David Simutis
When CTV inventory turns out to be fake, mobile app-classified inventory is almost always where it's hiding. Peer39's benchmark data shows that approximately 98% of Fake CTV volume originates from mobile app-classified inventory, accounting for 16.03 percentage points of the 25.2% total Fake Content share in open exchange CTV. This isn't spread evenly across the bidstream. It's concentrated in a small number of recognizable patterns.
Mobile app inventory drives nearly all Fake CTV
"Mobile app-classified inventory" are apps that declare themselves as CTV inventory without behaving like a streaming service. They might pass technical filters because the operating system checks out. What they don't pass, when anyone actually compares what they are to what a buyer expects, is a basic test of whether the person on the other end is watching streaming programming.
That distinction matters because of scale. Open exchange CTV bid requests classified as Fake Content run at 25.2% overall, and the mobile app share alone accounts for the overwhelming majority of that number.
Three patterns behind the number
Peer39's analysis identifies three recurring shapes this inventory takes:
The utility trap. Remote control apps, screen mirroring tools, file managers, and flashlight apps that pass technical filters because they run on a TV operating system. The person using them is configuring a device, not watching content, but the ad still serves and still completes.
The ambient void. White noise, fireplace, and aquarium apps, along with dynamic wallpapers, that post high completion rates and abnormally long session durations because they're designed to run unattended, often while the viewer is asleep.
The mobile leak. Casual games like solitaire or match-three titles ported to TV interfaces. A video ad here is a rewarded interruption the player sits through to get back to the game, not content they chose to engage with.
Different mechanics, same outcome: an ad serves, a completion event fires, and the budget moves, with no actual television viewing behind any of it.
Why completion rate can't tell the difference
This is the part that makes the problem hard to spot from a dashboard. Mobile app-based inventory misclassified as CTV often delivers completion rates near 99%, because the viewer is captive rather than engaged. Authenticated, real programming posts similarly high completion rates, drama at 98.39%, comedy at 98.74%, action and adventure at 96.65%, but for the opposite reason: people are actually watching.
The same number means two different things depending on what's behind it, and completion rate alone has no way to distinguish which one a buyer is looking at.
What the fix looks like
Two checks catch what completion rate misses, and one control removes most of the exposure:
Device and screen mix. Authenticated CTV delivery runs above 95% on actual TV screens. If tablet or mobile device share climbs above 1 to 2% of a line item's "CTV" delivery, that's a signal the buy includes mobile app leakage.
Channel name field. Identifiable, authenticated channel names correlate with genuine TV inventory. Missing, generic, or numeric channel values correlate strongly with Fake CTV.
Pre-bid categorization. Peer39's Q4 2025 benchmarks show that campaigns using pre-bid categorization see Fake CTV exposure fall from 25.2% in the open exchange to 2.21%, a reduction of approximately 92%.
Before the next optimization cycle
Peer39 reporting shows the first two fields to check: device type and channel name. If either one shows a pattern consistent with mobile app leakage, that's exposure a pre-bid filter can remove before the next dollar goes out the door, not after the report comes back.
