AI-generated storyboard illustration; a conceptual scene, not a client-production still.
A production company can deliver an excellent film that your team cannot evaluate. The files arrive, the campaign launches, and the first report raises questions nobody settled: Which version ran? What counts as a view? Did the landing page work? Was the goal attention, a qualified inquiry, or revenue?
Before commissioning the shoot, attach a measurement agreement to the production brief. It should name the creative question, the assets needed to answer it, the reporting definitions, and the people responsible for deployment and analysis. Crucially, separate accepting the producer’s deliverables from judging marketing and business outcomes. Those are different decisions, with different owners.
This is not a replacement for campaign strategy or a promise of video ROI. It is a practical handoff document for a marketing director buying commercial video production—in Dallas–Fort Worth or anywhere a production partner must coordinate with a media, web, or sales team.
Separate production acceptance from campaign performance
Production acceptance asks whether the company supplied what you commissioned: the approved cuts, correct messages, agreed versions, accurate end cards, usable files, and any specifically scoped measurement handoff. You can inspect these before launch. Define them in the scope and use clear approval gates; our guide to reviewing a video edit covers the creative and technical review itself.
Campaign performance asks what happened after distribution. It depends on the creative, but also on placement, audience, spend, frequency, offer, landing experience, tracking, and sales follow-through. A producer cannot independently guarantee those conditions unless its scope actually includes controlling them—and even then, outcomes are not assured.
Do not use “delivered two approved opening variants” and “increased qualified opportunities” as interchangeable acceptance criteria. The first is a verifiable vendor obligation. The second is a business objective requiring a defined observation method and wider accountability. If performance-based commercial terms are intended, negotiate them explicitly rather than hiding an undefined sales target inside “successful video.”
Start with one decision the evidence could change
Write the decision before choosing the metric. “We want to know whether people liked it” is too vague to commission against. A better question is: Should the next paid flight lead with the product demonstration or the customer problem?
That question creates a specific production requirement: two credible openings that lead into the same core argument. It also creates a reporting requirement: a comparison run under sufficiently comparable conditions, with a chosen success metric and a rule for an inconclusive result.
If your project is a brand film with no feasible comparison or reliable tracking, say so. You can still agree how distribution, relevant audience response, and subsequent inquiries will be observed. Do not label an observational report a test that proves the film’s effect. For broader objective and story decisions, use the brand-film planning guide; the agreement here covers the narrower bridge between the commissioned assets and their evaluation.
Copy this measurement agreement into the brief
Use the following fields as a working attachment. Replace every bracketed instruction before approving production. One named person should own each responsibility, even where a whole team contributes.
1. Audience, action, and placement
Audience: [The specific buyer or user group, plus any important exclusions.]
Intended next action: [For example, request a product demonstration—not simply “engage.”]
Placement and destination: [Exact paid/owned placements and the page or workflow receiving the viewer.]
Decision owner: [Name of the marketing lead who can change creative or distribution after readout.]
2. Creative hypothesis and commissioned assets
Hypothesis: [Changing this creative element may improve this defined response for this audience.]
Comparison: [Variant A versus B; identify what changes and what stays consistent.]
Production requirement: [Shots, performances, edit options, end cards, and exports needed for that comparison.]
Asset acceptance: [Named approver checks each delivered version against its approved script, export requirements, and version identifier.]
3. Reporting definition
Primary metric: [Exact metric name and platform/report; define numerator and denominator if it is a rate.]
Business-quality check: [How a submitted inquiry becomes a qualified inquiry, who classifies it, and where that status is recorded.]
Baseline or comparator: [Relevant prior period or controlled comparison; document differences that could invalidate it.]
Window and settings: [Flight dates, readout date, reporting timezone, conversion window, and attribution model where applicable.]
Decision rule: [What evidence would justify a change; what happens if volume or comparability is insufficient.]
4. Ownership and launch evidence
Producer: [Supplies the agreed assets and a version manifest; fixes delivery deviations within the agreed scope.]
Media owner: [Maps asset IDs to placements, controls the comparison setup, and documents spend/audience changes.]
Web/analytics owner: [Implements and tests destination links, campaign tagging, player events, consent behavior, and form tracking as applicable.]
Sales/CRM owner: [Validates inquiry quality and supplies a defined outcome report without unnecessary personal data.]
