ChatGPT Ads Got a Real Bidding System

OpenAI shipped eight new advertiser tools for ChatGPT Ads on July 24, 2026. The headline addition is conversion-optimized bidding, a feature Google Ads and Meta Ads have offered for years.

For marketers who have been testing ChatGPT as a paid channel, this update closes a gap that made the platform hard to defend in a budget review.

When ChatGPT Ads Launched, and What They're For

OpenAI announced its advertising plans on January 16, 2026. The pilot went live three weeks later, on February 9, 2026, showing sponsored placements to logged-in adult users on ChatGPT's Free and Go ($8/month) tiers in the US. Plus, Pro, Business, Enterprise, and Education subscribers have never seen ads.

OpenAI needed a second revenue line beyond subscriptions to offset its compute costs, and advertising let it keep the free tier viable without gating access. CEO Sam Altman had called ads a "last resort" back in 2024. The February launch reversed that stance.

Ads appear as a labeled "Sponsored" unit below ChatGPT's response, not inside the answer itself. OpenAI has been consistent: ads don't influence what the model says, only what appears next to it. Early pricing ran on a CPM model, starting near $60 per thousand impressions.

The platform moved fast after that. A self-serve Ads Manager entered closed testing on April 7 and opened publicly on May 5, with CPC bidding available to any US business.

Bids in that model run $3 to $5 per click. The pilot's original $200,000 minimum spend dropped to $50,000 for CPC advertisers, then disappeared entirely once self-serve launched.

Revenue estimates vary widely depending on the source. OpenAI's ad business reportedly reached $2.4 billion in 2026, with internal projections putting it at $100 billion for 2030.

EMARKETER, by contrast, expects the entire US chatbot advertising market to stay under $1 billion this year. That gap says less about ChatGPT specifically and more about how immature measurement still is across the category.

ChatGPT's New, Stronger Bidding System

The July 24 release is the first time ChatGPT Ads has offered true outcome-based bidding. Here's what shipped, according to OpenAI's product update to Ads Manager beta participants:

Conversion-optimized campaigns (oCPC): Advertisers select a Conversions objective, and the system automatically shifts delivery toward clicks more likely to convert. Billing still runs on a per-click basis, so the pricing model hasn't changed, only what the algorithm optimizes for.

Average daily budgets: Starting the following week, daily budgets move to a rolling seven-day average instead of a strict daily cap. Spending can flex up or down day to day as long as it nets out within the weekly limit.

Automatic budget pacing: Daily spend now distributes across the day rather than front-loading or exhausting early, a standard feature on Google Ads that ChatGPT lacked until now.

Geographic exclusions: Advertisers can remove specific locations from targeting, not just add them. This matters for any brand running region-specific promotions or working around legal restrictions in certain states or countries.

Mobile measurement integrations: ChatGPT Ads now connects to AppsFlyer and Adjust, the two largest mobile measurement partners. App advertisers can finally attribute installs and in-app events to ChatGPT campaigns, something that had no supported path before this update.

Automatic Advanced Matching: Website conversion tracking improves through hashed customer data, similar to Meta's Advanced Matching. It's enabled under Tools > Conversions > Data Source in Ads Manager.

Bulk API updates: The Ads API now supports asynchronous bulk creation and edits across campaigns, ad groups, and ads, which matters for any team managing more than a handful of campaigns manually.

Refreshed product feed ads: Product feed campaigns are starting to show updated cards with pricing and star ratings, bringing them closer to Google Shopping ad units.

None of these are novel concepts. Every one mirrors a feature Google Ads or Meta Ads has run for years. That's the point. OpenAI isn't trying to reinvent ad tech; it's trying to remove the reasons a performance marketer would say no.

How ChatGPT Is Doing Things Differently Now

In early June, OpenAI was testing a multi-advertiser ad format, where more than one brand could appear in a single sponsored placement sold through a second-price auction.

At the same time, it added the ability to convert lifetime budgets to daily budgets and clone CPM campaigns into CPC campaigns with one click. Bulk editing existed in a limited form.

The July update builds directly on that groundwork instead of introducing a flagship feature. It's infrastructure work: bidding logic, attribution, budget mechanics, API scale.

The speed is also different from what advertisers are used to. Five months separate the February 9 pilot launch and the July 24 conversion bidding release.

Google Ads took years to build out equivalent conversion optimization tooling. ChatGPT Ads is compressing a roadmap that established platforms spent a decade on into a single fiscal year, which tracks with how OpenAI has approached most of its product surface.

The category difference also matters. Google Ads and Meta built conversion optimization around intent signals collected passively, such as search queries, browsing behavior, and page views. ChatGPT's signal is a live conversation.

A user asking "compare CRM platforms for a 15-person sales team" is handing over more explicit purchase context than a search query ever would. Conversion bidding on top of that kind of signal is a different proposition than conversion bidding on top of keyword match types, even if the mechanics look similar on paper.

Conclusion

Conversion-optimized bidding, mobile attribution, and bulk campaign tools bring it in line with the baseline advertisers expect from Google Ads and Meta Ads. That baseline is what turns a curiosity budget into a line item.

Attribution is newer and less battle-tested, inventory is still capped to Free and Go tier users, and independent revenue estimates for the category still disagree by billions of dollars.

For marketing leaders evaluating ChatGPT as a channel, the calculus has shifted. Testing no longer means accepting click volume as your only signal.

You can now optimize toward business outcomes, exclude geographies that don't convert, and manage campaigns at scale through the API instead of a manual dashboard. That's enough infrastructure to justify a real test budget, not just a curiosity spend.