Web brand safety assumes a page. CTV doesn't have one.
Buyers apply the same "safe" label built for the open web to CTV, assuming it means the same thing everywhere. It doesn't, and the gap between what a label claims and what it verifies is the subject of this report.
Web brand safety reads a page. CTV has no page, only sparse, self-declared bidstream metadata. In a single-title case study, 58% of tracked impressions passed only one genre tag, and none carried the complete authenticated set.
The gap holds at scale: 44.7% of rated feature-film impressions in CTV carry an R rating that says nothing about what's actually on screen, and 10.7% of impressions can't be confirmed valid or invalid at all.
This report traces where the gap opens, how far it compounds, and what closes it.
Inside the report, you'll find:
Why a "safe" classification built for the open web has nothing to read in CTV, and why one narrow genre tag (or a blank field) is what actually reaches the DSP
Why nearly half of rated feature-film inventory in CTV carries an R rating that describes format, not content, and how a series-level rating can hide meaningful per-episode swings
Why Sensitive Content and user-generated content clear self-declared metadata at multiples of the rate they show up in authenticated programming, and why a meaningful share of CTV traffic can't be confirmed valid or invalid at all
A practical, four-step framework for verifying "safe" with authenticated, program-level signals, before the bid is placed
