Skip to content

AI ad creatives: which tool to use

An ad creative is a picture or a clip, on-screen text and a hook. For static ads use image generation in the Photo section (GPT Image 2.5, Seedream 5.0 Pro), for motion a clip of three to fifteen seconds, and for headlines a prompt with the text in quotes. Test five to ten variants of one idea and change one thing at a time.

Updated:

How to build a batch

  1. One idea, several formats

    Make one scene as a static image and as a video. That shows whether the still or the motion works better.

  2. Change one thing at a time

    Background, presenter, headline or the first seconds. If you change everything, you cannot tell what worked.

  3. Name and compare

    Give each variant its own number in the file name. It is easier to match results in the ad dashboard.

Frequently asked questions

What picture do I need for a static creative?

A clean composition with space for text. For posters and banners the models in the Photo section are easiest, with a frame format that fits the placement.

How many variants should I make per idea?

Usually five to ten. The differences should be visible: another presenter, another background or another first frame.

Can I use ready prompts?

Yes, the prompt library has approaches for covers, posters and product cards. Change the name, colors and text to fit your product.

Do ad creatives need real people?

Not necessarily, but a person in the frame often catches the eye. Just do not present a generated face as a real customer.

How do I know which creative is better?

Look at the ad dashboard: clicks, retention and cost per result. One winning idea usually matters more than ten similar variants.