ChatGPT Ads Are Here. Here’s What Marketers Actually Need to Know.

ChatGPT Ads are now available to advertisers in Germany. The interesting part isn’t another place to buy clicks. It’s a potentially different way of identifying intent.

If you use the free version of ChatGPT in Germany, you may have noticed something new underneath an answer recently: an ad. Not an affiliate link hidden inside the response and not a product recommendation the model was paid to make, but a separate, clearly labelled sponsored placement appearing below ChatGPT’s answer.

For marketers, that is worth paying attention to. OpenAI started testing ads in the US in February 2026 and expanded ChatGPT Ads into 31 European markets in August, including Germany. Since August 31, advertisers in those markets have also been able to access the platform directly through OpenAI’s self-service Ads Manager.¹ That makes ChatGPT a paid acquisition channel. A very new one, with plenty we don’t know yet, but a paid acquisition channel nonetheless.

The scale has developed quickly. By the end of August, OpenAI said ChatGPT Ads was already being used by tens of thousands of advertisers and had reached a $1 billion annualized revenue run rate. That does not tell us whether the channel performs well for any particular advertiser, but it does tell us this has moved beyond a tiny advertising experiment remarkably quickly.⁹

The interesting thing, however, isn’t really the ad format. It’s the context in which the ad appears.

So, what are ChatGPT Ads?

The basic concept is relatively straightforward. Someone is having a conversation with ChatGPT. They might be researching software, comparing products, trying to understand a problem or figuring out what to buy. ChatGPT generates its answer as usual, and below that answer an eligible user may see a clearly marked sponsored placement.²

Importantly, the advertiser does not buy their way into ChatGPT’s answer. OpenAI says its advertising systems are separate from the chat model, advertisers cannot influence the answer, and they do not get access to users’ private conversations or personal data.² The current ad itself isn’t especially revolutionary either. It can include the advertiser’s brand, headline, copy, image and landing-page link.³

Ads can currently appear to users on ChatGPT Free and Go, while Plus, Pro, Business, Enterprise and Edu accounts remain ad-free.² That distinction becomes particularly important when thinking about B2B advertising, because “people who use ChatGPT” and “people who can currently see ChatGPT Ads” are not the same audience.

What makes the platform more interesting is how OpenAI determines whether an ad might be relevant in the first place.

This isn’t just Google Search with a chatbot attached

Search advertising has trained marketers to think in keywords. Someone searches for employee engagement software, an advertiser bids to appear against that demand, and the quality of the campaign depends partly on how well we understand and capture the searches that indicate useful commercial intent.

ChatGPT creates a different environment because users don’t necessarily express their needs in neat two-to-five-word queries. They explain situations, add constraints, ask follow-up questions and compare alternatives. Someone might say that they run HR for a German company with 800 employees, that engagement scores have fallen over the last year, that they want better employee surveys, and that they need something their existing HR team can actually manage. That is considerably richer than employee engagement software.

OpenAI’s advertising system can consider the context and intent of the current conversation alongside things such as the landing page, ad copy, targeting selections and advertiser-provided context hints.³ Those context hints allow advertisers to describe what their product does, who it helps and the situations in which it might be useful. They are explicitly not exact-match keywords.⁴

So instead of merely targeting a phrase like employee engagement software, an advertiser might describe a situation such as HR leaders at mid-sized companies looking to improve employee feedback, understand engagement problems or reduce employee turnover. That sounds like a small technical difference, but strategically it could be significant because it moves paid media away from matching a phrase and towards matching a problem in context.

This makes positioning unusually important

That part immediately caught my attention because it connects paid media much more directly to something marketers should have figured out long before opening an ad account: when, exactly, is our product relevant?

If contextual advertising works the way it is intended to, vague positioning becomes harder to hide. An advertiser needs to explain not only what it sells, but who needs it, what is happening when they need it and what problem it actually solves. Traditional advertising platforms have often allowed us to compensate for some of that fuzziness with increasingly sophisticated targeting machinery. Here, the relationship between positioning and media execution becomes unusually visible.

Interestingly, OpenAI’s own advertiser guidance now pushes in much the same direction. Its recommendations for context hints essentially ask advertisers to explain three things: what the product offers, who it helps and when it is useful. Ad groups are also supposed to separate meaningfully different customer needs and use cases. In other words, the platform is effectively asking advertisers to do some positioning work before the targeting system can do its job.⁴

If we cannot clearly describe the customer situation in which our product becomes useful, what exactly are we asking the advertising system to match? That is one reason I find ChatGPT Ads strategically interesting beyond the novelty of putting advertising inside ChatGPT. The mechanics push marketers back towards understanding the actual customer problem.

What does it cost?

Right now, the barrier to experimenting is relatively low. ChatGPT Ads supports campaigns focused on views, clicks and conversions. Depending on the setup, advertisers can be billed on impressions or valid clicks, while conversion campaigns optimise delivery towards a selected conversion event.⁵

For accounts billed in euros, OpenAI currently sets the minimum daily campaign budget at €15.⁶ For CPC campaigns, OpenAI recommends starting with a maximum bid of around $3–$5 per click, although the actual price depends on the auction.³ The Ads Manager also supports conversion measurement, including OpenAI’s Pixel and Conversions API, so advertisers aren’t limited to impressions and clicks.¹

There is one number I would not spend much time looking for yet, though: a reliable benchmark. OpenAI explicitly says ChatGPT Ads is still too early to provide meaningful performance benchmarks across advertisers, industries or campaign types.⁷ So anybody confidently telling you what a “good ChatGPT Ads CPC” or expected conversion rate should be at this stage is getting ahead of the evidence.

