Ask 5 CTV sellers for a delivery report and you get 5 reports. Impressions counted at different moments, completion defined differently, some lines filtered for invalid traffic and some not, and almost nobody telling you whether the television was on when the ad played. Putting those side by side is a day of spreadsheet work, and the answer is still a guess.
IAB Europe's Connected TV Measurement Framework and Transparency Principles is an attempt to fix the vocabulary. The draft opened for public comment on 30 April 2026, with comments accepted until 12 June. The final version followed on 1 October 2026.
Why it exists
The trust gap is measurable. In IAB Europe's own research on CTV in Europe (2024, a survey of 422 advertiser and publisher executives across 29 markets), only 30% reported full visibility into where their ads run, fewer than half used quality verification tools, and 27% lacked consistent insight into brand suitability. IAB Europe cites those figures as the reason for the framework.
The same gap shows up in outcome reporting. Research from a CTV buying platform, reported in The Drum, found that only about a third of marketers fully trust the performance numbers platforms report, and more than 70% said they would invest more if the proof were stronger. Whatever the exact share, buyers hold budget back when they cannot check what they bought.
The framework was shaped in workshops with buy-side and sell-side companies, among them the broadcaster sales body EGTA, IAB UK, Samsung Ads and the sales houses of MFE ProSiebenSat.1, Publitalia, RMB Belgium and RTÉ.
What is in it
The final framework sorts CTV metrics into 3 layers.
Foundational delivery. Ad impressions, filtration of general and sophisticated invalid traffic, viewability, and TV-off detection: whether the television was powered on when the ad was delivered.
Exposure and audience. Reach, device identification, view-through rate, video completion rate, and information about the programming around the ad.
Outcomes. Conversions, return on ad spend, incrementality, brand uplift, ad recall and attention.
The point of the layers is the dependency between them. If the foundation is wrong, everything built on it is wrong too. A ROAS figure calculated on impressions that include invalid traffic, or ads that played to a switched-off screen, overstates the result no matter how good the attribution model is.
Alongside the metrics sit 5 transparency principles: definitions that are unified and aligned with existing standards, full disclosure of measurement limitations, support for data collection initiated by the advertiser, visibility into device power state as a default, and granular reporting.
The framework is voluntary. It sets out what a seller should report and disclose; it does not certify anyone, and the public material we have seen does not set new numeric thresholds (for example a viewability standard specific to CTV). Read it as a shared checklist, not a rulebook.
What changes in negotiations
The useful shift is that buyers and sellers now have a common reference to point at. A few places where that shows up in a deal:
Definitions go into the insertion order. Instead of accepting "impressions per the seller's ad server", the buyer can ask which framework definitions the seller reports against and write them into the IO. The publisher states its methodology and its limits once, up front, rather than defending it after the flight.
TV-off becomes a line in the report. Power-state visibility is listed as a default. A buyer can reasonably ask for it on every line, and a publisher that can supply it has a concrete way to show its inventory is watched.
Outcomes are discussed after the foundation. If a seller offers incrementality or attention metrics, the first question is what the impressions underneath were filtered for. That puts publishers who already filter and disclose at an advantage, which is fair.
Buyers can bring their own measurement. Support for advertiser-initiated data collection means the seller should accept the buyer's own tags or measurement partner, so the buyer is not limited to the seller's report.
Say a buyer is splitting a EUR 200,000 CTV budget across 4 publishers. Before the framework, each line came back on its own terms and the shift of budget between them mid-flight was argued from incompatible numbers. With the same definitions on every line, the trader can compare delivered impressions, TV-off rates and completion side by side, and move money with the client's confirmation on a basis both sides recognise.
Why it matters outside Europe
Nothing obliges a seller in Singapore, Jakarta or Sydney to follow an IAB Europe framework. It still matters there, for 2 reasons.
Regional buyers often work for global advertisers whose European teams will start asking for these metrics, and those requests travel. And for markets with no local CTV measurement standard, the framework is a ready-made list of what to ask for. A buyer in Southeast Asia can send it to a streaming publisher today and ask which items it can report. A publisher can answer the same list and use it to show what it already does well.
Where agents fit
A framework is only useful if the agreed definitions follow the deal from plan to final report. When a buying agent and a selling agent trade through a shared record, the metric definitions sit on the confirmed plan, and delivery is read against them line by line, shortfall first. Skopa and Fanera work from that record; the CTV insertion order guide covers which terms to write down before a flight starts.