IP-Based CTV Targeting Fails 3 In Every 4 Times, New Research Finds

The ad industry has for years billed IP-based connected television (CTV) targeting as precise and deterministic. But that promise is not backed up by reality, according to new research from identity resolution vendor Adstra and indie media agency InterMedia Advertising. 

In fact, standard IP targeting accuracy is so poor that just 23% of residential IP addresses reached their intended geographic target, per the study. Meanwhile, CTV identifiers, which, like mobile ad IDs are device-level, were 24% more consistent over time than IP addresses (71% compared to 57%, respectively). 

The findings suggest that IP-based mapping methodologies used by most of the leading providers in CTV and other environments are far less accurate than advertised. 

“Identity sellers often win tests based upon either match rate or scale. But the matching that is performed is opaque to the customer. It’s been a fundamental business problem since I’ve been in this industry,” said Adstra’s chief data officer Andy Johnson. He would know; over a career spanning three decades, he’s worked at some of the world’s leading identity providers, including LiveRamp (which Publicis intends to acquire for $2.2 billion), Acxiom (since folded into Omnicom as part of its acquisition of IPG), and Experian. 

This “black box matching,” where methodologies are opaque and identity vendors aren’t paid on the basis of accuracy, disadvantages advertisers and wastes media spend, Johnson implied. 

“If you don’t understand how matching is done, you have no ability to turn the dials to understand, ‘Do I want to drive up my scale so I can have more reach? Do I want to drive it down so I can have more fidelity for a measurement application?’” he said.

While the problem is pronounced in CTV where there aren’t persistent cookies, IP addresses are used across digital environments for everything from managing frequency to measuring attribution and incrementality.

A proposed remedy

The Adstra-InterMedia thesis is supported by other industry research; a Truthset study published in February found that nearly 40% of all open auction CTV media spend is wasted due to inaccurate identity data. Another Truthset study from last year, conducted on behalf of the Coalition for Innovative Media Measurement (CIMM) and Go Addressable, found similar shortcomings in IP mapping accuracy. 

It’s worth noting that both Adstra and Truthset have skin in the game: Adstra sells identity resolution products, while Truthset offers data quality and measurement tools. Both could theoretically benefit if advertisers move away from IP-based targeting and invest more in identity accuracy, something Johnson openly admitted. 

Johnson said the study isn’t just a case of Adstra grading its own homework, because the data reviewed came from a different third-party identity seller. Even so, it could be said that Adstra is seeking to diagnose a problem in the industry while simultaneously peddling a remedy. 

That proposed remedy is a new framework from Adstra called ID Connection Strength (ICS), designed to rate how reliably a given digital identifier (whether an IP address, a CTV ID, a cookie, or a mobile advertising ID) is accurately linked to a real person or household. 

The framework assigns a 1-6 rating across five distinct criteria: recency, frequency, signal integrity (or traffic quality), congruency (the extent to which various signals about the same device match across different identity data categories), and cardinality (how many other identifiers a given signal is linked to). Each score is determined using user-consented signals from Adstra’s network and is validated by a third-party fraud prevention firm, though Adstra declined to name them.

ICS was used to guide the research with InterMedia. It has been in testing since January with a handful of brands, agencies, and platforms, who are using it to help validate identity data, improve audience targeting, and adjust media budget allocations. Adstra declined to name these clients and partners. 

ICS will become more widely available in the third quarter. While it’s tied to Adstra’s network for now, Johnson said the company is “open” to partnerships, including collaboration with industry trade groups to help seed adoption 

The industry needs to become “more honest…about the confidence we have in the signals we’re using,” said David Nyurenberg, svp of digital at InterMedia. “We should stop treating identity quality as an assumption and start treating it as a measurable standard. That’s how we build greater confidence in targeting, measurement, attribution, and every AI system that increasingly depends on those signals.”

It’s too early to tell whether ICS will meaningfully move the needle on ID mapping accuracy, but early signals are promising; the team claims that 71% of CTV IDs with high ICS scores maintained their connection to the same person month over month. 

For the time being, Johnson has some simple advice for advertisers: “Demand more visibility in how matching is done, because for most partners, there isn’t any today.”