Home / Ad testing / How to pre-test a video ad
How to pre-test a video ad: the four weeks before launch
To pre-test a video ad, start four weeks before launch. In week four, write down the one decision the test must support and choose an animatic or rough cut to test. In week three, show the animatic to a matched sample of viewers and read the second-by-second response to find where the story builds and where it flattens. In week two, test two or three finished edits, each with its own sample of 15 or more viewers per audience segment, and check how people react to the brand moment and whether what they say matches what their faces did. In the final week, use the response curve to make the 20, 10 and 6-second cut-downs and vertical versions. In launch week, lock the edit and set up tracking so you can compare in-market results with the test. The checklist for each week is below.
Updated 17 October 2026 by Jonathan Prescott, Founder, Cavefish.
The four-week plan
Each stage has a short checklist. Tick everything off before you move on.
Set the decision
4 weeks before launch
- Write the one decision the test must support, such as which of two edits to run, or whether the brand reveal lands.
- Choose what you will test first: an animatic, a rough cut or a storyboard.
- Define the audience in a sentence and estimate how easy it is to reach.
- Agree what result would change the plan before you see any data.
Test the animatic
3 weeks before launch
- Run the animatic with a matched sample of viewers.
- Read the second-by-second curve: where does response build, and where does it flatten?
- Note the scenes behind each drop and agree fixes with the creative team.
- Compare animatics only with animatics.
Test the finished cut
2 weeks before launch
- Test two or three finished edits, each with its own sample of 15 or more viewers per audience segment.
- Check the brand moment: how do viewers react in the seconds around the logo or product?
- Set what viewers said in the survey against what their faces did (stated vs felt).
- Pick the edit, and list the weak seconds to fix.
Make the cut-downs
1 week before launch
- Use the curve to choose which moments carry the 30-second ad.
- Cut the 20s, 10s and 6s versions around those moments, keeping the brand moment intact.
- Make vertical versions for TikTok, Reels and Shorts and check the first two seconds.
- Re-test any cut-down that will carry significant spend.
Lock and launch
Launch week
- Lock the final edit and archive the test results with it.
- Plan post-launch tracking so you can compare in-market results with the pre-test.
- Note which test findings you acted on, so the next test starts from evidence.
When to skip a formal pre-test
If the media budget behind the ad is small, a test can cost more than the risk it removes. For social ads you can run cheaply, test in-market instead: launch two or three variants with a modest budget, keep the winner and iterate. Pre-testing earns its cost when one edit will carry a large TV or online video spend.
Predicted vs measured
Predictive tools score an ad in minutes with a model trained on past ads, and no one watches it. Use them to screen many rough ideas early. For the edit that carries the budget, a measured test with real viewers shows what happened in each second and why. See how the approaches compare on the ad testing page, and what they cost on ad testing cost.
Pre-testing questions
How the second-by-second curve is built is set out in the methodology.
How long does it take to pre-test a video ad?
Plan for four weeks before launch if you want to test both an animatic and the finished cut, and leave time to re-edit. A single test of a finished cut can run inside a week, but it leaves little time to act on what it finds.
How many people do you need to pre-test an ad?
EchoDepth recommends 15 or more viewers per audience segment for each ad or edit. Every group you want to report on separately, such as under-35s and over-35s, is its own segment and needs its own 15. Each edit gets its own sample so the comparison is fair.
Should I test an animatic or the finished ad?
Both, if you can. An animatic test checks the idea and story before you pay for production. A finished-cut test checks the execution: pacing, casting, music and the brand moment. Compare animatics with animatics, because they draw weaker reactions than finished film.
What is a monadic ad test?
In a monadic test each viewer sees only one version of the ad, so one edit cannot colour their reaction to the next. It is the fairest way to compare edits. If viewers must see several versions, rotate the order.
Can AI predict how my ad will perform instead of testing it?
Predictive tools score an ad with a model trained on past ads, in minutes and without viewers. They are useful for screening many rough ideas. For the edit that will carry the media budget, a measured test with real viewers tells you what actually happened in each second and why.