Advertising on ChatGPT: What Every Marketer Needs to Know Right Now

Tarryn Wensley

Digital Manager

Team AI, team anti-AI, or firmly on the fence?

If you’ve spent any time in a marketing team meeting over the past year, you’ll know the room usually splits three ways. There’s the “team AI” camp, already automating their way through tasks throughout the day. There’s the sceptical middle, cautiously testing the water while keeping one eye on the exit. And then there’s the “team anti-AI” contingent, who have valid concerns – the environmental cost of powering these models, the murky data privacy and security questions, and a general unease about where all of this is heading.

Wherever you sit on that spectrum, here’s the thing: love it, loathe it, or feel somewhere uncomfortably in between, AI is fast becoming part of the everyday marketing toolkit. And when it’s used thoughtfully, considering both its capabilities and its limitations, it can be a very useful tool. Nowhere is that truer right now than in advertising, where AI platforms are opening up new channels to reach audiences.

The opportunity: advertising inside AI conversations

For years, a “new channel” in digital marketing meant a new social platform. What’s happening now is different.

People are increasingly using AI tools like ChatGPT to research products and services, compare options and make decisions, all within a single conversation. That matters because it’s where intent is starting to surface. Someone asking an AI assistant to compare two instruments or research tools is often close to deciding.

For science and healthcare agencies, this is an exciting opportunity to provide credible, evidence-based information at exactly the right time.

Exclusive early access to the ChatGPT Ads beta: what we’re seeing so far

OpenAI has started testing ads inside ChatGPT. The mechanics will feel familiar to anyone who’s run paid search or social. A clearly labelled, visually separated sponsored card can appear beneath a response when the conversation shows genuine commercial relevance, matched to the context of that conversation rather than to a static keyword list.

OpenAI has been firm that ads run on entirely separate systems from the AI model that generates organic ChatGPT answers. Advertisers can’t buy their way into ChatGPT’s answers, only into a clearly signposted placement alongside them.

What began as a US-only pilot has expanded steadily into new markets, with advertisers accessing the platform through OpenAI’s self-serve Ads Manager Beta. Sciad has been accepted into that beta, putting us among a select group of agencies with early, hands-on access, and giving us a head start on learning what works for science and healthcare communications.

How it works, and why it isn’t Google Ads with a fresh coat of paint

No keywords. OpenAI handles targeting by matching ads to conversation context, meaning the sentiment, topic and intent of the live conversation, rather than to keywords you bid on.

Minimal format. One unit: a small image and a line of text beneath the AI answer, clearly labelled as sponsored.

Bidding has matured fast. The Ads Manager now supports CPM and CPC bidding, a measurement pixel for on-site conversion tracking, and a Conversions API for server-side attribution.

The audience is structurally limited. Ads only appear for logged-in users on Free and Go plans, so you’re reaching non-paying users only. That skews the audience in ways worth factoring in.

An EU privacy differentiator. Under OpenAI’s EU ads policy, updated on 2 June, personalised ads only serve to users who explicitly opt in, using consent rather than “legitimate interest” as the legal basis. Opted-in users can be targeted using past chats, memory and ad history. Everyone else sees generic contextual ads based only on the current conversation, location and time. In practice, a chunk of your targeting depth depends on opt-in rates.

What’s new across targeting, optimisation and measurement

The platform is evolving quickly and latest updates fall into two camps: reaching the right people, and proving that what you spend, delivers.

On targeting, advertisers now get more control. You can upload audience lists to include or suppress specific audiences and apply bid adjustments at ad-group level to flex spend for segments. Account health recommended tasks and key metrics appear in a single view, with a larger trend chart that makes patterns easier to spot. There’s also a suggested ad drafts feature, where a new ad is pre-filled with existing data from your own website.

The bigger shift is in optimisation and measurement, where the channel is starting to behave like a mature performance platform:

Bidding to conversions, not just clicks. Campaigns can now optimise for a conversion event rather than paying purely for clicks, so budget leans towards people more likely to act after they land. Access is rolling out gradually.

Cleaner tracking. Placeholders for campaign, ad group, ad and account IDs can be dropped into your landing-page links and populated automatically when the ad serves, which keeps attribution tidy. When a tracked event doesn’t come through, Ads Manager now tells you why, which field caused it, and how to fix it.

More measurement options. ChatGPT Ads now plug into some third-party tools, so you can view performance alongside your other channels and send conversion data back to OpenAI to sharpen optimisation.

Advanced matching on by default. This uses securely hashed details from your website forms to match more conversions back to ads. It’s on by default for new pixels and is being extended to existing ones, so if you’d rather not use it, review the opt-out in your conversion settings sooner rather than later.

The format keeps growing too. A more compact ad card is rolling out across web and mobile and a carousel that shows several of an advertiser’s products in one unit is being tested. Availability is also expanding into new markets.

What matters most is where this is heading. The channel is maturing from a simple contextual placement into a measurable performance channel, and the sooner you understand how conversions are tracked and optimised here, the better placed you’ll be as it scales.

Our advice: worth a look, but no need to rush.

With OpenAI projected to generate $2.5B in ad revenue in 2026, brands that invest early in ChatGPT ads could gain a valuable first-mover advantage.
As part of the Ads Manager Beta, we’re exploring how this channel can work for scientific and healthcare organisations, where messaging accuracy, tone and credibility matter just as much as click-through rate.

ChatGPT ads could be a natural fit for research-led sectors like B2B, SaaS and science and tech, because they reach people at the exact moment they’re digging into a topic. That intent is valuable. The catch is that ads only show to users on the free tiers, so the senior decision-makers who tend to pay for Plus or Enterprise are largely out of reach. With targeting and measurement still behind Google and LinkedIn, it’s a channel to test with a small budget rather than a big bet. For now, treat it as a complement to your existing search and social, not a replacement.

While paid AI placements remain limited and the platform continues to mature, optimising content to gain visibility in organic AI responses should still be a priority. Through Answer Engine Optimisation (AEO), building the credibility and authority that gets you cited is a benefit that compounds over time and delivers longer-term value.

Explore ChatGPT Ads for your own scientific communications

Interested in ChatGPT Ads? We can help!

If you’re curious about what a ChatGPT ads campaign could look like for your scientific organisation, or you’d like to explore applying for beta access for your own advertising, we’d love to talk it through. Get in touch with the team and let’s figure out whether this new channel has a place in your marketing mix.