AI discovery stopped being a visibility problem this month and became a revenue problem.
In the space of a few days, OpenAI let businesses claim profiles inside ChatGPT, and the IAB published a framework for deciding who gets credit when AI shapes a purchase. Add the checkout buttons Google and OpenAI have been building into their assistants, and the picture is clear. The place where customers ask what to buy is becoming a place where they can buy it.
Most brands I talk to are still asking one question about AI: are we getting cited? That is the right first question. It is the wrong last one.
The short answer: being cited by AI is only discovery. The brands that win agentic commerce will design the whole path - Discovery, Authority, Conversion and Ownership - so that a recommendation turns into a customer they keep, instead of a sale an assistant made on their behalf.
What actually changed this month
Three things landed close enough together that I stopped treating them as separate stories.
Business profiles came to ChatGPT. On November 10, OpenAI launched Businesses in ChatGPT in beta. Through Business Hub, a company can claim and verify its business, update key details, and add product feeds with current pricing and availability. It gets aggregated insights, including mentions, website clicks and discovery topics, the themes of conversations where its site is cited. Eligible US ecommerce and online service businesses can also launch a Business Agent that answers customer questions inside ChatGPT.
Read that list again. A profile, a product feed, an analytics dashboard and a branded chat. That is not a search result. That is a storefront.
Checkout is moving into the conversation. In September 2025, OpenAI and Stripe launched Instant Checkout in ChatGPT, built on the open Agentic Commerce Protocol. In January 2026, Google introduced the Universal Commerce Protocol, which Google says will soon power checkout on eligible product listings in AI Mode in Search and the Gemini app. One of those is live and narrow; the other is announced. Both point the same way: the path from question to purchase is starting to close without the customer ever loading your website.
Measurement admitted it has a gap. On November 12, the IAB released its AI-Era Attribution Blueprint. It asks who or what created the intent, made the decision and completed the purchase, then sorts each conversion into Human-Owned, AI-Assisted or AI-Executed. The IAB's VP of AI, Caroline Giegerich, noted that AI-influenced sales are sometimes reported as search or direct traffic, which can send the next marketing investment to the wrong channel.
That last point should worry every CMO. If your dashboard is calling AI influence "direct," you are probably underinvesting in the channel that is doing the work.
Why "get cited" is the wrong finish line
Because a citation is an impression with better manners.
Search trained us to think in a straight line: rank, get the click, convert on the site. AI discovery breaks that line in two places. The assistant can answer the question well enough that nobody clicks. And increasingly it can complete the purchase itself, so even a sale can happen somewhere you do not control.
A brand that optimizes only for citations will show up more often and still lose. It gets recommended, the assistant handles the comparison, the platform handles checkout, and the brand ends up as fulfillment for someone else's customer relationship.
That is why I treat GEO as the way in, not the whole job. I have spent 16+ years on the operating side of brand and marketing, mostly in categories where the usual playbook breaks. The lesson that keeps repeating is that getting attention was never the hard part. Keeping the customer was.
Being merchant of record means you process the order. Owning the relationship means they come back without asking an assistant first.
What the full path looks like
This is the framework I use with brands, and the one the rest of this series builds on.
| Stage | The question it answers | What it covers |
|---|---|---|
| Discovery | Does AI find and recommend you? | GEO, AI visibility, conversational search, category discoverability |
| Authority | Why should AI trust you? | Earned media, answer content, entity signals, citations, schema, reviews |
| Conversion | What happens when interest arrives? | Landing pages, CRO, ecommerce, local retail, booking, app acquisition |
| Ownership | Do you keep the relationship? | First-party data, CRM, email and SMS, loyalty, lifecycle, retention |
Discovery: claim the profile, then audit the answer
A claimed ChatGPT profile with an accurate product feed is becoming table stakes, the way a verified Google Business Profile did for local retail. Do it. Then go further and read what the assistants actually say about your category, your competitors and you. Business Hub's discovery topics will show you where your site is cited. They will not show you the conversations where you were never mentioned.
Authority: give the model a reason to pick you
Assistants recommend what they can verify. That means earned media, reviews, consistent facts across every source a model reads, and content with something original in it. If your website says the same thing as every competitor's website, there is nothing for the model to prefer. Authority is the stage most brands skip because it is slow, and it is the stage that decides whether discovery turns into a recommendation.
Conversion: design for a customer who arrives half-sold
When someone does click through from an assistant, they have usually already compared options. They do not need your homepage pitch. They need the specific answer to the question that brought them, a clear next step, and nothing in the way. And if the purchase happens inside the assistant, conversion design now includes your product feed, your Business Agent's answers and your checkout integration. Those are brand touchpoints. Most brands have not assigned anyone to own them.
Ownership: merchant of record is not the same as owning the customer
Both checkout standards keep the retailer in charge of the transaction. Stripe's description of Instant Checkout has merchants accepting or declining the order, charging the payment method, remitting sales tax, and handling fulfillment and returns as they normally would. Google's Sundar Pichai put it more directly at the National Retail Federation, saying the retailer remains the merchant of record, "allowing them to own and shape the customer relationship."
I would take that as an opportunity, not a guarantee. Being merchant of record means you process the order and handle the return. Owning the relationship means the customer opts into your email or SMS, joins your loyalty program, downloads your app, and buys from you next time without asking an assistant first. Nobody hands you that second part. You design for it.
I have built loyalty from zero, and the principle carries straight over. The first transaction is where you earn permission for the second one. In agentic commerce the first transaction may happen somewhere else, so everything after it has to work harder: the packaging, the confirmation email, the reason to come back direct.
How to measure this
Start with the IAB's three pathways, even roughly. The blueprint asks brands to estimate the share of conversions that are Human-Owned, AI-Assisted or AI-Executed. You will not get it exactly right, and the IAB is candid that much of the evidence sits inside AI platforms. But a rough split beats letting AI-influenced sales hide inside "direct."
Then measure each stage separately:
- Discovery: how often you are mentioned and recommended for the questions that matter in your category.
- Authority: whether what the assistants say about you is accurate, and which sources they draw on.
- Conversion: what happens to AI-referred visitors and AI-completed orders.
- Ownership: how many of those customers you can reach again directly, and how many come back without an assistant in the middle.
Ownership is the one most teams are not tracking at all. It is also the one that tells you whether AI is building your business or renting it back to you.
Is this only for big ecommerce brands?
No, but the timing differs. Business Agents start with eligible US ecommerce and online service businesses, and OpenAI says it will expand over time. Regulated and restricted categories should expect to be later and more constrained, which is exactly why the Authority and Ownership work matters more for them. When a platform limits what you can do inside it, the customer relationship you hold outside it is the asset that survives.
The takeaway
AI discovery is becoming a commercial channel with its own storefronts, its own checkout and now its own attribution model. Treating it like search means optimizing for a citation and handing everything after it to the platform. Design the whole path instead: get found, earn the recommendation, convert the customer who arrives half-sold, and make sure the relationship ends up with you.
The rest of this series takes each stage in turn, starting with authority. The offer behind it is the AI marketing systems service.