In short: OpenAI has rolled out nine ChatGPT Ads updates in one of the largest batches since launch. The most important is a new recommended buying model that optimizes for conversions from both clicks and views, and bills per impression. On top of that come flexible attribution windows, per-platform targeting, Shopify and HubSpot integrations, AI-assisted ad creation, text customization to the conversation, a quality score for your conversion tracking, and testing of sponsored agents. Here we go through all nine, one by one, with what they mean in practice.
From new ad surface to mature performance system
When we started our own ChatGPT Ads campaigns shortly after the channel opened in our market, Ads Manager was a young tool: Click-based buying, basic reporting and few integrations. This week's updates show the platform moving quickly from "new ad surface" toward a mature performance marketing system, and the pace is high.
We will take the updates one by one, with what each means in practice and what we plan to do with them ourselves.
1. Conversion optimization with impression billing
This is the most important update, and it is now generally available as OpenAI's recommended model for maximizing performance.
Previously you could think of optimization roughly like this: Person sees ad, clicks, converts. Now the campaign can optimize for conversions that happen both after clicks and after ad views, while you pay based on impressions rather than clicks alone.
An example: A marketing manager sees your ad in a ChatGPT conversation but does not click. Two hours later she looks up your company and books a demo. With the right attribution, that registers as a view-through conversion, and the campaign gets credit for a customer it actually helped create.
Our advice: Test the model against your current campaigns if your tracking has collected enough conversion data. But be aware that view-through conversions can make ads look better than they are. Keep watching actual bookings, qualified leads and what a customer costs you, not just the Conversions column in Ads Manager. The ladder from our buying models guide still holds: Clicks first for controlled learning, conversion optimization once the data is in place.
2. Flexible attribution windows
This one ties directly into point 1. You now decide how long after an ad interaction a conversion should be credited to the ad: A 7, 14 or 30 day click window, and a 0 or 1 day view window. You choose when customizing columns in Ads Manager, and the choice applies to all conversion columns displayed.
Why it matters: Say someone clicks your ad on September 1 and books a demo on September 12. With a 7 day click window, the booking is not attributed to the ad. With a 14 day window, it is. Same campaign, completely different numbers.
For B2B, where the decision journey tends to be longer than an impulse purchase in a webshop, this is especially relevant. We would start with a 14 or 30 day click window and analyze the difference. The 1 day view window can be useful, but evaluate view-through conversions separately from click-through when judging real effect.
3. Per-platform targeting
You can now separate Android app, Android web, desktop web, iOS app and iOS web, both in targeting and in reporting, in Ads Manager and in the Advertiser API.
It may sound trivial, but it becomes valuable as the data comes in. You might discover that desktop users produce fewer clicks but far more bookings, while iOS delivers high click-through rates and few conversions. Then you move your budget to where the business value is actually created, instead of guessing.
For B2B products, desktop web is often the surface to watch most closely, but do not conclude until you have enough data for the differences to mean something.
4. Shopify integration
This is bigger than a practical integration. US-based Shopify merchants can now install the ChatGPT Ads app from the Shopify App Store, connect their product catalog and measurement directly, create campaigns and track performance, all from Shopify. International availability is announced for later this month, so merchants elsewhere should watch this space over the coming weeks.
The flow potentially becomes: Shopify, product data, ChatGPT Ads, purchase, measurement, in one connected chain. That dramatically reduces friction for ecommerce advertisers.
From a GEO perspective, this is also a signal worth noting: ChatGPT keeps growing as an information surface, a product discovery surface and an ad surface at the same time. For ecommerce, organic AI visibility, product data and paid AI visibility are starting to converge.
5. HubSpot integration
The equivalent update for B2B. Businesses using HubSpot can connect their ChatGPT Ads account, create ads, track performance and follow up on leads directly in HubSpot.
That gives a much cleaner flow: ChatGPT conversation, ad, lead, HubSpot, sale, customer. And it is particularly interesting because you can eventually evaluate ChatGPT Ads on pipeline and actual revenue, not just impressions and clicks.
