Video measurement

CTV and video campaign measurement checklist

Published July 3, 2026. Updated July 7, 2026. Status: evergreen source page.

CTV and digital video campaigns can look persuasive because the report has many strong-sounding signals: completed views, attention, household reach, brand lift, visits, attributed sales, and search movement. Those signals are useful only when each one is kept in its lane.

A common readout problem looks like this: a streaming video flight reports a high completion rate, a lift in search activity, stronger site visits after exposure, and a favorable brand survey. The meeting pressure is to call the campaign a business driver. The safer first move is to ask which signal describes delivery, which describes exposure quality, which describes measured response, and which actually has a comparison strong enough to support an incrementality claim.

This checklist helps marketers, publishers, agencies, and analysts read a video campaign report without turning exposure quality into causal proof or treating a modeled household match as a clean experiment.

CTV campaign report review desk with delivery, exposure, survey, and outcome evidence separated into distinct lanes. — Read the report as separate evidence lanes before accepting one broad campaign-impact story.

The plain-language rule is simple: a video metric says what it directly measured. It does not become a stronger kind of proof just because it appears next to a stronger-sounding metric on the same page.

Start with the campaign job

A video campaign can serve different jobs. The report should not use the metric for one job as proof that another job succeeded.

Campaign job	Useful primary signal	Important comparison	Boundary to keep
Reach a defined audience or household context.	Qualified reach, frequency, completion, device mix, and content context.	Planned delivery, excluded inventory, and frequency cap.	Reach is delivery quality, not behavior change.
Improve message exposure quality.	Viewable or audible exposure, completed views, attention signals, and creative length.	Prior video flights, placement norms, or creative variants.	Exposure quality is not the same as incremental demand.
Move brand memory or consideration.	Brand lift with field dates, sample base, exposed-control balance, question wording, and uncertainty.	Control respondents or a pre-set matched baseline.	Survey movement is not automatically sales lift.
Drive qualified site or app activity.	Incremental visits, engaged sessions, search movement, or app events tied to a stated window.	Holdout, geo baseline, prior trend, or matched market.	Post-view activity can select people already interested.
Prove business impact.	Incremental conversions, revenue, store visits, or pipeline with assignment, leakage checks, and uncertainty.	Protected holdout, geo test, switchback, or credible model calibration.	Attribution alone is not incrementality.

Report checks

Inventory and context

Name where the video ran by content group, app or site class, placement type, device class, and buying path. A report that pools CTV, desktop video, mobile video, and social clips can hide the context that created the result.

Reach and frequency definition

State whether reach is measured by device, household, account, browser, panel, or modeled person. A strong reach number is weaker when identity resolution, co-viewing, duplicate devices, or high-frequency tails are invisible.

Exposure quality

Separate served impressions, measurable impressions, viewable impressions, audible exposure, completion, and attention. Completion can reflect format constraints as much as message persuasion.

Creative and length

Break out results by creative, duration, message, offer, and landing page. A six-second reminder, a fifteen-second product message, and a thirty-second brand spot should not be judged as if they had the same job.

Outcome window

Match the window to the expected response. Immediate site visits, branded search, store visits, and long-cycle sales need different windows and different confidence levels.

Concurrent activity

List major price, promo, search, social, retail, email, PR, seasonality, and distribution changes during the flight. Video often runs as part of a broader plan, which can make simple before-and-after reads fragile.

What common video metrics can and cannot say

The mistake is not using completion, attention, lift, or attribution. The mistake is letting those signals merge into a single proof chain. Keep each lane separate until the comparison design earns a stronger claim.

Video measurement lane diagram with guardrails separating completion, exposure, attention, survey response, post-view activity, and incrementality evidence. — Completion, attention, survey movement, post-view activity, and incrementality are neighboring signals, not interchangeable proof.

Metric	What it can show	What it cannot prove alone
Completion rate	How often the video reached the end in the delivered environment.	That the viewer paid attention, understood, or changed behavior.
Viewability or audible exposure	Whether the ad had a defined opportunity to be seen or heard.	That the exposure created incremental visits, lift, or sales.
Attention score	Relative exposure quality across formats, creative, and contexts.	That attention is a universal business outcome.
Brand lift	Movement in surveyed awareness, recall, favorability, or consideration among measured respondents.	That the same movement exists outside the surveyed population or caused revenue.
Post-view attribution	Observed activity after an impression inside a reporting window.	That the impression caused the activity.
Search lift or site lift	Directional movement in demand signals during or after exposure.	That video caused the movement without a credible comparison.
Incremental conversion lift	An estimate of changed outcomes if assignment and comparison are credible.	That the result will generalize across all audiences, creative, seasons, and inventory.

