TLDR;
ChatGPT Ads has added practical tools for running larger retail campaigns and reading more of what happens after an ad. Advertisers can build product campaigns in bulk, inspect product review problems and report conversions beyond the event used for optimisation. The upgrade makes the platform easier to operate, but attribution settings can change reported conversion totals, so enterprise teams should lock their measurement definitions before comparing performance.
What happened
Search Engine Land reported the advertiser rollout on 30 September. The update includes spreadsheet-based bulk creation for product-feed campaigns, review-status detail for rejected products, more conversion reporting and access in Thailand, Indonesia and Malaysia. OpenAI did not publish a separate launch post, so the correct description is a reported rollout backed by updated product documentation.
OpenAI’s Advertiser API changelog confirms a dated measurement change. Advertisers can now select click and view attribution windows and choose the attribution time basis across four Insights endpoints. If those settings are omitted, reports default to a 30-day click window, a one-day view window and ad-event time. OpenAI warns that conversion totals can differ from earlier reports even when impressions, clicks and spend stay the same.
The product-feed workflow is also more operationally useful. OpenAI’s help documentation says feeds can be uploaded manually, pulled from a hosted URL or sent through SFTP. Items expire after two weeks, which makes automated refresh a dependency for enterprise catalogues. Products remain eligible only for the advertising beta and do not gain organic ChatGPT visibility from the feed.
Reporting can now separate product impressions, clicks, spend and card-level engagement. OpenAI’s technical reporting guide also supports goal and non-goal conversion detail. That matters when a campaign optimises for a lead but the business still needs to see later sales, or when several products share one conversational ad.
Why it matters
This is the kind of unglamorous release an ad platform needs before enterprise budgets can move. A retailer with thousands of products cannot build campaigns one item at a time, and a performance team cannot make investment decisions from a single blended conversion total. Bulk creation, rejection reasons and product-level reporting reduce operational friction and make catalog problems visible.
The measurement change also creates a trap. Adding view-through conversions or changing the time basis may improve the number in the interface without changing a single customer outcome. OpenAI’s measurement guidance says view-through reporting uses a one-day window when enabled and that reporting changes do not alter bidding or billing. Finance and analytics teams therefore need a stable comparison basis before they call the rollout an improvement.
Product feeds are becoming part of paid AI visibility, but they remain separate from organic recommendation. Brands should not assume that buying feed ads teaches ChatGPT to recommend the same products in an independent answer. Paid delivery, organic shopping inclusion and on-site conversion are three different systems.
How your brand can benefit / be affected
Start with feed hygiene. Use stable item IDs, current price and availability, complete images and clear product titles. Automate refreshes well inside the two-week expiry window and route rejection reasons to the team that owns the source field. A campaign spreadsheet is useful only if it does not become a second, stale catalogue.
Define the reporting contract before launch. Record the click window, view window, attribution time basis, campaign goal and the non-goal events you want to inspect. Keep those settings beside exported results and compare them with first-party analytics and order data. When a definition changes, mark the break rather than joining the periods into one trend line.
Use bulk tools to create controlled experiments, not uncontrolled volume. Group products by margin, stock position and customer need, then compare product-card engagement with completed sales. Keep a separate view of organic ChatGPT discovery so paid campaign results do not hide a weak catalogue, unclear product evidence or poor landing-page experience.
News date: 30 September 2026. Editorial review: 2 October 2026.