Ads Are Now Live in ChatGPT: A Breakdown

OpenAI Ads Manager Is Live: A Marketer's Walkthrough of Ads in ChatGPT

OpenAI just made it official. Ads are coming to ChatGPT, and there's now a self-serve platform to buy them.

The landing page went live at ads.openai.com. The Ads Manager is in beta. The help center has quickstart guides. And OpenAI shipped a walkthrough video showing exactly how to launch your first campaign.

For anyone working in marketing, this is the moment AI search stopped being a "future channel" debate. ChatGPT now has an ad inventory, a bidding system, and performance metrics. It's a paid channel.

Below is a walkthrough of the platform, the parts that matter for marketers, and the question OpenAI's launch quietly raises about organic AI visibility.

What OpenAI announced

Three things shipped together:

The Ads Manager Beta, a self-serve platform where advertisers create accounts, build campaigns, and monitor results. It lives at ads.openai.com and looks structurally familiar to anyone who has used Google Ads or Meta Ads Manager.

A set of help center articles covering the basics of ads in ChatGPT, an overview of the Ads Manager, and a quickstart guide for launching a first campaign.

A walkthrough video from Katie on the OpenAI ads team, showing the full flow from account creation to performance monitoring.

Nothing in the announcement says "Google should be worried." But the structure of what was shipped tells you OpenAI is serious about building a real ads business, not a one-off experiment.

The setup, step by step

The launch flow has three steps: create an account, build campaigns, then launch and monitor performance. Here's what each one looks like in practice.

Step 1: Account setup

You can log in with a work email already tied to an OpenAI account, or sign up fresh. Once inside, there's a standard onboarding flow: business information, account verification, and a few configuration steps before you can run anything.

Three settings matter most before you create your first campaign:

Advertiser identity. Under Settings → General → Account Info, you set your advertiser name and favicon or brand logo. This is what shows up next to your ads in ChatGPT, so it needs to match exactly how you want users to see you. There's no "test it later" option here. This is the brand surface.

Billing profile. Under Billing → Overview, you add a card, billing address, and an invoice email. Standard stuff, but worth noting: OpenAI is collecting card data and sending invoices, which means this is a fully transactional ads platform from day one.

Team access. Under Settings → Users, you can invite teammates with admin, member, or view-only permissions. The role structure is what you'd expect from a managed ads platform.

Step 2: Building campaigns

There are two ways to create campaigns: a guided UI flow, or a bulk template you can download, fill out, and re-upload. The bulk option matters more than it sounds. It tells you OpenAI is building for advertisers who run dozens or hundreds of campaigns at once, not just first-time experimenters.

The hierarchy is the same as every major ad platform: Campaign → Ad Group → Ad.

At the campaign level, you set:

• A campaign name
• An objective: clicks (optimize for CTR) or reach (optimize for impressions)
• A campaign budget
• Optional conversion events
• A start date and an optional end date

At the ad group level, you set:

• An ad group name
• A CPC bid, if you chose clicks as your objective
• A default destination URL
• Context hints, the most novel part of the system

Context hints let you describe the conversations or topics that are relevant to your ads. This is the ChatGPT-native targeting layer. Instead of keywords, you're describing the kinds of conversations where your ad should be eligible to appear. Whether this maps to actual user prompts, embedding similarity, or some hybrid system isn't documented yet. But it's the mechanism that decides when your brand shows up in a chat.

For anyone who has spent years optimizing for search intent, this is a meaningful shift. You're no longer matching against a query. You're matching against a conversation.

At the ad level, you configure:

• An ad name
• Headline and ad copy
• An image
• A live preview of what the ad will look like inside ChatGPT

The preview is on the right side of the UI as you build, so you can see your ad in context before you ship it.

Step 3: Launch and monitor

Once campaigns are live, the platform shows a status indicator at every level: campaign, ad group, ad. If something isn't serving, you get a descriptive message explaining why. Most issues are fixable with edits. Some ads can be rejected for policy reasons.

