New research from identity resolution vendor Adstra and independent media agency InterMedia Advertising has found that just 23 percent of residential IP addresses reached their intended geographic target, a gap that separate analysis from data validation firm Truthset now estimates costs the connected television advertising sector 7.4 billion dollars this year alone.

The Adstra-InterMedia study, first reported July 15, 2026, tested one of the advertising industry's most stubborn assumptions: that IP addresses can reliably identify a specific household. For years, media buyers treated IP-based CTV targeting as functionally deterministic, a precise pointer from an ad impression back to a real home. The new data suggests otherwise. According to the study, standard IP targeting missed its intended household in roughly three out of every four instances, while CTV identifiers, which function at the device level much like mobile advertising IDs, held steady over time at a considerably higher rate. Device-level identifiers proved 71 percent consistent month over month, compared with 57 percent for IP addresses, a 24 percent gap that Adstra says undercuts the accuracy claims long associated with IP mapping.

A black box at the center of programmatic buying

Andy Johnson, Adstra's chief data officer, has spent three decades inside identity companies, including stints at LiveRamp, Acxiom, and Experian, according to the study. That vantage point shapes his diagnosis of the problem. Identity vendors, he said, often compete on match rate or scale rather than on the accuracy of the underlying match itself, and buyers rarely see how that matching actually happens. The opacity, he suggested, is not incidental. It is structural, because vendors are not compensated based on how correct their matches turn out to be, only on how many matches they can produce or how large the resulting audience looks on a media plan.

That dynamic, Johnson argued, leaves advertisers unable to calibrate their own campaigns. Without visibility into how a given identifier was matched, a buyer cannot decide whether to prioritize reach over precision, since both outcomes look identical on the surface even though they imply very different tradeoffs for measurement and frequency management. IP addresses remain embedded across digital advertising far beyond CTV, touching functions from frequency capping to attribution modeling, which means inaccuracies of this kind ripple well past the connected television environment where the study concentrated its testing.

Adstra's study did not emerge in isolation. It builds on a body of research from Truthset, a separate data quality and measurement firm, that has been probing similar terrain for more than a year. A Truthset study published in February 2026 found that nearly 40 percent of all open auction CTV media spend was wasted due to inaccurate identity data, and an earlier Truthset study conducted on behalf of the Coalition for Innovative Media Measurement and Go Addressable turned up comparable shortcomings in IP mapping accuracy. Both firms have a commercial stake in the outcome: Adstra sells identity resolution products, and Truthset sells data quality and measurement tools, meaning both stand to benefit if advertisers shift more spend toward identity verification. Johnson has acknowledged that overlap directly. Even so, the underlying data in the new study came from a distinct third-party identity seller rather than from Adstra's own systems, a distinction the company has pointed to in arguing the exercise amounts to more than grading its own homework.

What the numbers show, linkage by linkage

Truthset's own contribution to this picture, published as its State of Data Accuracy 2026 report, breaks the inaccuracy down by specific data linkage rather than treating it as a single undifferentiated figure. Among identity signals, email-to-postal matching reached 51 percent accuracy, the strongest of the three pairings measured. IP-to-email matching came in at just 16 percent, and IP-to-postal matching, arguably the pairing most central to household-level CTV targeting, reached only 13 percent, a figure other industry voices had already cited as evidence that IP-based CTV targeting can miss the large majority of households. Demographic signals told a similarly uneven story. Gender segments proved 61 percent accurate, the most reliable of the demographic categories tested, while age segments collapsed to just 13 percent accuracy. Presence-of-children segments landed in between, at 41 percent.

Those linkage-level figures translate into a dollar estimate at the center of Truthset's report: 7.4 billion dollars in wasted U.S. connected TV programmatic ad spend in 2026, calculated by pairing Truthset's own accuracy analysis with eMarketer's U.S. digital ad spend estimates. Truthset frames total U.S. CTV ad spend for the year at 38.0 billion dollars, with 18.4 billion dollars of that flowing through open auctions specifically, the segment where the firm concentrated its waste calculation. Put differently, waste equivalent to roughly 40 percent of open-auction CTV spend traces back to data signals that do not hold up under independent validation, a figure consistent with the nearly 40 percent open-auction waste finding Truthset had already published in February.

