The ChatGPT Ads Manager beta has now arrived in the UK, and we wanted to get in early and test it. That’s what we’ll be discussing here.
It is not yet as advanced as the advertising tools available through Meta or Google, but the current beta is already giving us useful insights into what the platform can do today, what it may be gearing up to support in the future, and how it differs — and will no doubt continue to differ — from established search and social advertising platforms.
We thought it was a great time to put together a short explainer.
Brands can already create campaigns, set budgets and bids, upload creative, measure performance and manage activity through a self-service interface. Sounds great! But the devil is in the detail:
OpenAI does not allow advertising to influence organic results
The organic response and the ad placement operate separately, and advertisers cannot shape, rank or alter what ChatGPT tells the user. They do not receive access to private conversations, chat histories or personal details. Ads appear beneath relevant conversations and are clearly separated from the organic answer.
Advertisers can currently reach only users on the platform’s Free and Go plans
Plus, Pro, Business, Enterprise and Education accounts remain ad-free. While there are no official figures for the UK market, only approximately 5% of global ChatGPT users pay for a subscription, so the potential reach should be healthy. However, it may be harder to target medium and large businesses, as they tend to use Enterprise accounts (if ChatGPT is their AI of choice).
The ads themselves share some similarities with Google campaigns
The current ad format includes the advertiser’s name and logo, a headline, a description, an image and a landing-page link.
Conversion optimisation is now available
Advertisers should begin by testing conversion optimisation against mid-funnel actions, where performance drivers such as engagement, clicks, and traffic generation are already well established. Rather than optimising immediately towards bottom-funnel conversions, a mid-funnel approach is likely to provide the platform with a greater volume of conversion signals and a stronger learning environment, helping to build confidence in the algorithm before progressing to lower-volume, high-intent conversion goals.
In terms of data, we can already see some useful, albeit standard, metrics within the tool
Advertisers can measure impressions, clicks, spend, click-through rate, average CPC, average CPM and attributed conversions. Performance can be viewed by campaign, ad group and individual ad, with chart views and CSV exports available.
First-party data is already part of the thinking
Retail advertisers can connect product feeds, while first-party customer or prospect lists can be used to include or exclude audiences and adjust bids, subject to minimum audience thresholds. This is less applicable to categories such as real estate or healthcare, but it could become an interesting ABM-style retargeting strategy for B2B brands with robust prospect databases!
Restricted categories already exist
Real estate is not one of them, but financial, legal and healthcare services are. It’s not a hard no for those types of campaigns, but they require manual approval from OpenAI, which is granted on a case-by-case basis. This means setting up a campaign would take longer.
Brand safety is also being considered
Ads do not appear near sensitive topics such as health, mental health and politics. We have no doubt this list will evolve over time.
ChatGPT ads are not targeted based on keywords
Advertisers can provide context hints describing the types of conversations, subjects and user needs for which an ad may be relevant. These signals help the system understand intent, but they do not allow a brand to purchase a specific prompt or guarantee that an ad will appear in response to a particular question. It is impossible to predict what a user may ask when they can ask anything, so brands need to stay flexible about where they would like to appear.
The system is particularly sensitive to context when deciding which ad is the best fit
It considers the user’s apparent intent alongside the advertiser’s context hints, landing page, headline and ad copy. Advertisers are therefore not simply choosing who to target or what question to bid on. Instead, they are defining the situations in which their product or service could be useful, and they need to match what they offer to that.
That distinction matters. A person asking a detailed question or comparing alternatives may reveal more about their immediate need than a conventional audience segment or isolated search term. Understanding the language, questions and decision points within a category therefore becomes a central part of campaign planning. It is worth spending more time on topic research and conversational intent analysis, as doing so can provide an important advantage.
There is a clear first-mover advantage in trying ChatGPT ads early, before the platform’s conventions, costs and competitive dynamics become established. ChatGPT is undeniably widely used, but its advertising environment is still much less competitive than it is likely to become, even if the current set-up is somewhat rough around the edges.
Our Activation team have already begun examining how conversational context affects campaign structure, delivery and traffic quality, so more insights will follow.
If you are interested in finding out whether your campaigns would fit within the current beta and how they could realistically be run, reach out to us.