Triage the report before the decision

Use the first pass to sort evidence. Use the second pass to find the missing comparison. Use the third pass to choose the next test or rewrite. That sequence prevents a strong delivery story from becoming an unsupported renewal claim.

Readout pass	Question to ask	Useful next action	Claim boundary
Delivery	Did the campaign reach the planned audience, context, device mix, and frequency range?	Repair inventory, frequency, or exclusion rules before debating impact.	Delivery quality supports reach language.
Exposure	Did the ad have a real opportunity to be seen or heard, and did creative length change the result?	Compare exposure quality by format, creative, and placement before changing budget.	Exposure quality supports attention or completion language.
Response	Did visits, search, survey movement, or conversions change in the expected window?	Check timing, source fields, match rates, and suppressed cells before crediting video.	Response supports observed-movement language.
Comparison	Was there a credible holdout, market test, switchback, baseline, or model calibration?	Upgrade to lift language only if leakage, balance, and uncertainty are documented.	Only comparison evidence supports incrementality language.

Better test designs

The strongest design depends on what can be withheld cleanly. When withholding is not possible, the report should use careful descriptive language instead of upgrade words like proved, caused, or drove.

CTV and video test design workflow showing holdout, geo, switchback, matched baseline, and model calibration paths through risk checks. — Pick the design by what can be withheld or compared cleanly, then inspect leakage, balance, timing, and uncertainty before upgrading the claim.

Design	Best video use	Main risk to inspect
Audience or household holdout	Addressable campaigns where eligible users or households can be withheld from exposure.	Leakage through other devices, household members, other campaigns, or untracked inventory.
Geo or market test	CTV or video campaigns bought heavily enough to vary spend by market.	Treatment and control markets differ in pre-period trend, distribution, promo, weather, or local competition.
Switchback	Campaigns where spend, creative, or delivery can alternate across pre-set time windows.	Carryover, seasonality, news events, or learning effects contaminate the next window.
Matched baseline	Directional readouts when randomization is unavailable and the stakes are moderate.	The match balances demographics or size while ignoring intent, prior brand demand, or media mix.
MMM calibration	Budget planning where video interacts with search, social, retail, promotions, and seasonality.	Video spend is too correlated with other drivers for the model to separate cleanly.

Readout language

Cleaner language does not weaken a strong campaign. It protects the next decision from being built on a metric that answered a different question.

Video campaign readout board showing an overclaim rejected on one side and a bounded evidence packet accepted on the other. — Rewrite the readout so delivery, response, and comparison strength each keep their own boundary.

If the evidence shows	Cleaner wording	Overclaim to avoid
Strong completion and no outcome comparison.	The flight delivered high completed exposure in the planned video context.	The campaign drove incremental demand.
Higher site visits after exposure.	Observed visits rose after exposure; causal confidence depends on the comparison described below.	Every post-view visit was caused by video.
Brand lift among surveyed respondents.	The study measured a lift in the surveyed population and window, with the stated sample and uncertainty limits.	The campaign proved profitable brand impact.
One creative led on attention.	This creative produced stronger observed attention signals and should be tested against the intended outcome.	This creative is the best sales driver.
A designed holdout showed lift.	The campaign produced measured lift for this audience, design, and window, with stated leakage and uncertainty checks.	The same lift should be expected in every future buy.

Minimum disclosure checklist

Campaign job, target audience, inventory context, buying path, device mix, and exclusion rules.
Reach definition, frequency distribution, completion, viewability, measurability, and attention methodology.
Creative rotation by length, message, offer, destination, and delivery volume.
Primary outcome, reporting window, attribution window, comparison group, and readout threshold.
Concurrent marketing, price, promo, distribution, and seasonality changes during the flight.
Uncertainty, sample sizes, suppressed cells, match rates, leakage risks, and limits on generalization.

Pair with

Use this checklist with the campaign KPI dictionary for metric definitions, the brand lift study readout checklist for survey claims, the attention measurement decision guide for exposure-quality signals, the geo lift test design checklist for market tests, and the source library for official measurement references.

Keep reading

Choose the next readout check

Move from this page into campaign readout QA, evidence triage, and renewal language before a report becomes a decision.

Readout workflowSeparate delivery from impactUse the staged workflow to keep delivery, response, quality, comparison strength, and action language apart. Readout QAAudit the evidenceCheck delivery, comparison rules, outcome quality, slices, and causal language before a budget decision. Evidence triagePick the next actionTurn weak delivery, source fields, outcomes, comparisons, and uncertainty into the next campaign step.