What should a CTV pilot brief include?

Why write a brief

Most CTV pilots are judged on numbers that were never defined in advance: one platform reports a 7 day window, another 30 days; one counts view-through conversions from retargeted visitors, another does not. A short written brief, sent to every platform you test, fixes the rules before money is spent. It also forces each vendor to answer questions its public pages leave open.

The template

Copy the sections below into a document and fill them in.

1. Objective. One sentence. Example: "Measure whether CTV adds online orders among new customers in the US at an incremental cost per order below our paid social benchmark."

2. Primary metric. One metric, defined. Example: "Incremental orders from new customers, measured by geo holdout."

3. Measurement design. Holdout type (geo, audience or none), control share, regions, attribution window for reporting, and which tool records conversions (your analytics, an MMP or a named partner).

4. Budget and dates. Total media budget, flight dates, and the read date after the flight ends.

5. Audience. Plain-language definition and exclusions (for example, exclude current customers and site visitors from the last 30 days).

6. Creative. The video files, lengths and any end cards or QR codes. Same files for every platform.

7. Inventory constraints. Publishers required or excluded, and frequency cap.

8. Commercial terms to confirm. Ask each platform to answer in writing:

• Minimum spend for this pilot.
• Pricing model and any platform, data or management fees.
• Commitment and cancellation terms.
• Cost of the incrementality test, if any.

9. Data you will receive. Report granularity (daily, by app, by region), export format, API or log-level access.

10. Decision rule. What result leads to scaling, adjusting or stopping.

Why section 8 matters

Our research for this site shows how often commercial terms are missing or need confirming. tvScientific and StackAdapt did not publish a minimum on the pages we read; MNTN stated no minimums but no CPM or fees; Vibe's pages describe both a $50 per day figure and no minimums; confirm current terms with Vibe; Amazon stated a typical $50,000 (US) minimum for managed service; Disney Campaign Manager stated $500 per campaign and no setup fees. Asking every platform the same written questions turns these gaps into comparable answers.

Why section 3 matters most

The measurement design decides whether the pilot can prove anything. If one platform runs its own lift study and another reports only attributed ROAS, the results cannot be compared. Agree the design first, then ask each platform whether it supports it. Platforms that describe incrementality or holdout testing on their pages, such as Vibe, tvScientific and Tatari, should be able to answer quickly; others may need to check.

Gaps we see most often

Three gaps cause most unreadable pilots. The first is no written decision rule: the team sees the result and then decides what counts as success. The second is mixed audiences: one platform is allowed to retarget site visitors while another prospects, so attributed results reflect who was targeted rather than what the platform did. The third is a short flight: TV effects often arrive days later through search and direct visits, and a two-week test may end before they show. A brief that fixes the decision rule, the exclusions and a flight of at least four weeks avoids all three. If budget only allows a short test, say so in the brief and treat the result as directional.

After the pilot

Keep the brief with the results. When you review the platform six months later, the brief tells you what was tested, under which terms, and why you decided as you did.