Exactly who is watching connected TV? Well, that’s hard to determine

The connected TV advertising market has grown into a multibillion-dollar business built on a promise: that streaming can deliver the reach of traditional television with the targeting precision of digital.

But multiple executives across the adtech and streaming supply chain have noted the industry still cannot reliably identify who is on the other side of the screen.

The problem is not a lack of data. CTV platforms generate vast amounts of viewer information.

The problem is that the identity signals underpinning audience targeting, measurement and attribution remain inconsistent across platforms, devices and households. That inconsistency means advertisers may be paying for impressions that look efficient on paper but reach the wrong people in practice.

Justin Rosen, SVP of data and insights at Ampersand, said the issue becomes more visible as inventory increases across the CTV landscape.

“As more inventory enters the market, lower CPMs can create the appearance of efficiency, but that doesn’t always hold up under scrutiny,” Rosen said in the Industry Insights roundtable on streaming, FAST and CTV strategy. “If identity signals are unreliable, advertisers may be reaching the wrong households more often than they realize, which erodes the value of those impressions.”

That erosion is not hypothetical. It affects how campaigns are planned, how results are reported and how much confidence advertisers place in CTV as a media channel.

When identity breaks down, every layer of the advertising process that depends on it, including targeting, frequency capping, attribution and measurement, inherits the same uncertainty.

Advertisement

The device ID problem inside a single household

At the device level, the identity challenge is compounded by how connected TVs handle user identification. Unlike mobile devices or web browsers, where identity frameworks have had years of development and iteration, CTV operating systems vary widely in how they assign and manage device identifiers.

Yang Cai, CEO and president of VisualOn, said the fragmentation exists not just across platforms but within individual homes.

“Fragmented device IDs within a single household make cross-platform attribution an unsolved challenge,” Cai said. “Additionally, discrepancies between buyer and seller reporting persist, even as IAB standards improve impression-level transparency.”

A household with a Roku in the living room, a Samsung TV in the bedroom and a Fire TV in the kitchen presents three different identity environments. Each platform handles device identifiers differently, and none of them share a common framework for recognizing that all three devices serve the same household. For an advertiser trying to control how many times a household sees a particular ad, that fragmentation makes frequency management unreliable.

The buyer-seller reporting discrepancy Cai described adds another layer.

An advertiser’s demand-side platform and a publisher’s supply-side platform may count and attribute the same impressions differently, producing reports that do not reconcile. The result is a market where both sides of a transaction can look at the same campaign and arrive at different conclusions about what it delivered.

Measurement built on an unstable base

The identity problem does not exist in isolation. It sits underneath the measurement frameworks the industry uses to evaluate campaign performance, and those frameworks reflect the instability.

Rachel Herbstman, VP of data innovation at Ampersand, said the misalignment across identity systems makes it difficult to connect ad exposure to outcomes in any consistent way.

“One of the biggest gaps comes down to the instability of identity across platforms,” Herbstman said. “Advertisers may think they’re reaching the same audience across environments, but in reality, those identity frameworks often don’t align, making it difficult to consistently connect exposure to outcomes.”

That disconnect has practical consequences for how advertisers allocate budgets.

Cross-platform campaigns, which are increasingly common as viewers move between linear, streaming and FAST environments , depend on the ability to deduplicate audiences and measure incremental reach. When identity signals do not persist across those environments, reported reach numbers may overcount unique viewers, and attribution models may assign credit to impressions that did not actually influence a purchase or action.

Herbstman added that measurement and attribution will continue to reflect approximations rather than reality until the industry addresses the identity foundation.

Advertisement

The word “approximations” is key. It suggests that current CTV measurement, despite its volume of data points and reporting dashboards, operates with wider margins of error than the precision-oriented language of programmatic advertising typically implies.

Standards efforts and their limits

The industry is not unaware of the problem. Several standards bodies and industry groups have pursued initiatives to improve measurement consistency and transparency across CTV environments.

Paul Davies, head of marketing and partnerships at Yospace, said measurement sits at the center of unlocking ad value but remains difficult to execute reliably at scale.

“A more standardized approach to measurement is needed to deliver consistency and build advertiser confidence,” Davies said. “Emerging standards like the SVTA Ad Creative Signalling and the latest iteration of CTA’s Common Media Client Data (CMCDv2) for Client-Guided Tracking seek to tackle the measurement challenge and create a more transparent ad ecosystem.”

Those standards address specific technical layers of the measurement stack – how ad creative is signaled, how client-side data is collected and shared. They represent progress on individual components of the problem. But identity resolution sits beneath those layers, and no single standard currently addresses how a viewer is recognized consistently across Samsung, Roku, LG, Amazon and Apple TV environments.

Cai noted that most measurement frameworks were originally built around browser and mobile app environments and then adapted for CTV. Device identifiers on connected TVs are inconsistent, background processing behavior varies by manufacturer and there is no universal standard for what constitutes a completed view across platforms.

“This makes cross-platform de-duplication genuinely hard and means reported reach numbers should carry wider error bars than most reports acknowledge,” Cai said.

What advertisers are starting to demand

The gap between what CTV measurement currently provides and what advertisers need to justify their spending is widening as budgets grow. Several respondents described a shift in advertiser expectations from basic delivery metrics toward verified audience identification and outcome-based reporting.

Rosen said the focus is moving toward more stable identifiers that can persist across screens.

“Broadcasters are starting to prioritize more stable, household-level identifiers that can persist across screens and environments,” Rosen said. “That shift helps reduce fragmentation and creates a more reliable foundation for both targeting and measurement.”

Advertisement

Household-level identification represents one potential path forward.

Rather than attempting to identify individual viewers on each device, the approach treats the household as the addressable unit, a model closer to how traditional television has always operated. The challenge is building that identification layer in a way that works across platforms that have no commercial incentive to share data with one another.

Herbstman said the emphasis needs to shift from impression volume to impression quality.

“Finding the right balance means focusing less on volume alone and more on ensuring those impressions are actually meaningful,” Herbstman said.

An infrastructure problem, not a reporting problem

The identity challenge in CTV advertising is often framed as a measurement or transparency issue.

But the responses from executives across the supply chain point to something more fundamental: the infrastructure on which CTV advertising operates was not designed with consistent identity resolution as a core requirement.

That design gap means improvements in ad serving, creative delivery and reporting standards, while valuable, cannot fully compensate for the uncertainty at the base layer. Until the industry establishes a more stable and interoperable identity framework across CTV platforms, the precision that programmatic advertising promises will remain ahead of what it can consistently deliver.

For advertisers increasing their CTV budgets, the implication is that the metrics they receive may look more definitive than the underlying data supports. For broadcasters and platform operators preparing their ad stacks for programmatic and addressable demand, the quality of the identity layer they build will likely determine whether that demand translates into sustainable revenue or a race to the bottom on CPMs.