When will ChatGPT visual ads launch? US test eligibility and conversion measurement

This is an announced US pilot, not universal availability

OpenAI's October 5, 2026 announcement says visual ads will initially be tested during ChatGPT image generation, starting later that month in the United States with initial advertisers. Ads are clearly labeled, separate from the image being generated, and do not influence ChatGPT's answers.

Brands can prepare creative and measurement plans, but cannot interpret this as every account already creating the format. No exact start date, initial-selection process, or complete creative specifications were given. October 5 is the announcement date, not universal activation.

Separate account access, advertising eligibility, and format eligibility

The Ads Manager availability help page requires the advertising and billing legal entity for self-service accounts to be in a supported country. The United States is listed, but account registration and self-service access differ. Registration cannot advance access for an unsupported location.

General Ads Manager access does not establish approval for the visual format. Check the correct account's review, verification, and availability information, then test eligibility for this pilot. Do not repeatedly create accounts when the format is absent or treat a different network location as satisfying entity requirements.

General ad rules do not establish this format's coverage

Current ad guidance says Free and Go users may see ads; Plus, Pro, Business, Enterprise, and Edu are ad-free, and accounts identified as under 18 do not see ads. These are general rules, not proof that every US Free or Go user will encounter this visual test.

Ads are labeled paid placements, not an OpenAI recommendation or endorsement. Advertising and measurement do not give brands private ChatGPT conversations. Messages users actively send to a merchant through an ad are content provided directly to that merchant.

Understand measurement through existing conversion data

The same announcement expands internal and partner measurement. Pixel and Conversions API, or CAPI, send purchase, lead, install, or registration signals. Hightouch, Tealium, and LiveRamp help connect existing enterprise-system data.

Partner guidance requires an active Ads Manager account, appropriate permissions, a suitable data source, and an event inventory. Some integrations need an Ads Manager API key; others use Pixel ID and CAPI key. Do not impose one integration on every partner. Successful event submission does not establish attribution eligibility.

Incrementality measurement is early, with geographic experiments being explored; brand-lift and suitability work also has pilots. Attribution assigns results under reporting rules; incrementality asks what extra business ads caused. One attributed conversion cannot establish causality.

How do 7-, 14-, and 30-day clicks compare with 1-day views?

Measure Results currently allows 7-, 14-, or 30-day click windows and view attribution set to 0 Day, off, or 1 Day. A view conversion requires a qualifying impression followed by conversion within a day without an eligible click taking credit for the same conversion. Eligible clicks take priority.

With a 1-day view window, total Conversions includes clicks and qualifying views, which can be inspected separately. Do not add the view column to that total again. With 0 Day, totals include click attribution only. Align settings before comparing reports.

Changing reporting windows affects attribution, not optimization, bids, or billing, and does not create multi-touch attribution. The documented path is Edit columns in Campaigns, Ad groups, or Ads, then Attribution windows under Conversion Events.

Check conversion latency and the reporting time basis together

Conversions can take 24–48 hours to appear, and recent totals can change as events are processed. By default, Conversions is assigned to the attributed click or impression time. Use Conversions (by conv. time) to report by the conversion's actual occurrence.

Record account, dates, time zone, event names, and both windows before comparing Ads Manager, site analytics, or partner results. Clicks, sessions, and attributed purchases differ. Page loading, consent, and reporting definitions can cause discrepancies; a same-day zero alone does not establish ineffectiveness.

What is known about Negative Phrases?

The announcement confirms Negative Phrases for eligible advertisers with narrow targeting requirements under their own brand policies. Verified materials do not specify eligibility thresholds, configuration, matching, count limits, or exclusion guarantees. Do not present a universal negative-keyword setup tutorial.

DoubleVerify and IAS suitability assessments remain pilots in controlled environments that do not access real users' private conversations. That does not independently certify every placement or guarantee absolute compliance with brand requirements at each impression.

What can you prepare now, and what must wait?

Confirm entity eligibility and account access, prepare real products, landing pages, and creative, and inventory conversion goals such as purchases or registrations with data sources and owners. Choose a measurement method already authorized and retain window and event settings for consistent reporting.

Wait for explicit account access or official details for visual-test participation, specifications, and Negative Phrases. Partner cases describe particular studies or brands, not your data or an ROI promise. When an entry appears, check formats, countries, and creative requirements. An absent entry is not automatically an account fault.