ChatGPT ads are live. Your AEO foundation decides if they work.

On February 9, 2026, OpenAI switched on ChatGPT ads. Free and Go users in the US see them first - small, labeled boxes at the bottom of a response, on the surface where hundreds of millions of people now ask what to buy. Paid tiers stay clean for now. The query that used to end on a Google results page now ends inside a conversation, and that conversation has inventory for sale.

For marketers, the reflex is to ask how to buy the new placement. That is the wrong first question. ChatGPT ads reward brands that already show up in the answer organically and punish the ones that do not. Here is what actually shipped, and what it takes to win the surface.

What OpenAI actually shipped

OpenAI is testing ads with US users on the Free and Go tiers. Plus, Pro, Business, Enterprise, and Education subscribers do not see them. Ads sit in clearly labeled boxes below the response, and OpenAI says they do not influence the answer itself - your conversation stays private from advertisers.

Targeting is contextual: the model matches ads to the topic of the conversation rather than to a keyword you bid on. In May 2026 the pilot expanded beyond the US to the UK, Mexico, Brazil, Japan, and South Korea, with Australia, New Zealand, and Canada following. This is not a quiet experiment. It is the start of an ad network on the fastest-growing consumer surface in a decade.

Paid placement on an empty foundation fails fast

The uncomfortable part for performance teams: ChatGPT ads do not behave like search ads. With Google, you can buy your way to the top of a query you have no organic presence on, and the traffic flows the moment your card clears. On a conversational surface, the ad is a layer on top of what the model already believes about you.

Brands that run ChatGPT ads with no organic answer presence are finding the spend has nowhere solid to land. The recommendation, the comparison, the follow-up question - all of that is still driven by what the model retrieves and cites on its own. Pause the campaign and visibility can collapse within days, because there was never an organic floor underneath it.

The foundation is your citation share

Showing up in ChatGPT's answer - paid or not - starts with the same work: being the source the model retrieves, trusts, and cites. We call this citation engineering, and it is the discipline that decides whether your brand is in the answer before anyone buys a single impression.

That means entity authority, clean structured content the model can parse, and a measurable citation share you actually track. Most teams cannot tell you what their citation share is today, let alone defend it. If you are buying ChatGPT ads to paper over that gap, you are renting visibility you could own.

Track the two separately. Blending paid impressions into your organic share of voice gives you a flattering, useless number. Measure organic brand mention rate and citation rate on their own, then layer paid on top and watch what it actually lifts.

This is operating work, not a campaign

Here is why the foundation is hard: it is not a project you finish. The model re-reads the web continuously. Competitors publish, sources change, your citation share drifts, and the answer to "best X for Y" shifts week to week. Defending your spot in ChatGPT's answer is a standing operation across content, on-page, technical, and off-page work, run continuously rather than shipped once.

That is the gap between marketing automation and agentic marketing. Automation fires the rules you wrote in advance. Agentic marketing watches the surface, decides what to change, and ships the work - new pages, structured data, updated comparisons, community presence - on its own. When the surface itself is a moving conversation, only the second model keeps up.

How Oz approaches the new surface

Oz is Claude Code for marketers: an agentic platform where marketing teams build and run AI agents that connect to their full stack - Google Search Console, Google Analytics, brand guidelines, SEO data, 100+ integrations - and take action end to end. Today Oz's agents operate across SEO and AEO: content, on-page, technical, programmatic pages, Reddit and community engagement, off-page, and competitor analysis. That is exactly the work that builds and defends the organic citation foundation ChatGPT ads sit on top of.

Ads, PDPs, and catalog management are on the near-term roadmap, so the same agents that win you organic placement extend into the paid layer instead of living in a separate tool. This is the toolchain Passionfruit refined over years of SEO and AEO delivery - $1B+ in incremental organic revenue shipped for clients like HP, Unilever, and Solawave - now opened up to in-house teams.

What ChatGPT ads are not

ChatGPT ads are not a replacement for organic AI visibility - they amplify it or expose its absence. They are not a keyword auction you can game in isolation, because contextual matching runs on top of what the model already understands about your brand. And winning them is not a content-generation problem: pumping out more articles does not move citation share, while the right structured page, in the right place, that the model actually retrieves does.

Which is why the tools built for the old surface miss. A content generator writes copy. A scorer grades a draft against a keyword. Neither builds entity authority, neither tracks your citation share across models, and neither takes action when the answer shifts. The operating layer that does all three is the one that wins the conversation, paid and organic.

ChatGPT ads do not buy you into the answer. They amplify the answer you already earned, or expose the one you never built.

The window is now

Self-serve access, contextual targeting, and a surface adding users faster than any platform in a decade mean the brands that move first will set the defaults in their category. But the ones that win will not be the brands that simply bought the placement earliest. They will be the ones that built the organic citation foundation underneath it, then bought amplification on top.

The foundation work starts now, and it does not wait for the ad spend. Oz is the agentic operating layer that builds and defends your presence in AI answers across every surface that matters. Join the waitlist and start owning the answer before you ever pay to appear in it.