Proof School · 10 lessons · 3 tracks
Proof School: lessons on buying and proving CTV advertising
Short answer
Ten lessons in three tracks: the basics of streaming TV buying, how to prove what a campaign added, and how to buy and run a pilot. Each takes under five minutes and links to the vendor pages it relies on.
By The Playback Desk · Best CTV Ad Platforms · Published 2026-10-15
Desk research from public vendor pages. Editorial assessment; we did not run campaigns on these platforms.
Basics
Lesson 1 of 10
What CTV advertising is and how it is bought
CTV advertising places video ads in streaming apps watched on a television. Buyers reach that inventory through self-serve platforms, programmatic DSPs or a publisher's own sales team, and each route changes what you can control and measure.
3 min read
Lesson 2 of 10
Self-serve, managed and demo-led CTV platforms
Self-serve platforms let you launch yourself, managed service puts an account team in charge, and demo-led platforms start with a sales call. Many platforms now offer more than one model, so check what each publishes before you plan a pilot.
3 min read
Lesson 3 of 10
How to read CTV pricing: CPM, minimums and fees
Most CTV is bought on a CPM, the price per thousand impressions, but platforms publish very different amounts of detail on minimums, fees and commitments. Read each term separately and confirm anything a page does not state.
3 min read
Proving results
Lesson 4 of 10
How CTV attribution ties TV ads to outcomes
CTV attribution links a household that saw an ad to later activity such as a site visit, purchase or app install, using a pixel, an identity graph or a measurement partner. It shows what happened after exposure, not what the ad caused.
3 min read
Lesson 5 of 10
Incrementality versus attribution
Attribution counts what exposed households did after seeing an ad. Incrementality compares them with a similar group that did not see it, to estimate what the ads actually added. Proof of CTV value needs the second.
2 min read
Lesson 6 of 10
How to run a geo holdout test
A geo holdout test runs a CTV campaign in some regions and withholds it in matched regions, then compares outcomes. With a clear hypothesis, matched markets and a fixed window, it gives a defensible estimate of what the campaign added.
3 min read
Lesson 7 of 10
Working with third-party measurement partners
Third-party measurement partners, from attribution tools to incrementality and mix-modelling firms, give a second view of CTV results outside the platform that sold the media. Check which partners a platform names and what data it can share with them.
2 min read
Buying and running
Lesson 8 of 10
How to pilot a new CTV platform against your current provider
Run the new platform and the current one side by side for a fixed period, with the same creative, audience definition and measurement method, and judge them on incremental results rather than reported ROAS. Choose a platform with terms that let you stop.
2 min read
Lesson 9 of 10
AI assistants, MCP and APIs in CTV buying
Some CTV platforms now let buyers plan, launch and adjust campaigns through AI assistants or APIs instead of only a web interface. Check what the assistant can read and change, how access is authorised and whether edits are logged.
2 min read
Lesson 10 of 10
Inventory: publishers, walled gardens and reach claims
CTV inventory differs by which apps and networks a platform can buy and how much of it is exclusive. Compare named publishers, check for the ones you need, and read household reach figures as the vendor's own statement.
3 min read
Frequently asked questions
Do I need to read the lessons in order?
No, but the proof track builds on itself: attribution, then incrementality, then the geo holdout test.
Are the lessons about specific vendors?
Mostly about methods. Where a lesson names a platform, the fact comes from that platform's own page, linked at the end.