Industry Insights: Connecting viewer behavior with content performance and business outcomes

Audience measurement is moving beyond traditional ratings and viewership totals as broadcasters and streaming platforms look for deeper connections between audience behavior, content performance and business outcomes.

Metrics such as retention, engagement, content visibility, and incremental reach are becoming increasingly important, while predictive analytics and AI offer new ways to anticipate demand and identify improvement opportunities. But understanding why content succeeds, rather than simply reporting how it performs, requires connecting audience insights with operational, financial and contextual data.

In part two of this Industry Insights roundtable, industry vendors explore how analytics inform programming, distribution and monetization decisions. The discussion examines the role of predictive modeling, the relationship between viewer experience and content performance, and the growing need for transparency, governance and collaboration as audience intelligence becomes more sophisticated.

Industry Insights: Closing the gaps in cross-platform audience measurement

Key takeaways from this Industry Insights roundtable

• Beyond traditional ratings: Media organizations are increasingly evaluating content through retention, incremental reach, on-screen visibility and business outcomes rather than relying solely on total views and audience size.
• Predictive analytics expands: AI and predictive models are helping organizations anticipate audience demand, optimize content placement and identify operational issues, although forecasting creative success remains more difficult.
• Context explains performance: Combining engagement data with playback quality, content metadata and distribution information helps distinguish programming weaknesses from technical problems and identify why particular content succeeds.
• Connecting business decisions: Integrating audience insights with content rights, distribution costs and advertising performance creates opportunities to improve licensing negotiations, promotional strategies and monetization.
• Transparency becomes essential: As measurement grows more granular and AI-driven, organizations need clear methodologies, explainable models and appropriate privacy safeguards to ensure audience intelligence remains reliable and actionable.

Which metrics are becoming most valuable for evaluating content performance beyond traditional ratings and total views?

Francesca Pezzoli, VP, marketing, Looper Insights: Beyond viewership totals, the industry is increasingly treating on-screen placement data as a performance signal in its own right — if a title isn’t visible on connected TV home screens, it’s unlikely to be discovered, let alone watched. Metrics that capture how prominently a title appears across CTV interfaces — its position, exposure duration, share of visible real estate and estimated impressions — are proving to be strong leading indicators of audience demand rather than lagging measures of it.

Peter Nummerdor, SVP, product management, VideoAmp: Reach and frequency remain important, but they’re increasingly being considered alongside measures of incremental reach and business outcomes. As viewing fragments, a large aggregate audience number can hide whether a media strategy is reaching new people, repeatedly reaching the same people or ultimately contributing to an advertiser’s objective. The next phase of measurement is about connecting audience exposure more closely to the value that exposure creates.

What approaches are helping customers measure retention, churn and long-term audience value?

Mark Paramor, director of operations assurance, Accedo: As the streaming ecosystem has grown more complex, services are having to adopt more sophisticated approaches to measurement around retention patterns, churn signals and viewer engagement. Churn analytics need to be combined with behavioral modeling, and lifecycle data to anticipate at-risk users and prioritize interventions. Quality-of-service monitoring and observability with advanced analytics have become critical business functions. Accedo Insights enables service providers to measure key components of their video service infrastructure, assessing how different components are performing and identifying bottlenecks that may be having a negative impact on performance and user experience.

How can audience analytics account for the different goals of live programming, on-demand content, FAST channels and short-form video?

Josh Heller, chief revenue officer, Coactive: They need different clocks, not different metrics. On demand you can afford to analyze after the fact, because the library is stable and the decision can wait. Live does not give you that, so the analysis has to happen while the content is still running, which is a harder engineering problem than most analytics stacks were built for and the main reason live gets measured more crudely than it deserves.

How can analytics platforms present complex audience information without creating additional noise or slowing decision-making?

Mark Paramor, director of operations assurance, Accedo: The key is to surface the data that matters, and that supports KPIs, and at the same time, ensure there’s enough context to understand what’s driving the data. That said, it’s still not always immediately obvious what changes are needed to deliver the desired outcome. Additionally, knowing how and when to implement change without inadvertently impacting other core areas such as performance, cost or user experience, can also be difficult. On top of that, even when video providers have the right data to hand and they know what action to take, they still need to move quickly.

