Yes, since September 2026. Conversion Optimization is generally available for standard click campaigns, and OpenAI's own list of eligible events includes leads, registrations, trials and subscriptions, which are the events a services business lives on. In August this guide said the platform was halfway across, and that the day conversion bidding reached click campaigns would be the day ChatGPT Ads became a performance channel for everyone. That day has come, with the first published benchmarks attached and small print worth reading before you move a budget.
What changed in August
Platforms rarely announce a change of category. They ship features, and one week the features add up to something different. The first week of August 2026 was that week for ChatGPT Ads. Conversion-optimised cost-per-click arrived in beta for product feed campaigns, with clone and bulk tooling to create those campaigns at volume. Dynamic URL macros landed, so campaign, ad group, ad and account IDs populate themselves at delivery. Pixel validation diagnostics appeared, showing which events were dropped and why. Triple Whale and Hightouch built integrations, and on 21 August OpenAI added a Maximize results bid strategy, one-day view-through conversion reporting and a third measurement vendor, WorkMagic.
The integrations are the real signal
Any ad platform can ship a bidding option. The more telling item was that Triple Whale and Hightouch both built for this channel in the same week. One is an ecommerce attribution stack and the other is a customer data platform sitting on the warehouse, so between them they cover how a serious advertiser measures and how a serious advertiser feeds data back. Vendors build connectors when their customers ask for them, and customers ask when they are already spending. The bulk and clone tooling pointed the same way: bulk creation only matters to someone running many campaigns, so building it was an admission that those advertisers now exist.
What changed for services in September
The August release left one qualifier in place: conversion bidding was available for product feed campaigns, which need a catalogue of roughly a thousand products, so a law firm, an agency, a clinic or a B2B software company could not use it. That qualifier is gone. OpenAI's September stat pack announces Conversion Optimization in general availability and lists the events it optimises toward as leads, orders, registrations, trials, subscriptions and content views. Two of the six accompanying case studies are services advertisers: a financial services advertiser optimising for leads, and a jobs and recruiting advertiser optimising for sign-ups.
The eligibility filter OpenAI used for its own benchmark is the practical checklist. An advertiser counted if it had at least five click-attributed conversions on the event, at least $100 of spend and three active days per bidding strategy. Read that as the minimum signal the bidder needs before it can do anything useful, and as the reason the pixel and the Conversions API come first. Every advertiser in the published sample had conversion tracking running before the switch.
The numbers, with their small print
Across 27 advertisers who ran both strategies on the same events in July 2026, conversion bidding showed a median 44% lower cost per result than click bidding, within the same advertiser. The six case studies are before-and-after or side-by-side comparisons over 14 or 28 days, and OpenAI labels every one of them observational.
| Advertiser type | Click-through rate | Cost per result | Basis |
|---|---|---|---|
| Software and technology | +27% | Cost per order down 50% | 14 days before and after, 98% of spend moved |
| Software and technology | +22% | Cost per order down 87% | 14 days before and after, 73% of spend moved |
| Jobs and recruiting | +180% | Cost per sign-up down 95% | 14 days before and after, 94% of spend moved |
| Retail and ecommerce | +7% | Cost per order down 65% | Same 28 days, side by side, 40% of eligible spend |
| Travel and hospitality | +72% | Cost per booking down 80% | 28 days vs prior 28, bookings up 123% |
| Financial services | +4% | Cost per lead down 50% | 14 days before and after, 70% of spend moved |
Read the click-through column together with the cost column. A 180% jump in click-through rate alongside a 95% drop in cost per sign-up does not mean the creative improved; it means the bidder chose different conversations to appear in. Conversion bidding changes where the ad shows up, which is the same lever our context hints pull by hand, automated against your own conversion data. The travel case even says so in its footnote: the gain coincided with a change in bid mix and the conversion signal ramping up.
The named advertisers in the pack say the same thing in prose. Figma reports higher click-through rates than traditional search for its Figma Make campaigns, HubSpot reports competitive CPAs and downstream revenue signals, and Best Buy reports click-through rates higher than anticipated and demand across a broader mix of categories than expected.
Two more figures belong next to these. Running the Conversions API alongside the pixel measured 57% more attributed conversions and a 28% lower measured CPA than the pixel alone, median advertiser across more than 1,000. And the product feed benchmark, a 74% higher click-through rate for feed-active accounts, was drawn from more than 10,000 advertisers above $100 of spend in a single month, which is the first advertiser count OpenAI has put in writing and a fair measure of how busy the auction already is. The full breakdown is in the Conversion Optimization guide.
What to do about it now
The sequence matters more than the switch. Every published result sits on top of tracking that was already working:
- +Install the pixel and the Conversions API together, and check the diagnostics panel once at install and after any change to your forms or checkout.
- +Pick the one event you want the bidder to buy and let it reach five click-attributed conversions on click bidding first. If your primary event is too rare, optimise toward the leading signal that fires often and move up once the primary has volume.
- +Clone the click campaign into a Conversion Optimization campaign, keep the original, and run both for two weeks so you have a within-advertiser comparison like the ones OpenAI published.
- +Move your ID tagging onto the URL macros and keep writing source, medium and campaign by hand, so the comparison is readable in your own analytics.
- +If you already run Triple Whale, Hightouch or WorkMagic, connect this channel so it is judged next to your others instead of in isolation.
The tracking setup is covered step by step in the conversion-tracking guide, and what to do about the AI traffic that never shows up in analytics is in the measurement guide.
The honest answer
A channel becomes a performance channel when you can point it at a business outcome, measure that outcome credibly, and compare it to your other channels on the same terms. As of September 2026 ChatGPT Ads satisfies all three for any advertiser with working conversion tracking, whether or not it sells from a catalogue. What it does not yet have is a body of independent evidence: the benchmark sample is 27 advertisers, the case studies are chosen by the platform, and everything is observational. Treat the 44% as the platform's claim about a small early group, and treat your own two-week side-by-side as the number that counts.
The reason to run that test now is the same as it was in August. This channel reaches the part of the buying journey that search never saw, the problem framing and discovery and comparison that happen before anyone types a product name, and the auction is still thin in the markets that opened this summer. When you want that test built, tracked and read properly, that is what our managed service does.
