15 questions to ask a CTV ad platform before a pilot

Public pages are written for many readers and change often. Your pilot depends on terms agreed for you: the minimum, the fees, the services your ads will run on, and how results will be measured. Asking in writing gives every shortlisted platform the same questions, makes the answers comparable, and leaves a record if something changes after launch. It also shows quickly which platforms answer clearly and which do not.

• 1. What is the minimum spend or commitment for a pilot, and for how long?
• 2. Which fees apply on top of media, and how are they shown on invoices?
• 3. Is there a contract, and can we stop at the end of the pilot without penalty?
• 4. Which payment methods and invoicing terms are available?

• 5. Which named streaming services and networks will our ads run on?
• 6. Can we include or exclude specific apps or content categories?
• 7. How much of a similar campaign's delivery ran on the services we care about?

• 8. How do you measure incrementality: household holdout, geo test or another method?
• 9. Can we set the holdout size and lock it before launch?
• 10. Which attribution window do your reports use, and can we change it?
• 11. Which third-party measurement partners can read our results, and is there a cost?

• 12. Which brand-safety or verification partners do you support?
• 13. Which of our first-party data can we use for targeting, and how is it protected?
• 14. Which security attestations do you hold?
• 15. Who supports the campaign day to day, and how fast do they respond?

Put the answers side by side in one table, with one row per question and one column per platform. Mark any answer that is vague or missing. Then weigh the answers against the goal of your pilot: if the pilot must prove incremental results, questions 8 to 11 decide; if one service is a must, question 5 decides; if procurement is strict, questions 3, 12 and 14 decide. A platform that answers every question clearly but fits your goal less well is still a weaker choice than one that fits and answers well.

Some answers are on public pages. For example, MNTN's homepage states no upfront commitments and no minimums; Vibe's MCP page describes incrementality testing built in, with geo holdout examples, and its FAQ states SOC 2 Type 2; Amazon's page describes self-service and managed-service options; StackAdapt's plans page lists net terms from its Grow plan. Each profile on this site shows what we found, with sources, so you can skip what is already answered and focus on what is not.

Source: MNTN homepage

Source: Vibe MCP page

Source: Amazon Ads: Amazon DSP

Source: StackAdapt plans and packages

Whatever the answers, keep the pilot simple: one goal, one metric, one comparison group, and terms you can walk away from.

Send the questions at the same time to every platform on your shortlist, and give each the same deadline. Platforms that need longer for some answers will say so; that is useful information too. If an answer points to a document, such as terms of service or an inventory list, ask for the document itself rather than a summary, and file it with the answers.

Keep the list short enough to answer. Fifteen questions is already a lot; add your own only where your situation demands it, such as category restrictions for regulated products, data-protection terms for your region, or creative approval rules. A long questionnaire sent too early can slow a pilot that was meant to be quick and low-risk.

When the answers come back, share them with the people who will approve the pilot: finance for terms and fees, legal or procurement for data and safety, and the measurement owner for the test design. Each group spots different gaps, and their sign-off before launch is worth more than their questions after it.