OpenAI to test visual ads inside ChatGPT image generation
OpenAI will begin testing a visual ad format in ChatGPT later this month, with the ads appearing during image generation rather than below a text response. The test starts with a small group of US advertisers, and OpenAI says the placements will be clearly labelled and kept separate from the image being created.
The format is the first expansion of ChatGPT Ads beyond the sponsored slots currently shown under responses. OpenAI says ChatGPT reaches 1.2 billion people a week, and that users split their time between asking questions, researching purchases, creating images and working with files. Putting ads into image generation opens the format to more of those sessions.
OpenAI has not disclosed targeting, pricing, creative requirements or reporting for the new placement, and there is no timeline for wider availability. Advertisers with placement restrictions can use a new Negative Phrases control, and OpenAI is running brand suitability pilots with DoubleVerify and Integral Ad Science that do not give the partners access to private conversations.
Alongside the format test, OpenAI added conversion integrations with Hightouch, Tealium and LiveRamp, and is working with Haus, Measured and WorkMagic on geo-based incrementality experiments. Early partner figures include a 15.3% lower attributed CPA than a blended paid search benchmark for WeightWatchers, reported by DV Rockerbox, and 93% of Portland Leather visitors from ChatGPT Ads being new, reported by Triple Whale. OpenAI describes the incrementality work as early stage.
Reported by Search Engine Journal ↗
Topics: ChatGPT, AI advertising, Measurement