OpenAI will start testing visual ads inside ChatGPT image generation later this month, with an initial group of US advertisers, according to a Search Engine Land report by Anu Adegbola published October 5. The ads are not live yet. Alongside the format, OpenAI named a long list of measurement and brand suitability partners, which suggests the company wants buyers to judge ChatGPT Ads with tools they already own.
The new format is built around imagery that shows products, services, or the experiences they enable. In the test, an ad appears while a user waits for ChatGPT to produce a requested picture. OpenAI says the ad is labeled and stays separate from the generated image. Search Engine Roundtable’s Barry Schwartz, who covered the same announcement, shared OpenAI’s mockup of an image mid-render, with an ad already loaded beneath it.
Measurement carries more of the news than the creative does. Hightouch, Tealium, and LiveRamp integrations let advertisers pass conversion data from their own systems into ChatGPT Ads. The attribution roster lists AppsFlyer, Triple Whale, Adjust, DV Rockerbox, Northbeam, Branch, Singular, Kochava, Airbridge, and Tenjin. For teams that already run those tools on paid search, ChatGPT can sit in the same reports.
Attribution alone cannot settle whether an ad caused a sale. For wider funnel measurement, OpenAI has lined up Fospha, Measured, and INCRMNTAL. Haus, Measured, and WorkMagic are piloting experiments split by region, which compare regions that see ads with regions that do not, to test whether ChatGPT Ads create conversions or only collect credit for them. OpenAI describes its incrementality work as early.
The first results come from partners, not from an independent benchmark. OpenAI relayed that DV Rockerbox found WeightWatchers paid 15.3% less per attributed acquisition on ChatGPT Ads than its blended paid-search average. WorkMagic found that 67% of Dose’s incremental purchases came from net-new customers. Triple Whale put first-time visitors at 93% of the traffic Portland Leather received from ChatGPT Ads. Each figure covers one brand and one method, and none shows what a typical advertiser should expect.
Brand safety is the remaining piece. Brand suitability pilots are in development with DoubleVerify and Integral Ad Science. Those firms will test OpenAI’s safeguards in controlled settings, and private user chats stay off limits to them. Eligible advertisers also get a Negative Phrases option for narrower placement rules. Neither pilot has published findings.
OpenAI claims 1.2 billion weekly users for ChatGPT. That is the company’s claim, and the announcement offers no ad-specific reach number to go with it. For a rough sense of delivery, Graphite’s report on ChatGPT Ads reach and click-through rates from September 30 is the nearest independent reading on our site.
The stake for search teams is budget. Every dollar that moves into ChatGPT Ads comes from a pool that also funds paid search, and the case for moving it now rests on attribution that the platform’s partners supply. Image generation is also a different moment from a research query, so a visual ad there may reach people who are not shopping yet. Outside measurement is what would show whether those impressions matter.
Search marketers who hold a test budget should ask for incrementality results before they accept attributed CPA as proof, and should keep a paid-search control group running through any ChatGPT trial.
Based on reporting by Anu Adegbola in Search Engine Land, published October 5, 2026, with additional detail from Search Engine Roundtable.
