Asset generation is the set of generative AI tools inside Google Ads that write headlines and descriptions and create or edit images for your ads, based on your website, a short prompt and any assets you already have. You review the suggestions and choose which to use before anything goes live.
How asset generation works
When you build or edit a campaign, typically Performance Max or Demand Gen, Google offers to generate assets. For text, it reads your landing page and any description you give and suggests headlines and descriptions in the right lengths. For images, you type a prompt such as “a bright kitchen with a freshly fitted worktop” and Google produces several options. You can also edit existing images: extending the background to fit a different shape, removing a background, or creating variations of a product shot.
This is different from automatically created assets, which Google assembles at serve time and which you switch on or off at campaign level. Generated assets are suggestions you approve at the time you build the campaign, and they then behave like any other asset you uploaded. Google has said that images made by its AI in Google Ads carry an invisible watermark and metadata identifying them as AI-generated. At the time of writing (October 2026) the exact features vary by campaign type and are changing quickly.
Why it matters
Automated campaigns perform better with a wide range of assets, and many small UK businesses do not have a library of photographs or a copywriter on hand. Asset generation fills that gap, letting a business supply the variety an asset group needs without a photo shoot for every campaign. It is also quick for producing the different image shapes each placement asks for.
The risk is accuracy. The UK’s CAP Code, enforced by the Advertising Standards Authority, applies to an ad whoever or whatever wrote it. A generated headline promising “same-day service across London” or a generated image showing a finish you cannot deliver is a misleading ad, and the responsibility is yours, not Google’s. Generated images of people, premises or results can also mislead if customers assume they are real.
Common mistakes
- Accepting every generated headline without checking it is true for your business, your prices and your service area.
- Using generated images that show products, finishes or premises that do not exist, which can breach advertising rules and disappoint customers.
- Letting generated copy drift from your brand voice, so ads sound unlike your website.
- Replacing real photographs that customers trust with generic AI scenes.
- Leaving American spellings or terms in generated text, such as a US postcode format or spelling, which look out of place to UK buyers.
How to act on it
Use asset generation as a first draft, not a final one. Give it a precise prompt that includes your service, area and the tone you want, then edit what it returns. Check every claim against what you can prove. Keep real photographs of your work, team and products as the core of your image assets, and use generated images for backgrounds, shape variations and scenes that do not misrepresent what you sell.
Once live, watch the asset ratings and remove generated assets that rate low or attract the wrong enquiries. Write down which prompts produced useful results so the next campaign starts from something proven. I use these tools, carefully checked, when building Performance Max campaigns that need a broad asset set.