Performance metrics currently available:

• Impressions
• Clicks
• Spend
• CTR
• Average CPM
• Average CPC

You can export performance data as a CSV, choosing between cumulative or daily values. There are also insight charts where you select a metric and see the trend over time.

For edits, there are two paths: an inline edit flow (hover, three dots, edit campaign), or a bulk export-edit-reupload flow that mirrors the bulk creation workflow. The bulk edit option works at the campaign, ad group, and ad level.

Four things to notice about how this is built

OpenAI is building an ads business, not a feature. The structure (campaigns, ad groups, ads, bid types, bulk workflows, conversion tracking, role-based access) is the structure of a platform that expects to scale. This isn't a sponsored-link experiment. It's an ads manager.

The targeting model is conversation-native. Context hints replace keyword targeting. That's a fundamental change in how brands buy attention. You're not bidding on what someone types into a search box. You're bidding on the kind of conversation they're having.

Measurement starts simple, but the framework is there. Today you get impressions, clicks, CTR, CPM, CPC, and conversion events. That's table stakes. But the platform is structured to add more (attribution, audiences, retargeting, lookalikes) over time. Plan for it.

Beta means the rules will change. OpenAI says updates are frequent. Anyone advertising now is essentially helping shape the platform. Early movers will have data, learnings, and account history that latecomers won't.

Ads only solve half the problem

Ads in ChatGPT buy you the ad slot. They don't buy you the recommendation. And the recommendation is where most of the value sits.

When a user asks ChatGPT "what's the best CRM for a 20-person sales team," the response will increasingly include both a recommendation and, eventually, an ad. The ad slot is now buyable. The recommendation isn't for sale.

That recommendation is what we'd call your organic AI visibility. It's how often ChatGPT, Gemini, Perplexity, and Claude mention your brand without being paid to. And it's still the most valuable real estate in AI search, because users trust recommendations more than ads, and because the recommendation is what gets cited, summarized, and remembered.

Paid ads are a new lever. Organic mentions are still the channel.

Most marketers are going to spend the next 12 months learning Ads Manager, optimizing CPCs, and writing context hints. That's good work. But the brands that will win are the ones doing both: buying the slots when it makes sense, and systematically improving the chance their brand gets recommended in the first place.

What to do this quarter

A short list of things worth doing now if you run marketing:

Get into the beta. Even if you're not ready to spend, set up the account. Configure billing. Invite your team. Familiarity with the platform is going to matter, and the early data you gather will be useful.

Test small campaigns against your highest-intent topics. Don't try to compete on broad terms. Use context hints to target specific conversations where your category gets discussed. Measure CTR and conversion the way you would on any other channel.

Audit your organic AI visibility in parallel. Before you spend on ads, know where you actually stand. Which prompts already mention you? Which mention competitors instead? What sources are AI models pulling from when they recommend a brand in your space? If you can't answer those questions, paid ads are going to be expensive guesses.

Treat AI search as a real channel internally. Get it on the dashboard. Get a budget line. Get someone responsible for it. The brands that institutionalize it now will be the ones with usable data when the channel matures.

The category just got real

OpenAI charging for placement in ChatGPT means AI search has graduated from emerging trend to monetized channel. Every brand now has to think about it the same way they think about Google Ads, Meta, or LinkedIn.

But unlike those platforms, the most valuable placement here isn't the ad slot. It's the recommendation. And the recommendation isn't for sale.

That's the part traditional SEO tools won't help you with. Ahrefs doesn't track ChatGPT. SEMrush doesn't tell you which prompts trigger your brand. The tools built for the search era are looking at the wrong surface.

If you're not already measuring how your brand shows up in AI responses, this announcement is your signal to start. The ad slot is new. The recommendation slot has been live for over a year, and your competitors might already own it.

GetMentioned tracks how your brand appears across ChatGPT, Gemini, Perplexity, and other AI models: the prompts that mention you, the sources that influence those mentions, and how you compare to competitors. If you want to see what your AI visibility actually looks like, that's where to start.