Scott McKinley, founder and chief executive of Truthset, framed the stakes in stark terms in the report's executive summary. "Whether they realize it or not, advertisers are paying an accuracy tax levied at every stage of the data pipeline, resulting in both wasted impressions and opportunities," McKinley said. "If marketers can't rely on the data guiding their decisions, they're flying blind." He tied the pressure directly to the broader economic environment shaping 2026 media budgets. "As inflation pressures continue and new tariff policies tighten marketing budgets across industries, media buyers are scrutinizing every advertising dollar," McKinley said. "And marketers are moving spend toward deterministic channels where they can reach authenticated users."

McKinley's closing point in the report reached beyond individual campaign performance toward a structural concern for the open web as a whole. "For the programmatic open web to remain competitive with walled gardens, validated data is essential to ensure accurate signals that connect ads to real people and drive marketing profitability," he said. That framing echoes concerns McKinley raised in earlier reporting, where he warned that walled gardens, which do not face the same identity-matching problems as the open internet, stand to capture audiences and the ad spend that follows them if open-web publishers and platforms cannot close the accuracy gap.

A ratings system for the data supply chain

Truthset's response to its own findings has been a tiered rating structure for audience data, marketed as Data Rated Audiences. The system sorts records into four accuracy tiers, labeled AAA, AA, A, and B, and draws on what Truthset describes as 95 percent census-level coverage sourced from participating members of its Data Collective. In practical terms, the system lets a buyer filter an audience file by rating before activation, trading scale for confidence or vice versa in a way that is visible rather than opaque. A companion tool, the Audience Validation Test, allows a buyer to upload a sample file and see how many records match against Truthset's own reference data; in one illustrative test cited in the report's own materials, 10,000 uploaded records were matched against 9,000 confirmed records, a 90 percent match rate, though the accuracy of any single demographic attribute within that matched set varied considerably depending on which attribute was tested.

Adstra's answer takes a different shape. The company has built a framework called ID Connection Strength, which assigns a rating of one to six across five separate criteria, according to the study: recency, frequency, signal integrity or traffic quality, congruency, meaning how well signals about the same device align across different identity data categories, and cardinality, meaning how many other identifiers a given signal is tied to. Each score draws on user-consented signals from Adstra's own network and is checked by an outside fraud prevention firm that Adstra has not named. The framework guided the InterMedia research itself and has been in testing since January 2026 with a handful of brands, agencies, and platforms that Adstra has likewise declined to identify. Early results, according to the study, show that 71 percent of CTV IDs carrying a high ID Connection Strength score kept their link to the same person from one month to the next, though Adstra has cautioned it remains too early to say definitively whether the framework will meaningfully shift industry-wide matching accuracy. The system is currently tied to Adstra's own network, but Johnson has said the company is open to partnerships, including with industry trade groups, to help the framework gain wider adoption. Wider availability is planned for the third quarter of 2026.

David Nyurenberg, senior vice president of digital at InterMedia, argued the industry's response needs to go beyond any single vendor's tool. "We should stop treating identity quality as an assumption and start treating it as a measurable standard," Nyurenberg said, according to the study. "That's how we build greater confidence in targeting, measurement, attribution, and every AI system that increasingly depends on those signals." Johnson, for his part, offered a more direct instruction to advertisers evaluating identity vendors. "Demand more visibility in how matching is done," he said, "because for most partners, there isn't any today."

Consolidation reshapes who owns the identity layer

The findings land at a moment when ownership of the identity infrastructure Johnson describes is itself in flux. LiveRamp, one of the companies where Johnson previously worked, is the subject of a pending acquisition by Publicis Groupe, announced May 17, 2026 at a total enterprise value of 2.2 billion dollars. Acxiom, another firm on Johnson's résumé, has already changed hands through a different consolidation: Omnicom completed its acquisition of Interpublic Group on November 26, 2025, folding Acxiom's identity assets into the combined company. Both moves concentrate identity resolution capacity inside a smaller number of large holding companies at precisely the point when independent research is calling the underlying accuracy of that infrastructure into question.

