A Place in the Model's Answer Is Now for Sale

Share of Model, a metric I defined in the summer of 2025, just became a line item in other people's commercial decks.

Last summer I described a metric that didn't exist in any report: Share of Model. The share of a brand inside the answer an AI gives when a shopper asks "what would work for me." Back then it was a thinking frame. This summer it is a product category.

Look at the sequence.

Sephora now sells directly inside Google: assessment, questions, routine building, checkout. The shopper never leaves the platform. In parallel, the Sephora app runs inside ChatGPT, linked to loyalty profiles.

Amorepacific opened its own storefront inside ChatGPT: a conversational environment where recommendations are assembled from the company's product data.

Revieve released a tool that lets a brand tune its own voice and brand logic inside AI recommendations. The vendor's argument deserves a second read: if a brand doesn't participate in these environments, the answer about it gets formed without it. The payment isn't called buying placement yet. It's called managing AI presence. The economic function is the same.

Payment protocols from Visa, Mastercard and Google matter in this picture not because they shape the recommendation. They turn the answer into a commercial surface: for the first time, the recommendation, the product and the checkout live in one environment. Once an answer can close a transaction, presence inside it becomes a measurable asset.

Let me be precise, so the thesis can't be picked apart. There is no unified ad auction yet where a brand buys the third line of an answer. The market arrived at this trade from different directions: owned storefronts inside models, managed brand interpretation, payment infrastructure. The mechanics differ. The commercial outcome is one and the same: the right to influence the answer now has a price.

This isn't a new mechanic. It's a very old one in a new environment

Shelf placement has always been for sale. Slotting fees, eye-level placement, the golden shelf: retail has monetized shopper attention for as long as retail has existed. Then placement in search results went on sale: the same shelf, just infinite.

Share of Shelf measured physical visibility. Share of Search measured visibility in results. Share of Model measures the probability that a brand ends up inside an already formed decision.

The difference isn't in the mechanic. It's in perception. The shopper could see the shelf and understood its rules. The shopper could scroll past the results, and the paid ones carried a label that said "ad." An AI recommendation arrives as an answer to a personal question, asked in the shopper's own words, about themselves.

Three rights are being formed right now

Strip the news cycle down to substance and the market is trading three distinct rights.

The right to be present in the answer. Sephora and Amorepacific are claiming it: an owned environment inside the model guarantees your catalog participates in the conversation at all.

The right to shape the brand's interpretation. Revieve sells it: not a slot in the answer, but control over which data and language the answer is assembled from.

The right to run the transaction after the answer. Payment protocols are building it: they connect the recommendation to the register.

For now these rights are sold separately, by different players. The day they merge into one system, Share of Model stops being a metaphor and becomes a line in a media budget.

Three consequences nobody is saying out loud yet

First. The recommendation acquires a cost basis. Once the right to influence an answer carries a price, the recommendation becomes a function not only of your question but of who paid for access to its formation. Search went down this road over a decade, and its final compromise was the word "ad" printed next to the result. There is no such label inside an AI answer. There is, for now, no requirement to add one.

Second. Multi-brand retail has started selling its core currency. Its power was always comparison: shoppers went where brands compete, and that's what separated it from a single-brand store. A retailer that rents its recommendation layer out to brands is converting accumulated trust into one-time revenue, at an exchange rate it will only learn later, when shoppers start asking questions about the coincidences.

Third. A vacancy has opened in the architecture. The more voices in the answer are shaped by commercial interest, the more valuable the layer whose voice cannot be bought. Not for ethical reasons. For market ones: trust in the recommendation becomes a scarce resource, and scarcity appreciates. I call this the Indifference Test: a decision layer passes it if it is indifferent to which option gets bought. Today almost none of the players above pass it, and that is exactly why the position is open.

Who takes it: a platform that decides to play arbiter, a regulator that mandates labeling of paid voices, or an independent layer built around indifference to the outcome. Open question. But the vacancy is visible now, earlier than most of the market has even articulated the problem.

What is actually new here

The shelf has moved into the chat, and the bidding is in full swing. Who's buying isn't the interesting part: the list is predictable. The interesting part is this. For the first time in the history of retail, the price of placement and the price of trust are moving in opposite directions. A place in the answer gets more expensive with every release. Trust in the answer gets cheaper with every bought voice.

Markets don't tolerate divergences like this for long. Positions get built on them.