Changing the colours in your ads isn't enough. Two ads that differ only in the shade of the button rarely tell you anything you will use again. Real testing is about the decisions that determine whether someone stops scrolling to look at the ad and whether they understand what you are offering them.
What is worth testing
- Hook: the first seconds of a video or the first line of a static ad. A problem, the product in use, a question.
- Narrative format: a demonstration, a before-and-after comparison, a person talking to camera, a carousel that explains step by step.
- Product angle: which feature or benefit comes first. Comfort, material, price, use on a particular occasion.
- Call to action: what you ask the viewer to do, and in what words.
Each of these changes answers a different question about your audience. That's why the findings can also be used outside your ads, on your website, in your newsletter or in your store signage.
Hypothesis, variable, criterion
Before you create the variants, write down the hypothesis. For example: “If the video opens with the product in use instead of the logo, more viewers will watch it to the end.” The hypothesis tells you what is changing and what you expect to see.
Change one thing at a time. If you change the hook, the music and the copy together and one variant does better, you won't know why.
Start with a few variants. Two or three versions of a hook give a clearer picture than ten different ads sharing a small budget.
Decide which event you will judge the test by. It might be the video being viewed up to a certain point, the click, the add to basket or the form submission. The criterion should match the campaign goal and be reliably measurable with the data you have.
Check the campaign structure as well. When variants run together in the same ad set, the platform may quickly give most of the delivery to one of them. For a clean comparison, use the testing tools that Meta Ads and Google Ads provide.
Small samples, hasty conclusions
Avoid drawing hasty conclusions from small samples. With only a few clicks or two or three conversions, the picture can reverse from one day to the next. Give the test enough time and budget to cover both weekdays and a weekend, and check whether the difference holds steady.
Also bear in mind that a result applies to that particular audience, period and offer. What worked during a sale won't necessarily work at regular prices.
A test doesn't produce a winner for ever. It gives you a better hypothesis for the next round.
From findings to the next cycle
The findings feed into the next creative cycle, not into a rule that only looks certain. Keep a simple record of each test: date, hypothesis, variants, criterion, result and what you will test next.
Over time, this record becomes more useful than any general advice, because it describes your own audience. It also shows the creative team which directions are worth developing and which have already been tested, so you don't pay twice for the same lesson.