Advertisement

This is where AI can add real value: AI orchestration and agent automation are becoming foundational layers that can accelerate time-to-insight, directly impacting business KPIs like churn, ARPU, and LTV. Modular, composable platforms also make it easier to act on those insights quickly without creating knock-on effects on other parts of the service. This shift not only enables agility, but also significantly lowers integration costs.

What role is predictive analytics beginning to play in forecasting content performance or audience demand?

Francesca Pezzoli, VP, marketing, Looper Insights: Predictive analytics is starting to close the gap between campaign planning and real-world outcomes by using historical placement and visibility data to model likely performance before a title goes live. On connected TV interfaces specifically, knowing which positions and page types have historically converted to viewership — and what those placements are worth as media value — lets distribution and marketing teams make smarter, evidence-based decisions rather than educated guesses. The shift is from reacting to what happened to anticipating what will.

Reinhard Grandl, chief product officer, Bitmovin: The reliable wins so far are operational, from predicting concurrency for a live event, effectively pre-positioning capacity, to catching a quality regression before support tickets arrive. Forecasting what an audience will want on the other hand is much harder and still mostly a content judgement with data attached. At Bitmovin, we see more value in developing models that shorten the reaction time on live streams, than in models that try to pick the next hit.

Peter Nummerdor, SVP, product management, VideoAmp: Predictive analytics is shifting audience intelligence from explaining what happened toward helping teams determine what to do next. With the right data foundation, traditional models and agentic systems can identify patterns across large datasets and surface signals that would be difficult for analysts to find manually, helping teams make faster decisions about audiences and media performance. The important distinction is that AI should support decision-making with transparent, reliable data rather than create another black box for teams to interpret.

How can technology help organizations identify why a program performed well, rather than simply reporting that it did?

Francesca Pezzoli, VP, marketing, Looper Insights: Predictive analytics is starting to close the gap between campaign planning and real-world outcomes by using historical placement and visibility data to model likely performance before a title goes live. On connected TV interfaces specifically, knowing which positions and page types have historically converted to viewership — and what those placements are worth as media value — lets distribution and marketing teams make smarter, evidence-based decisions rather than educated guesses. The shift is from reacting to what happened to anticipating what will.

Reinhard Grandl, chief product officer, Bitmovin: If quality-of-experience data is viewed alongside engagement data, video services get a much deeper understanding of what’s really going on. For example, a drop-off at minute four means something different when startup time was six seconds than when playback was clean. A lot of what gets read as weak content is actually a delivery problem, and separating the two is one of the most useful things video analytics can do right now.

Josh Heller, chief revenue officer, Coactive: You cannot explain performance with data that stops at the title. When content is indexed at the moment and scene level, you can look at what was actually in the episodes that overperformed, the talent, the settings, the story beats, the pacing, and test whether those attributes hold across the rest of the catalog. That turns a post-mortem into something repeatable.

How are audience insights being connected with content metadata, rights information and distribution costs to support better decisions?

Reinhard Grandl, chief product officer, Bitmovin: Much of this work is being done slowly, and usually by hand, with teams having to pull information from different systems and bring it together themselves. The interesting shift is that viewing data is starting to be joined with what each stream costs to encode and deliver, so a title can be judged on contribution rather than views alone.

Rights is what’s usually missing. Territory, window and exclusivity are likely in some contract system that nobody has joined to the audience insights, so a title gets judged on performance without the same people knowing where it was legally available or for how long. The rights holders doing this well can see a title’s cost to serve, its rights window and its viewing all in one place. That makes a rights renewal decision or negotiation data driven. Once you can see how a long-tail catalogue title performs and what it costs to serve, decisions about encoding profiles and windowing get a lot easier.

What examples have you seen of audience data directly changing a customer’s content or platform strategy?

Francesca Pezzoli, VP, marketing, Looper Insights: One pattern we see regularly is distribution teams using interface visibility data to renegotiate promotional agreements — if the data shows a title received materially less on-screen prominence than contracted, that’s a measurable discrepancy with direct commercial implications. We’ve also seen studios shift their release prioritization based on which streaming services consistently deliver stronger placement for comparable titles, moving away from anecdotal partner feedback toward data that’s objective and auditable. In both cases, the shift is the same: from assumption to evidence.

