Table of Contents
Search has click data. Social has conversion pixels. For years, television had neither. Sure, there was brand awareness, but not accountability. Connected TV (CTV) has changed the calculus. Measuring CTV campaigns today means tracking return on ad spend (ROAS), cost per acquisition (CPA), and verified conversions (CVR) with the same rigor applied to paid search.
The gap between what’s possible and what most brands currently do remains pretty wide. Many direct-to-consumer (D2C) advertisers still measure CTV the way they measured linear in 2012: through reach and frequency, and that’s about it.
This guide covers the metrics that matter today, including those that don’t always appear in your dashboard by default.
CTV Core Performance Metrics: Tracking ROAS and CPA on the Big Screen
CTV performance metrics are the same ones you track everywhere else: connected TV ROAS, CPA, and conversion rate (CVR). What changes is how you collect them.
You can’t click a TV screen, so without a clickstream to follow, CTV relies on a pixel-based tracking setup paired with a defined view-through attribution (VTA) window, typically 24 to 72 hours after ad exposure. (You use your purchase cycle to set that number).
When a user sees your CTV ad and converts on any device within that window, the conversion is attributed to the CTV impression. Get the window wrong, and all the conclusions you draw about CTV ROAS will be wrong, too.
Cost per completed view (CPCV) is also worth adding to your reporting. Cost per mille/one thousand impressions (CPM) counts an impression as soon as an ad loads; CPCV only counts users who watch to the end. Because you can’t skip most pre-roll and mid-roll placements on streaming programs, completion rates on CTV run higher than other video formats, so CPCV is a more meaningful efficiency metric than CPM for this channel.
Video completion rate (VCR) and reach metrics complete the baseline layer. Although they don’t directly measure bottom-funnel outcomes, they’re useful diagnostics. A high CPCV plus a low VCR usually points to a creative or placement problem worth investigating before drawing bigger conclusions about your D2C TV advertising measurement.
How Cross-Device Attribution Works
CTV is an impression-based medium; tracking CPA, CVR, and ROAS depends on VTA and cross-device matching. The connecting layer is an identity graph, or database, that maps different devices to the same household. It looks like this:
- You serve a streaming ad on a smart TV or a streaming stick (Roku, Apple TV, Fire TV Stick). The CTV ad network logs the device identifier for advertising (IFA), IP address, and timestamp.
- A viewer sees the ad. Later, they visit your site on their smartphone or laptop and hit “buy.”
- Your tracking pixel or conversion API (CAPI) fires, capturing the purchase value, timestamp, shopper’s IP address, and hashed email if they log in.
- The attribution platform matches the pixel data to the CTV impression log. A matching IP address, or a hashed email mapped to the same household profile, gives the CTV credit for the conversion.
Identity graph resolution is the variable that determines the accuracy of this data. Check carefully when evaluating CTV measurement partners, as a weak graph means missed connections and understated results.
It’s also worth noting that some performance platforms extend this logic beyond simple attribution. Realize, for example, enables advertisers to retarget and build lookalike audiences based on CTV exposure, extending campaigns from the television to the open web. This means a CTV impression doesn’t just get credit for a later conversion, it can also trigger a follow-up native or display campaign across publisher sites, creating a more direct path from TV awareness to web action.
Measuring the Halo Effect: CTV’s Impact on Search and Social
CTV’s influence doesn’t stay on television. According to MNTN Research, 65% of CTV viewers use a second device to look up information and navigate to an advertiser’s site while they’re streaming. That second-screen behavior is a direct pipeline from TV impression to branded search query, and it’s the mechanism behind what the industry calls the halo effect.
This cross-channel halo effect measures the boost CTV exposure provides to your search, social, and affiliate channels. The data specifically shows how much more likely someone is to convert when they see a CTV ad alongside your other ads, compared to seeing only those other ads. MNTN also showed that CTV boosts the performance of other marketing channels by up to 54%, and that brand recall more than doubles when a CTV ad accompanies a digital ad for the same brand.
Platforms designed for cross-channel performance can help capture this halo effect more deliberately. Realize, e.g., allows advertisers to extend CTV campaigns to the open web through retargeting and lookalike modeling, effectively turning a TV impression into the first touchpoint of a multi-channel journey, rather than a standalone awareness play.
Measuring the halo effect requires comparing exposed users against a control group. Watch these three signals:
- Search lift: An increase in branded search queries (your brand, product name, or branded keywords) just after CTV exposure. This increase signals that the CTV is performing well. Ara Ohanian, CEO of Aragil, describes the mechanism well: CTV campaigns actively generate new branded search queries that are less competitive, convert at higher rates, and signal a warmer audience than generic category searches.
- Social CPA/CPI reductions: Users who’ve seen your CTV ads tend to convert at higher rates when they see your social ads later. This metric appears as a lower CPA on Meta or TikTok for the exposed cohort, compared to the control group.
- Direct traffic baseline increases: Direct traffic to your site that increases during and after a CTV campaign also often reflects the halo effect. Users remembered your brand and searched your URL directly. Last-click attribution misses this metric entirely, hence why halo analysis exists.
Lennea Zielinski, head of content at Prescient AI, has stated that research underscores the measurement challenge, pointing out that, without measuring halo effects, you might see an increase in branded search or direct traffic, but traditional attribution would attribute those conversions to search or direct channels, completely missing the CTV campaign that initiated the customer journey.