There is a significant B2B caveat

The Free and Go limitation is probably the first thing I would discuss with a B2B client before testing this channel. ChatGPT may have enormous overall usage, but that doesn’t mean every ChatGPT user is part of the advertising inventory.

If you sell a consumer product, software for freelancers or something with a broad market, that distinction may turn out to be relatively unimportant. If you are trying to reach senior executives, enterprise buyers or people whose employers provide them with ChatGPT Business, it could matter considerably. Those users may simply sit outside the current ad-supported products.

We don’t yet know enough about the composition of the available advertising audience to know how important this will be for different industries. That is exactly why I would be careful with claims about ChatGPT Ads giving brands access to some enormous new professional audience. The potential reach of ChatGPT and the addressable reach of ChatGPT Ads are two different things.

Who should test it?

I wouldn’t put ChatGPT Ads into every client’s media plan tomorrow, but there are situations where I think it is already worth a controlled experiment. Products with a considered purchase journey are the obvious candidates: B2B software, professional services, complex consumer products and other categories where people naturally research and compare before making a decision.

It becomes particularly interesting when the customer problem can be articulated clearly. If people regularly ask questions such as What should I use for this?, How do I solve this?, What is the difference between these options? or What should I consider before choosing X?, conversational context could contain meaningful commercial intent.

Companies already running Google Search are particularly interesting test cases because there is an existing high-intent channel to compare against. Instead of asking whether ChatGPT Ads “works” in isolation, we can start asking whether the intent captured through a conversation behaves differently from intent captured through a search query.

The experiment I actually want to run

For the right client, I would start small. Not because €500 or €1,000 is magically the correct ChatGPT Ads budget, but because the purpose of the first campaign shouldn’t be to scale. It should be to learn enough to decide whether scaling makes sense.

I would build several ad groups around distinct customer situations rather than simply translating an existing Google Ads keyword structure into ChatGPT. The landing pages would need to reflect those situations, and we would set up proper UTM parameters and conversion tracking before spending anything meaningful. Then I would compare the resulting traffic with an existing high-intent channel such as Google Search.

CTR would be interesting, but it wouldn’t be the result. I would want to know whether the visitors are qualified, what they do after arriving, which conversational contexts produce useful traffic, whether that traffic converts and eventually how the economics compare with search. A cheap click from someone who was never going to become a customer is still a bad click.

The hypothesis worth testing isn’t whether ChatGPT can give us cheaper traffic. It is whether conversational context can identify a different or better kind of intent than a search keyword can. If it can, we are looking at something more interesting than another advertising platform.

And this could become much more interesting

The current sponsored placement may only be the beginning. In September, OpenAI introduced a broader set of AI-powered advertising experiences, including Sponsored Agents, new AI creative tools and integrations with platforms such as HubSpot and Shopify. Sponsored Agents are still in limited alpha, but they are particularly interesting because they let someone move from seeing an ad directly into a conversation with an AI representative of the advertiser.¹⁰

Conceptually, however, it feels much more native to a conversational interface than a conventional sponsored card. Imagine researching a complex software product and moving directly from your research conversation into a specialised agent that can explain implementation, answer product questions, compare plans or help you understand whether the solution fits your particular situation. There are plenty of questions about how useful, trustworthy and commercially effective that experience would be, but at least it begins to look like an advertising mechanic that could only really exist in this environment.

That is probably what I will be watching most closely. Putting an ad underneath an AI answer is new inventory. Turning the advertisement itself into a useful conversation could become a genuinely new format.

The Change Strategies position

ChatGPT Ads belongs in the toolkit, but it does not yet belong in every media plan. The platform is new, the available audience is constrained by subscription type, there are no meaningful industry benchmarks yet, and advertisers still have a lot to learn about what conversational intent actually means commercially.

None of that is a reason to ignore it. It is a reason to test it properly. For suitable clients, we would currently treat ChatGPT Ads as an experimental acquisition channel with small controlled budgets, clearly defined customer situations, proper conversion measurement and a comparison against channels we already understand. We would not create a novelty budget because somebody decided the company “needs to do something with AI,” and we certainly wouldn’t celebrate a cheap CPC before checking whether anybody useful actually came through it.

9. OpenAI. “Expanding access to AI with ChatGPT Ads.” 2026.

10. OpenAI. “Reimagining advertising with AI.” 2026.

We already have plenty of places to buy ads. The interesting development here is that, at meaningful scale, marketers may now be able to buy media against the context of a conversation rather than primarily against a search query, an audience segment or someone’s browsing behaviour.

Whether that produces better marketing remains to be seen. But that is exactly why this is worth testing now.

Sources

1. OpenAI. “ChatGPT Ads expands across Europe.” 2026.

2. OpenAI Help Center. “Ads in ChatGPT.”

3. OpenAI Help Center. “Ads in ChatGPT: The Basics.”

4. OpenAI Help Center. “Write Context Hints for ChatGPT Ads.”

5. OpenAI Help Center. “Conversion-optimized Campaigns.”

6. OpenAI Help Center. “Create Campaigns for ChatGPT Ads.”

7. OpenAI Help Center. “Frequently asked questions.”

8. OpenAI Help Center. “Sponsored Agents in ChatGPT Ads.”