, AdCOM, OpenDirect, and VAST together under agent interfaces. The Ad Context Protocol, now at 3.1, runs an agent-native pipeline from inventory discovery through creative production, and a video creative on that path ends as a VAST document handed to the sell side. Neither stack changes the endpoint. Every video transaction still resolves to InLine or Wrapper, MediaFiles, Impression pixels, and TrackingEvents.

The new signaling changes what the agents negotiate over. A seller agent will offer a pause ad, a buyer agent will bid on it by format, and the deal will clear machine to machine without a human reading the tag. The format is no longer an implied pre-roll. It is a declared contract with its own rendering behavior, and the VAST that fulfills it has to match the declaration.

A signal is a promise, not a proof

Standardized signaling solves the discovery and transaction problem. It does nothing about correctness. The AdCOM or VAST field that says pause ad is an assertion by the seller. It is not a check that the delivered payload is a well-formed pause ad, or valid VAST at all. The trade can clear on a signal while the tag underneath is malformed, mis-shaped for the format, or wrapped in a chain that drops the very signal that sold it.

This is the same gap the agentic protocols already carry, now widened by new surface. The protocol layer moves deal state and format signals between agents. The correctness of what actually ships is out of scope for the signaling, and out of scope for AAMP and AdCP alike. A new format vocabulary means new ways for the payload to disagree with the promise.

Where the new formats break at the payload

• A tag signaled as a non-linear pause or overlay ad but structured as a standard linear InLine, so the player renders it in the wrong slot or not at all.
• Interactive overlay and squeeze back formats that rely on SIMID for remote or QR interaction, a component with a documented history of failing validation.
• MediaFiles whose dimensions, delivery, or apiFramework do not match the format the impression was sold as.
• TrackingEvents mapped to a linear lifecycle that never fires for a non-interruptive format, silently losing measurement.
• Wrapper chains that resolve to an InLine stripped of the format signal or the universal ad ID, so the payload no longer proves what the deal promised.

Validate the artifact, not just the deal

The fix is not exotic. It is a validation step at the payload, before the tag enters a deal and before an agent commits to it. vastlint validates the VAST document these formats ship as: structure across VAST 2.0 through 4.3, MediaFiles, tracking, wrapper resolution, universal ad IDs, OMID, and SIMID. It checks what the tag is, independent of what a signal claims it is.

In an agentic loop that matters more, not less, because there is no human in the path to catch a payload that does not match its declaration. Run the vastlint MCP server inside a buyer or seller agent and the tag gets validated before the deal clears, in the same protocol the agents already speak. The signal tells the agents what was bought. Validation confirms that what ships is what was signaled.