A new study out this month should worry anyone buying CTV programmatically. Adstra and InterMedia Advertising just published research showing that roughly 40% of every open auction CTV dollar is wasted because of bad identity data. Not fraud. Not brand safety. Just identifiers that don’t actually map to a real person or household anymore.
The number that stopped me was the IP one. When campaigns targeted geography using only residential IP addresses, just 23% of impressions actually landed in the intended market. That’s not a rounding error. That’s most of a media budget missing its target before you even get to frequency, brand safety, or attribution.
The Industry Built a New Currency Without Auditing It
CTV identity has always been messier than desktop or mobile. There’s no single login, no persistent cookie, and the identifiers in play (CTV IDs, household IPs, hashed emails, UID2s, MAIDs, TTD cookies) each decay at different rates and get stitched together by different vendors with different logic. Everyone in the chain assumes the upstream identity graph did its job. Increasingly, it didn’t.
That’s the real story behind Adstra’s launch of ID Connection Strength, a scoring framework that rates identity links across five dimensions, recency, frequency, signal integrity, congruency, and cardinality, on a 1 to 6 scale. The interesting part isn’t the framework itself. It’s that a major identity vendor felt the need to build a transparency layer on top of its own graph. That’s an admission that “we have an identity match” and “we have a good identity match” have not been the same claim for a while, and buyers had no way to tell the difference.
The study’s stability numbers back this up. CTV IDs with high connection strength scores held their link to the same person 71% of the time across a measurement window, a 24 point premium over legacy IP based mapping. That gap is the whole ballgame for anyone running frequency capping, sequential messaging, or attribution off device level identity.
Why This Is Showing Up Now
Three things are colliding at once. First, CTV inventory volume has outpaced the identity infrastructure built to support it, so more dollars are flowing through graphs that were sized for a smaller, more deterministic world. Second, the post cookie shift pushed more of the ecosystem toward probabilistic and household level signals, which are useful but need to be validated, not assumed. Third, buyers have gotten more sophisticated about asking for proof, which is exactly why a vendor felt commercial pressure to publish a quality score instead of just a match rate.
None of this means identity resolution is broken. It means unvalidated identity resolution is expensive, and most of the market has been treating a match as binary when it’s actually a confidence interval.
What This Means for Buyers
If you’re running CTV programmatically, a few practical takeaways come out of this research. Ask your identity partner for a quality score, not just a match rate, on every identifier type in your stack. Treat household IP targeting as directional, not precise, unless it’s been cross validated against another signal. And build fraudulent or low quality ID filtering into your pipeline before activation, not as a post campaign audit. Catching a bad match after the impression served is a wasted impression either way.
This is also where a validated identity graph earns its keep. At BDEX, this is the exact problem our identity resolution work is built around, connecting MAIDs, hashed emails, and household IPs to a real, verified person rather than assuming a probabilistic link is good enough. Filtering out fraudulent or low confidence IDs before they touch an audience isn’t a nice to have feature. Based on this research, it’s close to half the budget.
The Takeaway
CTV isn’t going to get less fragmented. More devices, more identifiers, more vendors stitching graphs together. The winners over the next few years won’t be the platforms with the biggest reach claims. They’ll be the ones who can tell you, with evidence, how confident you should be in the identity behind every impression. A 40% waste rate is a wake up call, but it’s also a roadmap. The industry now has language and data to demand better identity, and buyers who start asking for it will see it in their numbers before their competitors do.
