Assisted Intent: What ChatGPT's New Ad Platform Could Mean For Sports Brands

Hey {{name}} ,

Welcome to another edition of The Sports Stack. This week, I’m focused on OpenAI’s new ad manager and whether sports organisations should consider it as a paid media channel. 

In my previous edition, I wrote about the continued decline in traffic that organisations are losing to AI apps, and now I’d like to focus on a potential (albeit costly) avenue for regaining that traffic. The twist is that ChatGPT (and others) don’t just know what your fans are searching for; they know what they’re trying to decide, which changes how paid media works within it.

What you'll find in this edition:

By the way, thank you to everyone who read and shared my last piece on the future of sports platforms. It was by far my best-performing article to date, with over 3000 reads across channels and driving a huge number of new subscribers - welcome to anyone new!

OpenAI's ad platform is still in testing, but it's making great progress. Self-serve buying started in May 2026, and five countries are already part of the rollout. The ad unit is also more powerful than early comments indicated. Here's what OpenAI has officially confirmed, straight from their own documentation.

Live in five countries, including the UK, as of this week. Ads are currently active in the US, Canada, Australia, and New Zealand. Fresh off the press, the ChatGPT ads pilot has gone live in the UK as of 5th June, so expect more countries very soon!

Only Free and Go users see ads. Plus, Pro and all Business plans are ad-free by design. Ads are restricted to logged-in adults, with under-18 accounts excluded by self-declaration or prediction. Free users can opt for an ad-free experience in exchange for lower usage limits. We’ll dive into this in a later chapter, as it’s important for your decision-making.

Whole environments are off-limits. No ads appear next to sensitive or regulated topics, e.g. health, mental health, politics. Third-party reporting also indicates that there are no ads in Temporary Chats, immediately after image generation, or in the ChatGPT Atlas browser.

Self-serve buying is live. The Ads Manager beta launched May 5, 2026, with the previous $50,000 minimum spend removed. Advertisers can now register, set budgets, upload creatives and manage campaigns directly as they would in other ad managers, e.g. Meta and Google. 

Two ways to buy: CPM or CPC. Reach objective for CPM (default max bid $60); Clicks objective for CPC (recommended starting max bid $3–5). Based on research from tutorials, anything under $3 is flagged as unlikely to be delivered to users.

The ad unit is richer than a text line but not up to Meta's creative standards. Each placement contains an advertiser name, favicon, headline, body copy, landing page and an image. Ads sit below the ChatGPT response, clearly labelled and visually separated from the organic answer.

Targeting runs on conversational context, not keywords. Advertisers can supply "context hints" (topics, conversations, or keywords) that indicate where their product may be relevant to guide ad matching. These hints aren't exact matches and don't guarantee delivery in specific conversations. I’ll share some examples of context hints as I walk through the step-by-step guide later in the article.

Reporting is functional, not exhaustive. Ads Manager Beta covers impressions, clicks, spend, CTR, average CPC, average CPM and conversions. UTM parameters persist through the click for integration with existing analytics.

OpenAI's 900 million weekly active users is the number you might immediately think about as your TAM (total addressable market), but it’s actually not that high, and for sports marketers, it's the wrong number to plan against. 

Strip the audience down to those who can actually see an ad, and the 900 million active users drop quickly. 

Those 50 million paying subscribers skew tech-fluent, higher-income and business-led, exactly the cohort premium sports brands pay top dollar to reach on Meta and Google. ChatGPT has built a product in which the most commercially valuable users pay specifically to avoid advertising. By design. 

There is still 81% of the user base that you can target via ads, but with a higher CPM than most other channels, it will be interesting to see whether the ROAS is as good as Meta, Google and others. 

In my opinion, this is probably the most defining reason why this channel will not suit all brands. In the future, ChatGPT may relax this decision, but that’s not a given. We’re also seeing other operators, like Meta, releasing ad-free subscriptions, so it could be that this is the new norm.

When thinking about writing this edition, one of the main areas I wanted to dive into was the differences in intent between someone searching on Google, Meta, and ChatGPT.

For two decades, paid media has been built around two intent models. Google captures expressed intent; you type a keyword or a range of keywords, and behind the scenes, an auction runs. Meta captures interrupted attention; you scroll through content, and a targeted ad slides into the feed. Both have worked and produced enormous ad businesses, but neither describes what I believe will happen inside ChatGPT conversations. 

We now search the internet in a different way; we ask questions, provide briefs and no longer just browse different sites that Google provides us - this is what you could call an assisted intent model. When using ChatGPT, Perplexity or Google (The new Google vs the old Google), we now provide it with things in our conversations like: 

To give you a sports example of this approach, a sports fan isn’t typing ‘Manchester United Tickets’ into ChatGPT; they’re now asking for a good way to take their family to a Manchester United game for under £300, including food and activities. Advertisers who appear next to these searches know more about your decision-making and desires than any keyword bids ever have.

I like to call this ‘assisted intent’

ChatGPT is charging a premium because they believe the intent is higher; every impression represents a user who is trying to make a decision or make a purchase, rather than a passing doom-scroll. 

The setup will feel familiar to anyone who has run campaigns on Meta or Google, but the difference lies in how you target and build the creative. 

