How to Build a CTV Measurement Framework

Why CTV needs its own measurement framework

You can't measure CTV with the same framework you use for Meta or Google. The mechanics are different. The attribution chain is different. The timeline from exposure to conversion is different.

Agencies that bolt CTV onto their existing digital measurement stack end up in one of two traps: either CTV shows zero attributed conversions (because last-touch ignores it) or the platform claims unrealistic credit (because view-through windows are too generous).

A dedicated CTV measurement framework prevents both traps. It defines the right KPIs, establishes baselines, layers attribution methods, and validates with experiments. Here's how to build one.

Step 1: Define the measurement objectives

Before selecting metrics, clarify what the CTV campaign is meant to achieve. CTV serves different roles depending on the advertiser's situation:

Awareness driver. For brands entering a new market or launching a new product. Primary metrics: reach, frequency, brand lift (awareness, consideration, intent).

Demand generator. For brands with existing awareness that want to drive action. Primary metrics: website visit lift, search lift, attributed conversions, incremental CPA.

Reach extender. For brands already running digital video but hitting frequency caps or audience saturation. Primary metrics: incremental reach (households not reached by other channels), unduplicated total reach.

Performance complement. For direct-response advertisers who want to amplify their paid search and social campaigns. Primary metrics: cross-channel conversion lift, assisted conversions, halo effect on other channel performance.

Each objective requires different KPIs and different attribution approaches. Define the objective first, then build the measurement stack around it.

Step 2: Establish pre-campaign baselines

You can't measure lift without knowing what "normal" looks like. Before launching CTV, measure and record:

• Weekly website traffic by source (direct, organic search, branded search, referral). Use 8 weeks of pre-campaign data to establish a stable baseline.
• Branded search volume. Track weekly branded search impressions and clicks in Google Ads and Search Console. CTV consistently lifts branded search, so this is one of the most reliable CTV performance indicators.
• Conversion rate by source. Baseline conversion rates for each traffic source. CTV can lift conversion rates across channels by building familiarity and trust.
• Total weekly revenue or conversions. The aggregate number you're trying to move. This is the denominator for ROI calculation.

Store these baselines in a central document. You'll reference them throughout the campaign to calculate lift.

Step 3: Set up the attribution stack

Layer three attribution methods from most granular to most comprehensive:

Layer 1: Platform-native attribution

Use each CTV platform's built-in measurement tools. This typically includes:

• Impression and completion reporting (100% coverage).
• Audience reach and frequency (deduplicated within platform).
• Conversion attribution via the platform's identity graph (coverage varies by platform, typically 30-60% of conversions).

Platform-native attribution is your baseline. It's available immediately, requires minimal setup, and provides campaign management signals (which placements, dayparts, and audiences are performing).

Important limitation: Each platform measures independently. Deduplicated reach across platforms requires a third-party measurement partner.

Layer 2: Cross-device attribution

Implement one or more cross-device attribution methods:

• Server-side pixel with IP matching. Place a server-side conversion event on your website. Match the converter's IP address to CTV impression logs. This captures conversions from any device on the same household network as the CTV device.
• Identity graph integration. Partner with a device graph provider (LiveRamp, The Trade Desk, TransUnion) to match CTV device IDs to web and mobile identities. Higher accuracy than IP matching, but requires data onboarding and has associated costs ($2-$5 CPM on matched impressions).
• CRM matching via clean rooms. If you collect customer email addresses at conversion, match them against the streaming platform's user database via a data clean room. Highest accuracy for identified customers, but only captures customers who provide contact information.

Layer 3: Aggregate attribution

These methods capture the total business impact, including effects that user-level attribution misses:

• Search lift analysis. Compare branded search volume during the campaign to the pre-campaign baseline, controlling for seasonality. Attribute the lift to CTV.
• Website visit lift. Compare direct and organic website visits during and after the campaign to baseline. Control for other marketing changes.
• Media mix modeling. If you have 12+ months of data, MMM provides the most comprehensive view of CTV's total revenue contribution.

Step 4: Build the KPI dashboard

Organize your CTV KPIs into four tiers:

Delivery metrics (reported daily):

• Impressions delivered
• Completion rate (target: 95%+)
• Unique households reached
• Average frequency per household

Engagement metrics (reported weekly):

• Website visit lift (% increase vs. baseline)
• Branded search lift (% increase vs. baseline)
• QR code scans or vanity URL visits (if applicable)

Attribution metrics (reported weekly):

• Platform-attributed conversions
• Cross-device attributed conversions
• Cost per attributed conversion

Incrementality metrics (reported monthly):

• Incremental conversions (from lift study or holdout test)
• Incremental CPA
• Incremental ROAS
• Brand lift survey results (if running)

The daily and weekly metrics guide campaign optimization. The monthly incrementality metrics validate the campaign's true business impact.

Step 5: Run validation experiments

No CTV measurement framework is complete without experimental validation. Plan for two types of experiments:

Conversion lift study (run in first quarter). Work with your primary CTV platform to set up a test/control experiment. The test group sees your ads; the control group sees PSAs or no ads. Minimum requirements: 500K+ impressions in the test group, 4+ week flight, clear conversion event (purchase, sign-up, lead).

The lift study serves two purposes: it provides a true incrementality baseline, and it calibrates your ongoing attribution. If the lift study shows 500 incremental conversions but your cross-device attribution only counted 300, you know your attribution captures about 60% of true impact. Apply that calibration factor going forward.

Geographic holdout test (run annually). Select 2-3 markets with no CTV advertising and compare total revenue performance to similar markets with CTV. This captures all CTV effects -- including offline, halo, and long-tail effects that platform attribution misses.

Step 6: Iterate and calibrate

After each reporting period, review and refine:

• Update calibration factors. If new experiments show your attribution is capturing more or fewer conversions than expected, adjust the calibration.
• Refine the baseline. As CTV becomes an ongoing channel, the pre-campaign baseline becomes less relevant. Transition to using geographic holdout markets or always-on incrementality testing as your baseline reference.
• Add new data sources. As new measurement capabilities emerge (ACR data, improved clean rooms, better device graphs), integrate them into the framework.
• Document and share learnings. Each CTV campaign teaches you something about measurement accuracy, channel dynamics, and optimization levers. Build an institutional knowledge base.

FAQ

How long does it take to build a CTV measurement framework from scratch?

Allow 4-6 weeks for the initial setup: 1-2 weeks for baseline measurement, 1 week for attribution stack configuration, 1 week for dashboard building, and 1-2 weeks for testing and validation. The framework then evolves continuously based on campaign data and experimental results.

What's the minimum CTV budget to make this framework worthwhile?

The framework scales down, but the experiments require minimum budget levels. For platform lift studies, plan for $50K+ per platform over 4-6 weeks. For geographic holdout tests, $100K+ across test markets. If your total CTV budget is under $50K per quarter, skip the formal experiments and focus on search lift analysis and aggregate attribution.

Should every client get the same CTV measurement framework?

The structure should be consistent (same four KPI tiers, same validation approach), but the emphasis differs by client. Brand-awareness clients need more investment in brand lift measurement. Performance clients need stronger cross-device attribution. New-to-CTV clients need more emphasis on establishing baselines and running initial experiments.

---

Go Funnel uses server-side tracking and multi-touch attribution to show you which ads actually drive revenue. Book a call to see your real numbers.