How ChatGPT decides which ad to show, and what changed in October

On 21 October 2026, OpenAI published a detailed explanation of how ChatGPT chooses which ad to show. It sounds like a document for advertising engineers. For a business deciding where to spend its customer-acquisition budget, it describes something more basic: a different way of matching an ad to a customer.

From search keywords to customer conversations

Someone searching Google for "emergency plumber Surrey" has a clear, immediate commercial intent. Now picture someone using ChatGPT to talk through recurring water pressure problems: asking about possible causes, comparing repair options, and eventually looking for a local professional. The commercial opportunity develops across the conversation.

OpenAI says that is exactly the problem it had to solve. In a conventional ad system, "a user searches for or views something online" and the system matches ads to that one event. In ChatGPT, OpenAI writes, "user intent develops across conversations and shifts over multiple turns of a conversation, sometimes without a fitting advertising moment." Treating every message as a separate search, it says, would risk misreading the goal, starting too soon and showing ads that get in the way.

How an ad gets chosen

OpenAI's system is run by a purpose-trained language model it calls the orchestrator, which works through the ad process step by step. In simplified form:

Two details matter for advertisers. First, in OpenAI's words, "returning no ad is an ordinary, successful outcome": the system is designed to show nothing rather than something weak. Second, an ad can win the auction and still not appear, because ChatGPT does not delay its answer to wait for an ad.

What changed in October

These changes show a growing advertising ecosystem. They do not establish that ChatGPT Ads will outperform Google Ads for every business.

What this means for advertisers

1. Describe the customer's problem, not just the service

The orchestrator summarises a conversation into intent before it looks for ads. A business whose campaign and landing page clearly say which problems it solves, for whom and in which situations gives the system more to match against. That is also the logic behind context hints, which we explain in ChatGPT Ads context hints.

2. Make the website trustworthy and understandable

Pricing where practical, the areas you serve, a clear description of the business and consistent service details help a customer decide, and give the ad and its landing page a coherent story.

3. Measure beyond the click

The October measurement additions exist because clicks are not the goal. Advertisers need to know whether traffic turns into qualified enquiries, customers and revenue. A low cost per click is not a success if those visitors never convert. Our guide to ChatGPT Ads conversion tracking covers the setup.

4. Test rather than assume

OpenAI shared early results from its measurement partners: according to DV Rockerbox, WeightWatchers' attributed cost per acquisition on ChatGPT Ads was 15.3% lower than its blended paid-search benchmark. These are partner-reported figures for large brands, not independent results or a forecast for a local business. Compare lead quality, acquisition cost and incremental revenue in your own account before moving significant budget. Our ChatGPT Ads versus Google Ads guide covers how to split the test.

Paid placement is not an AI recommendation

One point deserves to be stated plainly. OpenAI says ad selection "runs alongside ChatGPT's normal response generation to ensure the ads flow doesn't influence ChatGPT's responses," using "different services that use separate models and separate code." Advertisers cannot buy a recommendation inside ChatGPT's answer, and a sponsored placement is advertising, not an endorsement.

For Canadian businesses

• Canada has self-serve access, so a Canadian business can run its own Ads Manager account, billed in Canadian dollars, with a minimum daily budget of $25 CAD.
• The visual ad format is a US test for now, with an initial group of advertisers.
• Ads appear only to Free and Go users; Plus, Pro and Enterprise stay ad-free, which shapes who you reach.
• Some categories, such as health, financial and legal services, face extra restrictions outside the US. We keep the details current in our ChatGPT Ads in Canada guide.

The business decision

Conversational advertising adds another customer-acquisition channel. The fundamentals have not changed: a clear offer, a working website, reliable lead handling and an honest understanding of how much you can profitably spend to win a customer. Our lead calculator is a quick way to put a number on that last one.

New technology does not replace those fundamentals. It makes measuring them more important. At MOST AI Labs we judge advertising against customer economics, conversion and revenue, not clicks and impressions. If you want to test the channel properly, see how we run ChatGPT Ads management for Canadian businesses.

Sources

• OpenAI: How we built ChatGPT's novel ad stack for user trust and relevance (21 October 2026)
• OpenAI: New ChatGPT advertising formats and measurement (5 October 2026)
• OpenAI: ChatGPT Ads expands to 26 new countries as over 100,000 advertisers have used the platform (21 October 2026)
• OpenAI Help Center: Ads Manager availability

Editorial note: product capabilities and availability change often. Last checked against OpenAI's documentation on 24 October 2026; verify current access and campaign settings before making spending decisions.