Eight updates, one signal
OpenAI announced eight new capabilities for ChatGPT Ads in late July 2026. Taken individually, each looks like incremental platform work. Taken together, they answer the question performance teams have been asking since the platform launched in March: when does this have the infrastructure to run seriously?
The answer, as of this week: now.
The updates, confirmed by Search Engine Land and MarTech, cover bidding, measurement, attribution, campaign operations, and ad format:
- oCPC (optimized cost-per-click) bidding via a Conversions objective
- Geographic exclusions for location-level campaign control
- AppsFlyer integration for mobile app attribution
- Adjust integration for mobile app attribution
- Bulk API for asynchronous campaign edits at scale
- Automatic advanced matching
- Product card ad format with price and star ratings
- Shift from daily budgets to weekly average pacing
That's not a feature release. That's a platform maturity checkpoint.
What oCPC actually changes
The difference between CPC and oCPC is the optimization signal. With standard CPC, the platform delivers clicks. With oCPC, you select the Conversions objective and the system optimizes toward clicks more likely to result in a conversion — while still billing per click.
If your ChatGPT Ads tests so far have been optimizing at impression level, you've been running what amounts to a display awareness campaign with no conversion signal feeding back. oCPC changes that. You're giving the platform a performance target and letting it route toward users whose conversational context matches your conversion profile.
It matters most in categories where intent varies significantly within the same search query. A "best protein powder" conversation could be casual browsing or active purchase research. oCPC allows the system to differentiate based on surrounding context signals. Standard CPC can't do that.
The tradeoff: oCPC needs conversion data to optimize properly. Get the OpenAI pixel installed and conversion events flowing before switching to oCPC — otherwise the algorithm has no signal to optimize against and the Conversions objective won't outperform basic CPC.
Mobile measurement is now in the stack
The AppsFlyer and Adjust integrations are the most significant update for a specific group: brands running app install campaigns. Both platforms have published setup documentation in their partner directories, with integrations confirmed live as of mid-July 2026.
What the integration enables: app install and in-app event attribution from ChatGPT Ads flows into your existing MMP. If you're running UA campaigns across Google, Meta, and TikTok, you can add ChatGPT to that measurement stack without creating a separate reporting silo.
For app-first brands, this was the critical gap that made ChatGPT Ads unusable before. You weren't going to allocate app install budget to a channel where attribution was opaque. That constraint is gone.
The integration is available in the markets where ChatGPT Ads are live. Implementation lives on the MMP side — find the ChatGPT Ads entry in your AppsFlyer or Adjust partner directory, follow the setup flow, and verify the postback connection before spending.
Geo exclusions and the Bulk API
Geo exclusions let advertisers remove specific locations from campaign delivery. If you have regional shipping restrictions, legal constraints in certain markets, or territories where your LTV data is weak, you now have a lever for it. Previously, targeting was defined by market availability with limited granularity.
The Bulk API is primarily relevant for agencies and brands managing multiple campaigns or product lines. Asynchronous campaign edits at scale — bulk creative changes, budget shifts across multiple campaigns, targeting adjustments — are now possible programmatically. Smaller advertisers won't feel this immediately. Teams managing complex account structures will.
Weekly average budgets replace daily maximums. If your weekly budget is $700, the platform can spend $140 on a high-intent day and $60 on a slower one, as long as the weekly total holds. This is standard behavior on Google and Meta. Its addition normalizes how buyers plan ChatGPT Ads spend.
Who this changes the math for
App marketers who've been waiting for MMP integration: the barrier is gone. Run a four-week structured test with AppsFlyer or Adjust connected, set a cost-per-install target, and get a baseline CPI comparison against your current channels.
DTC brands with high-AOV considered purchases: ChatGPT's commercial query context — product research, comparison questions, brand evaluation — aligns naturally with longer consideration cycles. oCPC lets you optimize for buyers rather than browsers. If your average order value is above $80 and your customers research before purchasing, the conversion economics now have a real shot at working.
Brands running multi-market campaigns with regional exclusion needs: geo exclusions combined with oCPC gives you a functional performance setup with geographic control. That combination wasn't available before this week.
The group still on hold: impulse-purchase brands with sub-$30 AOV and thin margins. The optimization infrastructure now exists, but the math is still difficult at low order values.
ChatGPT Ads CPMs dropped from around $60 at launch to roughly $25, as reported by earlier analysis. The auction is not crowded. Low CPMs plus real performance optimization is the definition of a test window. oCPC won't keep costs low indefinitely — more sophisticated tooling attracts more budget competing for the same inventory.
If you want to know where your account measurement setup stands before testing a new channel, the free audit at Gromerce shows you which gaps matter most in three minutes.
Sources: Search Engine Land, MarTech, MediaPost, OpenAI Blog, July 2026
