As connected TV advertising matures, the industry faces a fundamental tension: more data than ever before, yet persistent challenges in translating that abundance into clarity. From AI-driven identity resolution to real-time fraud detection protecting premium inventory, the tech is reshaping how advertisers measure and optimise CTV campaigns. We asked ad tech leaders how AI is solving CTV’s toughest measurement challenges in practice.
Erika Loberg, Global Head of CTV, OpenX
“The real challenge SSPs are solving is consistent access to quality signals — what those signals are, and which ones matter most, often vary by platform — and then reconciling them into insights that are actually usable and actionable. While more signals are available than ever, transparency alone doesn’t tell advertisers which signals matter or how they should be interpreted. This is where AI can help — cutting through the noise of existing, transparent data to surface the signals that translate into meaningful, quality outcomes.
“The industry also tends to conflate transparency with quality. Because quality means different things to different businesses, there’s no universal definition of what ‘quality’ means. What the ecosystem needs isn’t new frameworks, but broader adoption, consistent enforcement, and real accountability around standards that already exist.”
Julie Selman, SVP, Head of EMEA at Magnite
“While AI doesn’t replace measurement methodologies themselves, it ‘joins the dots’ between ad exposure, audience behaviour, and outcomes, making attribution scalable. AI can help uncover nuances, such as whether the first ad in a pod outperforms the others and if creative variation can drive better engagement. This has helped give buyers a clearer picture of how CTV actually drives performance across the funnel.
“SSPs are uniquely positioned to enable this shift because they sit at the closest point to premium supply, to connect high-quality, first-party signals at the point of impression. Using AI to normalise, enrich, and connect these signals allows SSPs to improve match rates, targeting precision, and clarity for buyers navigating signal loss. This proximity to publisher data also improves the quality of inputs for measurement and attribution partners, reducing discrepancies and making downstream models more accurate and actionable.
“As a result, “better measurement” in CTV today looks very different from two to three years ago: buyers now have faster feedback loops, stronger exposure-to-outcome modelling, and more confidence in reach and incrementality. It isn’t about a single new metric; it’s about continuity. With AI’s capabilities quickly scaling, advertisers can move beyond surface-level metrics, optimising and delivering campaigns that not only perform but also build stronger connections with consumers.”
James Milne, SVP Business Development, Epsilon
“AI is improving measurement and attribution in CTV by doing three things better than traditional approaches: resolving fragmented exposure into consistent people-based signals, modelling delayed cross-channel outcomes, and turning those predictions into live optimisation.
“First, AI-driven identity resolution links CTV ad exposure across apps, devices and household contexts back to a persistent profile. Without that stability, attribution is a game of averages, with duplicated reach and missing downstream actions. With it, marketers can connect exposure to subsequent behaviours elsewhere, whether on mobile, on-site, or in-store.
“Second, predictive models can estimate incrementality when the path to action is non-linear and time-lagged. Rather than relying on last-touch or simple correlation, AI can learn which sequences and frequencies are most likely associated with genuine uplift, and which audiences were likely to convert anyway.
“Third, those models can be used operationally. As signals arrive, AI can rebalance delivery, manage frequency at an individual level, and prioritise placements which are more likely to drive the next meaningful step.
“In retail-linked environments, this becomes measurable in business terms. Our own data shows that, when CTV exposure is matched to shopper outcomes, brands have seen conversions rise by 46% and in-store sales by 20%, creating a feedback loop that makes optimisation credible.”
Chris Kleinschmidt, VP EMEA Advertising Sales, Xperi Inc (TiVo)
“Ironically, AI is moving CTV back towards more probabilistic outcomes, rather than the deterministic, 1:1 approach CTV has always tried to drive. AI is currently trying to act as the connective tissue between fragmented data points. It’s moving us away from simple ‘last-click’ models toward sophisticated, multi-touch attribution by identifying subtle correlations in consumer behaviour that traditional manual analysis would miss. But by no means is this fully accepted or widely implemented, at the moment.”
“Historically, the biggest hurdle in CTV was the lack of a ‘cookie’— or a unified ID. AI has turned what used to be ‘noise’, like varying IP addresses and disparate device IDs, into actionable signals. Technically speaking this could allow us to account for co-viewing patterns and cross-device journeys with a level of statistical confidence that was previously unattainable.”
“Transparency in CTV is sometimes associated with sophisticated ad fraud (given the frothy CPMs). AI-driven filtration systems are now the standard for identifying ‘spoofed’ signals in real-time, ensuring budgets aren’t wasted on non-human traffic. Beyond fraud, it’s also improving signal quality by categorising content at scale, giving advertisers a clearer picture of exactly what environment their ads are appearing in. Compare this with linear TV— which never dealt with this issue— a brand could simply know when/what time their ads were airing.”
