Performance · 8 June 2026 · 8 min
Testing and scaling video ads on Meta and TikTok
A good hook is the start, not the finish. This is how to set up a testing structure that finds a winning video ad in two weeks and then scales it.
Most video ads get killed too early or kept alive too long. Both cost money. If you take video ads seriously on Meta and TikTok, you need a testing rhythm that surfaces a winner faster and shuts down losers faster.
Week 1: test wide, small budgets
Start with four to six variants of the same product message, each with a different hook. Same offer, same landing page, only the first two seconds differ.
Keep the budget deliberately low (fifteen to thirty euros per variant per day). You're not chasing revenue, you're chasing signal: hook rate, click-through and cost per click.
Week 2: scale what works
The winners from week 1 get more budget and two or three derivative variants. Different music, different pacing, different call-to-action. That keeps you from getting stuck in a single execution.
What underperforms gets switched off. No sentimental ploughing on. An ad that doesn't land in week 1 rarely starts converting in week 3.
The tempo you need
To sustain this rhythm you need to produce new variants every week. That's where most brands get stuck: coming up with the idea is one thing, actually editing it is real time.
That's what we use AI video ads for. Ten to fifteen new variants a week without a shoot, in the same brand style. For the testing frame above, that's exactly the right volume. See also which hooks consistently work based on what we see perform.
Keep reading