The first three seconds: how we test creatives for app ads

On a social feed, your ad is competing with a friend's dog and a stranger's kitchen renovation. You have roughly three seconds before the thumb decides. Everything about creative testing follows from that number.

What goes in the first three seconds

The product, doing its one thing, for a specific person. A child's name appearing in a story. A scene being read aloud with the user's own city in it. A coffee card filling up at a counter. Not a logo, not a tagline, not an establishing shot. If someone watched only three seconds, they should know what the app does and for whom.

Formats that earn their place

• Screen recording with a voice. The app, used honestly, with someone talking over it like they are showing a friend.
• Reaction. A real person reacting to the first result. Works when the result is genuinely surprising, which is the test of the product as much as the ad.
• Problem first. Ten seconds of the problem, then the app. Risky, because the problem has to be instantly recognisable.

How to structure a test

One idea per creative. Variants change one thing at a time: the hook line, the format, the voice, the first frame. Three to five variants per idea, enough budget on each to produce trial starts rather than just views, and a decision at the end of the week.

What to measure

Thumb-stop rate and view-through tell you whether the hook works. They do not tell you whether the right people stopped. Judge creatives on cost per trial start, and keep an eye on trial-to-paid a few weeks later; a creative that attracts people with a promise the app does not keep looks great for a week and expensive for a month.

The mindset

Most creatives lose. That is not failure; it is the process. The point of testing is to find the few that work and then make ten variations of those. Teams that expect every creative to win stop testing after the first bad week, and that is where accounts go stale.