Two separate research releases this week put hard numbers on something buyers have suspected for a while: the identity layer underneath digital advertising is not performing anywhere near its sales pitch. Adstra research found that only 23% of residential IP addresses reached their intended geographic target — IP-based CTV targeting misses roughly three times in four. A separate Truthset study estimates nearly 40% of open-auction CTV media spend is wasted on inaccurate identity data. And an agency white paper reports match rates have degraded below double digits — under 10% for at least one client. Read those together and the addressable-advertising premium starts looking like a rounding error on a coin flip.
01What the research found
Adstra’s comparison is the most actionable piece. Device-level CTV identifiers proved 24% more consistent than IP over time — 71% versus 57% persistence. Adstra CDO Andy Johnson names the market failure directly: “Identity sellers often win tests based upon either match rate or scale. But the matching that is performed is opaque to the customer.” In other words, the industry has been buying on a metric nobody can audit. Adstra’s proposed answer is ID Connection Strength (ICS), scoring identifier reliability across recency, frequency, signal integrity, congruency and cardinality, widely available in Q3; it reports 71% of CTV IDs with high ICS scores kept the same person association month over month.
The agency side is blunter. Idea Peddler founder/CEO Cimin Ahmadi Cohen, writing with PJX Media, Share Local Media and Locality in a white paper titled “The Attribution Illusion: Inside the $1 Trillion Crisis Transforming the Future of Digital Advertising,” says the match-rate decline has already changed her buying: “we do very, very little pure programmatic display anymore because it’s just deeply ineffective.” Her recommended replacements are premium CTV, contextual targeting, brand-lift studies and MMMs — measurement that doesn’t require resolving a person to make a decision.
Be appropriately skeptical of all of it. Adstra and Truthset are identity and data-quality vendors publishing research that implies you should buy identity and data-quality products. The white paper is agency-authored. These are directional findings with commercial motives attached, and the Adweek dates here are derived from relative timestamps and are ±1 day.
02Why marketers should care
| You are paying an addressability premium for accuracy that is not there | Adstra found only 23% of residential IP addresses reached their intended geographic target: IP-based CTV targeting misses roughly three times in four. The audience CPM uplift on that inventory is priced against a precision the data does not support. |
|---|---|
| Match rates below 10% change what a segment means | A segment you can only match into single digits is not the audience on the insertion order. It is a small, non-random slice of it, and its performance says little about the audience you meant to reach. |
| The failure is specific to probabilistic matching | Deterministic, consented, logged-in relationships do not degrade this way. That is a real distinction to price, and most rate cards do not make it. |
The identity layer was sold as precision infrastructure and is measuring out as an expensive approximation.
03What marketers should do
04The bottom line
The identity layer was sold as precision infrastructure and is measuring out as an expensive approximation. That’s genuinely bad news for anyone whose business model is audience arbitrage, and genuinely good news for publishers who own real reader relationships and real editorial context. The numbers stopped adding up on probabilistic targeting some time ago; this week is just the first time the industry wrote them down.