The moral of this story? Don’t ignore the halo effect. Companies that cut budgets for channels appearing to perform poorly in direct attribution could see overall performance decline across all channels, because they were cutting campaigns that created valuable halo effects.
Attribution Models for Non-Clickable Media
CTV attribution models use different infrastructure than click-based channels. Since television ad views don’t have a UTM parameter attached, connecting TV impressions to conversions relies on three technologies working together:
- View-through attribution (VTA).
- Cross-device identity graphs.
- Multi-touch attribution models.
View-Through Attribution
When your CTV ad serves to a household, and conversion happens on any device within your defined window, VTA assigns credit to that impression. How well the ad performs depends on how accurately the identity data connects the TV to the converting device.
Identity Graph Resolution
The graph links device IDs, emails, IP addresses, and physical addresses to a household or individual profile. If someone sees your ad on a Roku and an hour later makes a purchase on their iPhone, this graph makes all the connections.
Multi-touch Attribution (MTA) Models
Instead of assigning conversion credit to one channel, MTA models distribute conversion credit across multiple touchpoints in the customer journey. For D2C brands running CTV, search, and social simultaneously, MTA prevents double-counting conversions that involve multiple channel exposures. Without it, your CTV platform, Meta dashboard, and Google Ads account may each claim full credit for the same purchase and hide what really happened.
Advanced Measurement: Incrementality and Lift Testing
Although view-through attribution will tell you a conversion happened after CTV exposure, it doesn’t tell you which ad led to the conversion — an important distinction when it’s time to talk about budget.
Incremental lift testing CTV campaigns (often measured as incremental ROAS or iROAS) is the methodology that answers the causality question. You compare a group exposed to your campaign against a matched control group that wasn’t. The difference in conversion rate between the two groups is the lift your campaign drove.
Simulmedia chief scientist Alex Papiu has previously noted that, “In CTV advertising, A/B testing involves randomly assigning viewers to a test group (exposed to your ad campaign) or a control group (either exposed to a placebo creative or generated synthetically). By comparing their behaviors and outcomes, you can establish a direct causal link between your campaign and its impact on viewer behavior.”
There are four main approaches to testing CTV incrementality.
PSA (Public Service Announcement) Testing
This public service announcement (PSA) runs as a control creative, giving you a clean comparison against your campaign. It’s easy to set up, and a good starting point for anyone running incrementality tests for the first time. The trade-off is that some of your media budget will go to non-promotional placements.
Geo Testing
Geographic testing divides your audience by geographic region into test and control groups. You can use it to see whether customers would have converted without the ad, and it avoids inflated return on investment (ROI) metrics that platform dashboards sometimes generate. Since regional differences can skew results, factor those differences in.
Ghost Bidding
Ghost bidding places non-winning bids on your active campaign to build a control sample without adding more cost. It’s a bit more complex than PSA testing, but it avoids budget trade-offs. The output answers the same question — did this user convert because of the ad, or were they going to convert anyway? — without requiring a portion of your spend to go to PSAs.
Synthetic Control Groups
These control groups build a statistical benchmark by combining data from multiple unexposed groups and applying weights to model what might have happened without advertising. Marketers who need to measure broad, hard-to-test formats (e.g., national TV, podcast placements, and out-of-home) use these groups when traditional holdout testing doesn’t make sense.
A caveat: if you don’t separate organic conversions from ad-driven ones, you could potentially overinvest in audiences with high baseline conversion rates and miss the incremental growth the channel can generate.
Incrementality data is the strongest case you can make when arguing for a larger CTV budget. Platform-reported ROAS provides the number; iROAS from a holdout test is evidence.
Key Takeaways
The infrastructure to measure CTV as a performance channel exists. You can track ROAS, CPA, and incrementality data with the right setup. What requires a calibration in rethinking is that CTV conversions rarely happen in the same session as ad exposure, and even more rarely on the same device. Platform-native attribution chronically undercounts CTV’s contribution for this exact reason.
What gives a more accurate picture of a channel’s production, then? Building your measurements around CTV view-through attribution, identity graph resolution, and cross-channel halo analysis. If you already run search and social at scale, the halo effect might also justify a CTV test, as the channel tends to reduce CPAs elsewhere even when direct attribution makes it appear like it’s doing nothing. Incrementality testing closes the loop, turning a correlation into a defensible claim when the CFO asks why you’re scaling TV spend.
Frequently Asked Questions (FAQs)
How do I set up conversion tracking for connected TV ads?
The typical approach combines a pixel on your confirmation or thank you page with a view-through attribution window defined through your CTV demand-side platform (DSP) or measurement partner. When a user sees a CTV impression and a conversion fires within that window, the platform uses its identity graph to connect them.
For more precise tracking, choose a measurement partner with strong household-level resolution. This layer connects the smart TV impression to the mobile or desktop conversion.
What is a good ROAS for CTV compared to Facebook?
Since CTV reaches users earlier on the path to purchase — and they routinely complete purchases on a different device or through a different channel — direct ROAS on CTV typically runs lower than on bottom-funnel channels like search or retargeting-heavy social. CTV’s halo effect often means that the blended return is comparable or superior to social alone.
How do I know if a conversion came from CTV or a Facebook ad?
You need a multi-touch attribution (MTA) model or measurement partner with identity graph resolution. The identity graph confirms that the smart TV serving the CTV ad and the device belong to the same household, so the credit isn’t duplicated across channels. Without either, your Meta dashboard and your CTV platform can each claim the full conversion, so your total reported run-of-site (ROS) is inflated, and you may not be able to tell which channel got the sale.