This is a development worth following even if you use a different CRM: The integrations are unlikely to stop with HubSpot.
6. AI-assisted ad creation
Fairly self-explanatory, but useful: When creating an ad, the system can use your landing page and campaign objective to suggest ad copy and imagery. You stay in control and edit the suggestions before adding them to a campaign. The feature is in testing, with broader availability expected in the coming weeks.
This makes Ads Manager more like modern Meta and Google advertising, where AI helps with creative production. Our advice: Use it for variations and testing, do not let the AI decide the entire message. You know your customers better than your landing page does.
7. Text customization to the conversation
This one is actually more interesting than the previous. You can now opt in to letting the system adapt existing headlines and descriptions to the context of the conversation the ad appears in, and translate the ad copy to the user's preferred language.
Imagine a base ad like "See how your business gets recommended in ChatGPT". Person A is talking to ChatGPT about digital marketing. Person B is talking about SEO. Person C is talking about how their brand shows up in AI answers. Instead of all three getting exactly the same wording, the copy can be shaped by the conversation each of them is actually in.
That is a fairly fundamental departure from classic advertising: The ad adapts to the conversation it appears in. And it confirms one of the clearest lessons from our own campaigns: Copy that fits the conversation gets clicked, while generic advertising language gets scrolled past. OpenAI is now building that insight straight into the product.
The setting is optional, and you decide whether it is enabled. We will be testing it, but keep a close eye on which phrasings are actually used and how they perform, it is still your brand on the sender line.
8. Event Quality Score
Check this one as soon as it is available on your account. Under Tools and Conversions, eligible Pixel and Conversions API setups get a score from 1 to 10, assessing among other things your user data coverage, with actionable recommendations for improving the implementation. Think of it as a health check for your conversion tracking.
It matters more than it sounds. Poor tracking does not just mean poor reporting: If your campaigns optimize toward conversion events, poor signals also mean the algorithm gets worse data to learn from. With the new recommended buying model from point 1, that becomes doubly important.
The minimum that should be measured as robustly as possible: Ad, website, booking or purchase, conversion back to the campaign. The full recipe is in our pixel guide, which now also covers Event Quality Score.
9. Sponsored Agents
Strategically, this is by far the most exciting update. OpenAI is testing with select advertisers in the United States that after clicking an ad in ChatGPT, users may be able to start a clearly labeled conversation with a business-sponsored agent.
Instead of the classic journey, ad, click, website, hunt for information, convert, it could become: Ad, conversation with the business's agent, ask questions, evaluate the product, take the next step. Such a conversation could look like this:
- User: How do I know if my company is visible in ChatGPT?
- Agent: We measure how your business is mentioned and recommended across the AI models, and compare you to your competitors.
- User: What does it cost?
- Agent: Here are the plans, and here is where you can book a demo.
The ad itself evolves from an entrance to your website into the start of a sales conversation. And it fits perfectly with how ChatGPT is used: The user is already in a conversational interface and never has to break out of the experience to find information. The test is limited to the US for now, but if it opens up more broadly, this could become a significant new way for B2B companies to convert interest into leads.
What it all adds up to
The most interesting part is looking at the updates together. ChatGPT Ads is now getting the building blocks we know from established ad platforms: Targeting, AI-generated creative, dynamic copy, conversion optimization, attribution, Pixel and Conversions API, CRM and ecommerce integrations, and conversational agents. The platform is going from "ads inside ChatGPT" to looking like the start of a fully fledged performance marketing platform.
Our priorities, both for our own campaigns and for our clients: Test conversion optimization with impression billing, set the attribution windows deliberately, segment by platform, test text customization, and check Event Quality Score right away. Sponsored Agents we are watching closely.
Want help putting the updates to work?
We run ChatGPT Ads for ourselves and for our clients, and adopt the new features as they roll out. If you want to know what these updates mean for your campaigns, or get started while competition is still low, book a meeting and we will walk through it together. And if you prefer to read up first, our pillar guide to AI advertising is the place to start.