Images & Video · Amazon announcement · source The page shows no publish date; its footnotes cite 'Oneisall, UK, 2025'.
A/B-Tested Ad Images with Amazon Ads AI Image Generation
Produce enough on-brand ad image variations to A/B test across ad types
“sales increased by more than 50% year-over-year”
Per source From Amazon itself, not independently tested.
Copy-paste prompt
Write a master prompt for Amazon Ads image generation for my brand [brand]. Include: color palette [colors], lighting [style], mood [mood], and things to avoid. Then write 6 scene variations for [product] (backgrounds and in-use settings) that keep those rules, so I can test them as separate Sponsored Products ad images.
The source didn't publish its exact prompt. We wrote this starter so you can try the same workflow. Replace anything in [brackets].
What they did
Oneisall, a China-based smart pet product brand selling in Europe, used Amazon Ads' AI-powered image generation across Sponsored Brands, Sponsored Products and display ads. The team wrote master prompts with brand guidelines (color palette, lighting, mood) for a consistent library, generated many variations per product with different backgrounds and settings, ran them in parallel campaigns to find winners, and made tailored display creatives for audience segments.
How it went
Amazon Ads reports sales increased by more than 50% year-over-year, ACOS fell 22% YoY, share of voice in the core keyword category rose by more than 50%, and ad recall beat the industry benchmark by 12 percentage points.
We only report what the source says. Results are self-reported unless noted.
Worth knowing
Amazon's own case study of its own tool, and the results cover a whole full-funnel strategy, not the images alone. The reusable part is the master prompt with brand rules.