Method
How Vesk Lab reads and tests video ads
Vesk Lab is a marketing-science lab for US brands and agencies running video ads on Meta and TikTok. It reads each ad second by second, ranks what it finds into hypotheses, and tests the strongest one against a holdout. Every claim carries a seal that says how far it can be trusted.
Updated · Igor Lima, founder
-
Read
Each ad is marked second by second: when a face turns to camera, when the product and the brand appear, what text is on screen, where the cuts fall, what is said. Where your account has the data, the marks are lined up with the Remain curve: the share of viewers still playing the video at each second.
-
Rank
What we find becomes a short list of hypotheses, strongest first. Each carries one of three seals:
- Observation: Describes what one ad did in one campaign. It does not compare ads and does not say why.
- Association: Among comparable ads, a pattern went together with a different result. Audience, bid or context the data doesn’t show may explain it.
- Test: A holdout result. The only seal that means “because”.
-
Test
The strongest idea goes against a holdout. Before launch we write and sign the plan: the change, what stays the same, the primary KPI and guardrails, the design, the duration, the minimum budget, the smallest effect the test can detect and its power. Media spend stays yours. If the answer is “no difference”, you get that answer.
What we don’t claim
We don’t measure where people look, we don’t promise a lift, and we don’t call a pattern a cause until a holdout says so.
The terms used here are defined in the glossary, and common questions are answered in the FAQ. The sample report (PDF) shows the method applied to one ad.