Launch approver: [Reviews the actual deployed experience and records pass/fail evidence before traffic starts.]
5. Limits and review
Known blind spots: [Untracked placements, small samples, delayed sales, missing identity, or other relevant limitations.]
Readout owner and date: [Who reconciles the reports and presents the decision.]
Change boundary: [Which additional cuts, tracking work, or reshoots require separate authorization.]
The attachment is operational, not a substitute for the production contract. Resolve conflicting responsibilities before the estimate is signed, then incorporate the agreed production obligations into the scope.
Ask for variants the shoot can actually support
“Give us something to A/B test” is not a production instruction. An alternate opening may require another setup, a different performance, an additional interview answer, or dedicated product coverage. An editor cannot always manufacture an honest comparison from footage designed for a single story.
For a product demonstration, commission two openings that both transition cleanly into the same demonstration and call to action. Capture each as a complete usable beat, including the action’s beginning and end. If only the opening is under evaluation, avoid quietly changing the offer, runtime, music, and destination as well; you would no longer know which change mattered.
For an interview-led film, ask for two self-contained opening statements supported by appropriate coverage. Do not chop a person’s words into a stronger claim they did not make. For an end-card comparison, capture a clean ending with room for the agreed treatments and verify that each call to action matches its actual destination.
These are illustrative production choices, not a universal test recipe. Ask the media owner which comparison the intended platform can run before buying extra variants. Google’s video experiment guidance describes a basic A/B approach with creative as the single variable and a selected success metric; availability and setup depend on the campaign. Use the separate cutdowns and vertical-versions guide for the full deliverables and framing plan. Here, every extra version needs a measurement reason.
Define “views,” clicks, and inquiries precisely
Never accept “video views increased” without a reporting source and definition. YouTube explains that paid video metrics vary by ad format. Counts from different placements are not automatically comparable measures of attention.
For a rate, record both parts. “Destination click-through rate” still needs the platform’s precise click and impression definitions. YouTube’s impression CTR guidance, for example, concerns views after registered YouTube impressions—not every possible exposure or channel view. Copying a percentage into a slide without its denominator creates false precision.
On your website, distinguish a player event from a business action. GA4’s built-in enhanced video measurement applies to embedded YouTube videos with JavaScript API support enabled; do not assume every player is automatically tracked. Likewise, a recorded form submission is not proof of a qualified lead. Define the qualification step and its owner separately.
Run a deployment rehearsal before launch
Have the web and media owners test the actual destination, not just the delivered movie. Follow each intended path, check the correct creative and end card, submit a permitted test inquiry, and confirm the expected records appear in the designated reports and workflow. Document what was tested and any limitations. Keep this separate from approving the master file.
Use a version manifest with a simple creative ID, filename, hypothesis, placement, and destination. The media owner should preserve that mapping when uploading; otherwise the report may identify an ad nobody can match to the approved cut.
Agree campaign naming with the analytics owner. Google’s campaign-tagging guidance supports consistent source, medium, and campaign parameters, with utm_content available to distinguish creatives. UTMs label referral traffic; they do not prove that a video caused a sale. Do not assume a landing-page report will display every UTM value, either—select the appropriate reporting dimensions with the analyst.
Put attribution limits in the agreement, not the footnote
An attribution report assigns credit under its configured rules. Google’s attribution overview explains that credit depends on the model and user-path information. An attributed conversion is not, by itself, proof that the conversion would not have happened without the video.
If the business needs incremental impact, discuss an appropriate controlled study with the media and measurement specialists before production. Google’s Conversion Lift explanation describes exposed and control groups and notes that access is not available to every account. Do not promise a lift study your campaign cannot support.
At readout, separate delivery findings, deployment findings, observed performance, and causal conclusions. Where volume is too low or conditions changed, record “inconclusive” and the next decision—not a winner manufactured from a small difference.
The decision before the camera rolls
Ask the producer, media lead, and analytics owner to walk through the same attachment. Can the crew capture the comparison? Can the editor deliver it without inventing missing coverage? Can the deployment team identify each version and observe the intended action? Does everyone know what the evidence cannot establish?
When those answers are clear, measurement becomes a production input rather than a disappointed conversation after launch. Accept the work against the agreed deliverables. Evaluate the campaign against the agreed evidence. Keep those judgments connected, but never confuse them.