1. Create your account. Sign in at ads.openai.com with a work email, upload your brand logo, submit your business details and set up billing. There's a verification wait before you can build campaigns. According to research, using a work email will get you verified faster.

2. Build your campaign. Choose an objective, either Reach (CPM) or Clicks (CPC), then set your budget, dates and target countries. Structure each campaign around a distinct goal: one for ticketing, one for merchandise, one for hospitality, etc. Conversions are coming soon as an objective.

3. Set your targeting — this is where it gets different. 

At the ad group level, you need to write context hints. These are natural-language examples (one to two sentences each) of conversations where your product is relevant. No negative keywords and no exact-match toggles. 

The goal here is to describe what your core user writes when you want to appear in the conversation. My advice is to use the insights you might have from Google search terms, virtual assistants and customer support tickets to understand what your customers typically search for, write or query to find you. Use tools like Claude to help you build these as well.

For a ticketing brand, that might look like:

Fan travel & hospitality:

The auction considers four signals: your context hints, landing page, ad title, and ad copy, and combines them to improve relevance relative to your bid. A highly targeted ad with a lower bid can outperform a broader one with a higher bid. Being specific is key.

One ad group per audience-and-intent combination. Use the audience + intent + topic formula: name the person, the active verb (comparing, planning, evaluating), the category, and a constraint.

4. Build your creative.

Each ad includes a title, a description, an image, and a landing page. The specs are deliberately tight:

Build multiple variations per offering. OpenAI's system tests different creative combinations across conversations, so a diversity of angles outperforms a single "perfect" ad repeated.

5. Submit and monitor. All ads go through review and are typically approved within 24 hours. Once live, Ads Manager reports impressions, clicks, spend, CTR, CPC, CPM and conversions. Treat the first 30 days as diagnostic: learn which context hints perform, which creative converts, then shift budget to what's working. 

More information on building your first campaign is below:

Ticketing is the clearest fit. StubHub, SeatGeek and Ticketmaster all launched ChatGPT integrations between December 2025 and April 2026. Ticketmaster has explicitly confirmed it's running sponsored ad placements for queries such as "what events are happening this weekend?" 

Fan travel and hospitality rank just behind, as the complex, multi-constraint queries fans ask ChatGPT are ideal for this channel. DTC sports merchandise with direct checkout completes the list. 

If you are a league or organisation expanding your presence in English-speaking markets, then ChatGPT offers you a targeted route into relevant conversations without Meta’s creative overhead. Fantasy sports platforms (not real-money, as I believe it’s banned) fit nicely here, too. Fans are always asking about lineup decisions, player comparisons and draft strategies. The downside is it’s expensive.

Sportsbooks and gambling operators are entirely locked out. OpenAI’s Ad Policies currently prohibit advertising for online sports betting, casino promotions, and platforms involving real‑money wagering, so sportsbooks cannot run standard ChatGPT ads today. Major European leagues running brand campaigns to grow their global profile, e.g. La Liga and Serie A, won’t find what they need here either, as the ad format cannot carry emotionally-led storytelling and the five-country footprint excludes core audiences. The same applies to all sports organisations promoting their website and app; no purchase funnel, no fit, in my opinion.

It’s important to remember that every ad served on ChatGPT will appear below the answer. All brands appearing within the answer arrived there organically. We’re still early in the AI search engine era, so it’s entirely achievable for sports orgs and brands to focus on strong structured data, merchant feeds and answer engine optimisation to enable them to be served organically. Paid placement should always be a secondary lever. 

Sports organisations that want to rank high should focus on four things: 

As you read through this analysis, I’m sure you’ve been thinking, “Why would I run ads on the most expensive channel, where the most commercially valuable users are excluded by design. Wouldn’t it be better to increase my spend on mature channels like Google or Meta, where my audience is proven and I know what a good ROAS looks like?” 

It’s a fair challenge! And if the question is “should you move all your paid media budget to ChatGPT ads?” - the answer is obviously no, and that’s not the question this piece is asking. 

The question is whether ChatGPT merits consideration, and early evidence indicates it does. According to Criteo’s May 2026 ChatGPT ads update, over 1,000 brands running campaigns via Criteo’s integration with OpenAI are seeing AI‑referred conversion rates approaching 2x traditional search in categories like consumer electronics and lifestyle & wellbeing, with budgets largely additive rather than cannibalising existing channels.

The first-mover advantage in paid media is well-documented. Brands that tested Meta ads early, entered Google Shopping first, and experimented with TikTok before CPMs increased gained an early understanding of the mechanics before competition and costs normalised. ChatGPT's ad platform is only four months old, and CPMs have already dropped from $60 to $25 in observed cases. 

Worth flagging, though: measurement is still maturing. OpenAI launched its Conversions API and pixel-based tracking in May and rolled out conversion-optimised campaigns on June 5, but independent third-party verification doesn't yet exist,  so for now, the platform is marking its own homework. 

I’m not suggesting you pull budget from what's working. But if you're a sports organisation with a transactional funnel, a presence in English-speaking markets, and even a modest test-and-learn budget, the cost of ignoring this entirely is starting to outweigh the cost of running a small experiment. 

Thanks for reading!

Mark