ChatGPT Will Show Ads While It Generates Your Images
OpenAI begins testing with early advertisers in the US in October. The company says it is expanding its ad measurement tools and filtering ads out of sensitive conversations.
OpenAI announced on October 5 local time that it is bringing a new ad format to ChatGPT: advertising that uses images to show a product in use or a related experience. The first test will place ads in front of users while ChatGPT generates an image for them. The company is also expanding the tools advertisers use to measure results.
What ads during image generation look like
The test starts in October with early advertisers in the US. Businesses can apply to advertise on ChatGPT starting the day of the announcement. OpenAI said ChatGPT has 1.2 billion weekly users.
Ads carry a label identifying them as advertising, and they appear separately from the image the user is creating. The layout is meant to prevent confusion between the requested image and the ad. OpenAI said advertising does not influence the content of ChatGPT’s responses.
Measuring whether ads led to purchases
Advertisers can send conversion data stored in their own systems, meaning records of outcomes such as purchases, to ChatGPT Ads in order to review ad performance. To support this, OpenAI is integrating with the customer data companies Hightouch, Tealium and LiveRamp. It is also adding more partners that analyze which ad a purchase on the web or in an app originated from. The integrations support data transfer, performance reporting and post-click attribution analysis.
OpenAI also shared advertiser results. For WeightWatchers, the cost of acquiring a single customer through ChatGPT Ads came in 15.3 percent below its paid search benchmark. At Dose, 67 percent of incremental purchases came from new customers. For Portland Leather, 93 percent of the visitors arriving via ChatGPT Ads were first-time visitors.
The company is also working on measuring how much advertising actually increases purchases. OpenAI said it is exploring geography-based experiments with partners. The aim is to move beyond tallying post-click outcomes and gauge the effect the advertising itself creates. The company described this measurement work as still being at an early stage.
Will ads appear in sensitive conversations?
OpenAI said it evaluates whether the content of a conversation is an appropriate place to show advertising. The company said this mechanism helps screen out ads when a user is in emotional distress or discussing sensitive topics. Qualified advertisers can also exclude their ads from conversations containing specific phrases. OpenAI is preparing a pilot program to check whether ads appear in brand-appropriate contexts, with third-party verification partners conducting the review in an environment where they cannot see users’ private conversations.
For users, the change means seeing a separately marked ad while an image is being generated. For advertisers, it means more tools to track purchases and new customer acquisition. What to watch next is how the ads actually surface during the US test, and how this early-stage work on measuring ad effectiveness develops.