The timing carries a practical implication for buyers. As deterministic identity products get absorbed into fewer, larger platforms, the visibility problem Johnson describes, where a buyer cannot see how a match was actually made, becomes harder to route around through simple vendor switching. If the leading identity providers are consolidating faster than they are opening up their matching methodologies, advertisers may find fewer independent points of comparison against which to benchmark accuracy claims of the kind Adstra and Truthset are both now publishing.

Why the accuracy question matters beyond CTV

Although both the Adstra-InterMedia study and Truthset's report concentrate their testing on connected television, the underlying identity infrastructure they examine extends well past that single channel. IP addresses, email hashes, and postal linkages sit underneath frequency management, attribution modeling, and incrementality testing across most programmatic environments, meaning the accuracy rates documented here plausibly understate the total scope of the problem rather than overstate it. Marketers evaluating identity vendors for campaigns running outside CTV face the same fundamental visibility gap Johnson describes, even where the specific numbers have not yet been measured and published.

For an industry publication audience specifically, the significance sits less in any single statistic and more in the pattern those statistics form together. Three independent research efforts, spanning Adstra and InterMedia in July 2026, Truthset in February 2026, and an earlier CIMM and Go Addressable-commissioned Truthset study, have each converged on the same conclusion: identity matching in CTV performs considerably worse than its deterministic reputation would suggest, and vendors have limited commercial incentive to volunteer that gap on their own. Whether newer scoring frameworks such as Data Rated Audiences and ID Connection Strength meaningfully close that gap, or simply add another proprietary layer to a supply chain already criticized for opacity, remains an open question that neither company's own data can fully answer.

Timeline

  • December 1, 2024: Truthset begins collecting data across six providers and three validation sources for a study commissioned by the Coalition for Innovative Media Measurement and Go Addressable.
  • February 28, 2025: That 90-day data collection period concludes.
  • January 2026: Adstra's ID Connection Strength framework enters testing with a handful of unnamed brands, agencies, and platforms.
  • February 2026: A separate Truthset study finds that nearly 40 percent of all open auction CTV media spend is wasted due to inaccurate identity data.
  • May 17, 2026: Publicis Groupe announces an agreement to acquire LiveRamp, a former employer of Adstra's chief data officer, at a 2.2 billion dollar enterprise value.
  • July 15, 2026: Adweek reports on the Adstra-InterMedia study, detailing findings that only 23 percent of residential IP addresses reached their intended geographic target and that CTV identifiers proved 24 percent more consistent over time than IP addresses.
  • 2026 (full-year estimate): Truthset's State of Data Accuracy 2026 report puts total U.S. CTV programmatic ad spend at 38.0 billion dollars, with 7.4 billion dollars of that wasted due to inaccurate data signals.
  • Third quarter 2026: Adstra's ID Connection Strength framework is scheduled to become more widely available.

Summary

Who: Adstra, an identity resolution vendor, and InterMedia Advertising, an independent media agency, conducted the study, which Adweek first reported; Truthset, a separate data validation firm, published the accompanying accuracy and waste figures; Andy Johnson, Adstra's chief data officer, David Nyurenberg, InterMedia's senior vice president of digital, and Scott McKinley, Truthset's founder and chief executive, provided on-record commentary.

What: New research found that only 23 percent of residential IP addresses reached their intended geographic target in connected television advertising, while device-level CTV identifiers proved 24 percent more consistent over time. Separately, Truthset estimated that inaccurate data signals will waste 7.4 billion dollars in U.S. CTV programmatic ad spend in 2026.

When: Adweek reported on the Adstra-InterMedia study on July 15, 2026. Truthset's State of Data Accuracy 2026 report and its associated waste estimate cover the current year, building on a separate Truthset study from February 2026 that found nearly 40 percent of open auction CTV spend wasted.

Where: The research concentrates on U.S. connected television advertising, including open auction programmatic transactions, though the underlying identity infrastructure examined extends across broader digital advertising environments.

Why: The findings matter because IP-based targeting has long been treated as a reliable, near-deterministic method for reaching specific households in CTV, a channel that continues to draw rising ad investment. Both studies suggest that assumption does not hold up under independent testing, with implications for how media buyers allocate budgets, measure frequency, and evaluate identity vendors, particularly as ownership of major identity platforms consolidates under fewer holding companies.