CTV measurement & attribution: How to measure what actually matters

Why CTV measurement is genuinely hard

CTV has solidified itself as a core advertising channel, but throughout its consistent growth one issue has remained: measurement. There are many aspects that make connected TV ad measurement more difficult than other digital channels.

For one, it is a fragmented ecosystem. Today’s audiences don’t just watch content in one place; they move across multiple streaming services, frequently subscribe and unsubscribe, and watch on different screens throughout their homes. This makes it difficult to get a single, consistent view of who’s watching what, where or when.

Also, many of the leading streaming platforms such as Netflix, Disney+ and Amazon Prime Video operate as walled gardens, which means that reporting is limited

There is also technically no “click” on a TV screen like there would be on other channels. While new solutions like QR codes and pause ads are being tested, it is still more of a challenge to trace actions back to specific moments with connected TV. Measuring delivery is easy, but measuring real impact requires a more thoughtful approach.

The good news is that while these challenges are real, they are not unsolvable. But building one streamlined CTV measurement strategy can be quite complex. Continue reading to learn more about the CTV attribution framework that is helping marketers better understand their CTV performance to make the most out of their ad spend.

The metrics that actually matter on CTV

CTV ad measurement starts with understanding the difference between what was delivered and what actually drove results for your brand.

• Delivery metrics show whether an ad ran as expected. These include impressions, completion rate, viewability, in-target reach and frequency. They help confirm that a campaign was executed properly.
• Outcome metrics go a step further. They focus on what changed as a result of exposure, such as site visits, app installs, conversions, in-store visits and brand lift.

The right CTV metrics depend on the campaign objective:

• For awareness, reach and frequency paired with brand lift are more relevant.
• For consideration, engagement signals like site visits or search activity are more relevant.
• For conversion, the primary metrics would have to do more with actions taken after exposure.

When marketers use both delivery and outcome-based metrics, they provide a more complete view of performance. Delivery metrics validate that your ads were seen, while outcome metrics help determine their effectiveness.

The 4 CTV attribution models—and when to use each

There is no single CTV measurement model that fully captures performance. Most marketers rely on a combination of approaches:

These models are rarely used in isolation. Oftentimes, brands combine methods to balance speed, accuracy and strategic insight. For example, one model may help with in-flight optimization, while another supports longer-term planning.

Cross-screen attribution: The hard problem no one solves alone

Another reality to face with CTV is that audiences do not experience media in silos. A single household may move between linear TV, streaming platforms and digital channels throughout the day. From a measurement perspective, this means that the same audience may be counted multiple times or not connected at all.

Without a way to unify these touchpoints, it becomes difficult to answer some fundamental measurement questions:

• Is your brand reaching new audiences, or the same ones repeatedly?
• Which channels are actually driving results?
• How much incremental value is CTV delivering within the broader mix?

The answer to this unification doesn’t lie in one solution, but in many. Effective cross-screen measurement depends on:

• A consistent way to recognize audiences across environments
• Deduplicated exposure data to avoid overcounting
• Privacy-safe ways to collaborate with platforms that restrict data access

When these elements come together, marketers can get closer to a complete picture of performance.

How identity powers every measurement method

No matter which attribution model your brand decides to use, one thing remains the same: a strong identity solution is key.

• VTA needs to know the same household was exposed and later converted
• MTA needs persistent IDs across every touchpoint
• MMM benefits from cleaner channel-level exposure data
• Brand lift requires identity-resolved survey panels for control vs. exposed

So, while it is important to choose a strong measurement partner, it is arguably even more important to find a solid identity partner.

Building a CTV ad measurement stack

Given the complexity of CTV, building a measurement strategy is less about choosing a single solution and more about aligning multiple components.

Here is a five-step framework for assembling a CTV ad measurement program:

1. Define the objective and the corresponding success metric. Not every campaign should be measured the same way. Awareness campaigns should prioritize reach and brand outcomes, while performance campaigns should focus more on actions.
2. Anchor on identity. Measurement depends on the ability to recognize audiences across touchpoints. Establishing an early on helps connect exposure to outcomes and supports more consistent attribution.
3. Layer methods. No single attribution model tells the full story. Using a combination of approaches allows marketers to be more confident in their measurements.
4. Negotiate walled-garden access. Many CTV environments limit data access, making it necessary to incorporate other technologies. Clean room environments and privacy-safe data sharing approaches help marketers work within these constraints.
5. Set the cadence. Delivery metrics are often reviewed more frequently, while outcome-based insights can take longer. Establishing a consistent reporting cadence helps teams balance tracking their short-term performance with analyzing the longer-term impact.

Bringing these elements together requires the right combination of technology, data and expertise. Epsilon's CTV advertising platform does this by taking a more unified and effective approach to measurement. Recognized in the IDC MarketScape, Epsilon combines in-flight visibility through our partnership with iSpot.tv with our industry-leading identity solution, COREid. This enables more accurate cross-screen deduplication and audience understanding. Clean-room integrations help advertisers extend measurement into difficult environments such as walled gardens.

Together, these capabilities make it easier to move beyond fragmented reporting and toward a more complete view of CTV performance.

FAQ

What's the most accurate way to measure CTV advertising?

Incrementality testing is the gold standard for causal measurement, but it's expensive and slower than other methods. Most advertisers combine identity-resolved MTA for in-flight optimization with periodic incrementality tests for strategic validation.

What's the difference between MTA and MMM for CTV?

MTA assigns credit at the touchpoint level using log-level data (needs identity resolution); MMM is a top-down statistical model that estimates channel contribution from aggregate data (privacy-safe but less granular). They're complementary, not substitutes.

How do you measure CTV inside walled gardens like Netflix or Amazon?

Through the streamer's clean room (Amazon Marketing Cloud, Disney Real-Time Ad Exchange, NBCU One Platform Total Audience) — where you can match your first-party data against their viewership data in a privacy-safe environment to attribute outcomes.

Can you measure ROI on CTV the same way as digital?

Not exactly. There's no click. But identity-resolved view-through attribution, incrementality lift, and sales-lift studies all produce ROI-comparable measures when set up correctly.

What is incrementality testing in CTV?

A measurement method that compares exposed and unexposed (control) groups to isolate the causal effect of the CTV campaign on outcomes. Considered the most rigorous form of CTV measurement.

What's the typical completion rate on CTV ads?

Typically 95%+ — CTV ads are generally non-skippable and run in lean-back environments, so completion rates are much higher than display or social video.