Paul Davies, head of marketing, Yospace: Live streaming with one-to-one ad addressability at scale is challenging. Take the 2026 FIFA World Cup, where viewers go to ad breaks simultaneously, and the timing of those breaks depends on the state of play. An intelligent prefetch solution can use real-time audience data to anticipate demand, make ad requests earlier, and pace them over a longer window, giving the ad tech more time to trade and respond and helping avoid unfilled ad spots.

Real-time performance data is equally important once those ads are being delivered. A live dashboard provides ad operations teams with real-time visibility into ad performance, enabling them to identify issues and respond on the demand side.

Advertisement

How can vendors help connect data teams with programming, editorial, product, advertising and executive stakeholders?

Mark Paramor, director of operations assurance, Accedo: Many media companies are still operating with legacy infrastructure, and with siloed teams and data, isolated initiatives and reactive product cycles. This mode of operation is not sustainable in the AI era. Data needs to be useful and accessible across the entire organization. This starts with agreeing shared KPIs, so different teams are all working towards the same outcomes, while still being able to see the data through the lens most relevant to their role. Clear ownership and feedback loops are also important, so insights can lead to actions, and the results of those actions can feed back into a process of continual evolution.

First, streaming companies need to shift to architecture that enables OTT platforms to continuously evolve, yet this move requires deep operational expertise and structured, scalable foundations. Building and evolving modern OTT platforms takes more than integration capability; it requires deep understanding of how streaming architectures behave under continuous change.

Paul Davies, head of marketing, Yospace: Vendors can help by turning measurement from a technical function into evidence of business value. Ad insertion alone doesn’t generate sustainable revenue. Broadcasters also need to prove whether an ad was seen, heard, completed, and engaged with, and make that information meaningful to teams beyond ad operations. A thoughtful measurement strategy creates a shared source of insight that product, advertising, and executive teams can use to understand and improve monetization.

What governance, privacy and transparency considerations should organizations address as audience measurement becomes more granular?

Reinhard Grandl, chief product officer, Bitmovin: Streaming services capture a huge number of data points and signals, but many of these are not used to make decisions or trigger actions because there simply isn’t enough context to understand what the data means. To remove the noise that makes analysis overly complex and that also risks privacy infringements, video platforms should decide what data they actually need before they collect it. Session-level quality data answers most operational questions without touching identity and keeping those two data sets apart makes compliance a design choice rather than a cleanup project. Context is also essential because the more granular the metrics become, the more questions there are about what sits behind them.

The practical test for transparency is whether you can show how a number was produced: the methodology. The definition, the sample and the collection method all need to be written down and pointed at. If two teams can’t agree on what a view is, no amount of additional measurement will settle it.

Josh Heller, chief revenue officer, Coactive: Most of this conversation defaults to consent and identity. That matters, but the underexamined piece is explainability: as measurement gets more granular, more of it is produced by models, and a system that cannot show its reasoning is a governance problem no matter how clean your consent posture is. Our standard is that every classification is inspectable and exportable, with timestamps and the exclusion rationale attached.

What are we not discussing that we should be?

Reinhard Grandl, chief product officer, Bitmovin: The industry is refining audience measurement for a world where a human sits in front of a screen and chooses something, and that assumption is quietly expiring. Once recommendation and discovery run through AI agents, a growing share of “audience behavior” is a machine acting on someone’s behalf, and the panels and dashboards we have today have no way to tell the difference.

The second thing we measure in only one direction is advertising. Every stack reports how many impressions were delivered and none of them report how many viewers left during the break. Ad load gets decided as an inventory question and booked as revenue, while the cost lands somewhere else entirely, in completion rates and return visits that then get blamed on the programming. Until those two numbers sit on the same page, the industry will keep optimizing for the ad that played and ignoring the viewer who didn’t come back.

Josh Heller, chief revenue officer, Coactive: That AI gives you the map, not the destination. A model can surface every family scene in a library, but it cannot decide which ones are on brief for your brand, and every brand has its own definition of premium and its own tolerance for edge cases. We prompted a model with “dirty face” once and got Halloween makeup rather than honest grime, because it was guessing without a brand to anchor it.

Advertisement