OpenAI ties ChatGPT's new image-side ads to 18 outside measurement and data partners

The partners fall on two sides of the data flow. Hightouch, Tealium and LiveRamp carry an advertiser's conversion data into ChatGPT Ads ^[11]. OpenAI also lists pixel-based measurement and Conversions API support ^[3]. Data that flows into OpenAI's system improves what OpenAI can report, but it cannot also be the independent check on that reporting.

I think the measurement layer is the right first build. A channel gets real budget only when it shows up in the same report as the channels it competes with. That check comes from the second group. OpenAI names ten attribution and measurement vendors, AppsFlyer, Adjust, Northbeam and Kochava among them ^[9]. It also names three full-funnel partners: Fospha, Measured and INCRMNTAL ^[10]. According to the dev.to summary of OpenAI's announcement, this gives advertisers a way to tie ad activity to outcomes in their own measurement systems instead of relying only on platform reporting ^[17]. The same writer notes that new inventory is hard to evaluate when marketers cannot compare it with established channels ^[15].

OpenAI has named 18 outside firms ^[18] and published no rates ^[8]. Cost-per-click bidding now sits alongside the existing CPM option ^[7].

The placement narrows what a test can show. The ad sits beside the generated image, not inside it ^[5]. The person who sees it has just asked ChatGPT to make a picture. The dev.to writer advises against assuming that placement suits every offer ^[16].

OpenAI cited early findings from WeightWatchers, Dose and Portland Leather on lift and new customer acquisition ^[13]. The dev.to writer calls them early pilot results, not broad benchmarks, and warns against reading them as a forecast for any future campaign ^[14]. For those results to carry over, a new advertiser would need an offer that fits the moment after someone generates an image. It would also need to reach an audience like the one in the first test.