How does agentic commerce for retailers decide who gets the sale?
A multi-brand retailer sells products other shops sell too, so an agent compares the offer rather than the product: price, stock, delivery time, returns and service. OpenAI says ChatGPT ranks merchants selling the same item on availability, price and quality. A retailer wins when each offer is matched by GTIN and its terms are declared on every page.
What the customer sees
Sunglasses
Half-frame sunglasses, gradient lenses
CHF 210.00
What their agent sees
"@type": "Product",
"name": "Half-frame sunglasses, gradient lenses",
"category": "Sunglasses",
"availableAtOrFrom": "Store, Geneva",
"additionalProperty": { "name": "Lens category",
"value": "3" },
"offers": {
"price": 210.00,
"priceCurrency": "CHF",
"availability": "InStock",
"deliveryTime": { "maxValue": 1 }
}Price, stock and delivery can be compared
Who this page is for
- Retailers carrying many brands, online or online and in store
- E-commerce teams managing large catalogues and frequent price changes
- Retailers with store stock that agents could reserve or collect
Usual starting point: Buyer Journey Check
What does a retail customer ask an assistant?
A typical request reads like this one: "Where can I buy these trail running shoes in size 43 today in Geneva, or with delivery by tomorrow?" The agent needs to match the exact model, then compare stock in size 43 per store, delivery times and price across the shops that carry it.
Where can I buy these trail running shoes in size 43 today in Geneva, or with delivery by tomorrow?
Which product fields decide the purchase in multi-brand retailers?
In multi-brand retailers, five fields decide whether an agent can shortlist a product: identifiers, store stock, delivery and returns, price accuracy, catalogue scale. The table shows why each one matters to the shopper's request here and, for your developers, how it is declared so an agent can read it.
| Field | Why the agent needs it | How it is declared |
|---|---|---|
| Identifiers | Matching the same product across shops | gtin13, mpn, brand |
| Store stock | Same-day and click-and-collect prompts | availableAtOrFrom with store locations |
| Delivery and returns | Where retailers actually differ from each other | shippingDetails, MerchantReturnPolicy |
| Price accuracy | Frequent changes create gaps between page and markup | Markup rendered from the live price |
| Catalogue scale | Fixes have to apply by template, not page by page | Template-level JSON-LD generation |
What does the data say about AI shopping for retailers?
In Switzerland the comparison already runs against marketplaces: 13 of the 30 largest B2C online shops are marketplaces, according to HANDELSVERBAND.swiss, and Galaxus alone reported CHF 3.764 billion in platform sales for 2025. In the US, AI-referred visits converted 42% better than other traffic in March 2026, which makes the comparison worth winning.
| What was measured | Source, date and scope |
|---|---|
| 13 of the 30 largest Swiss B2C online shops are marketplaces. | HANDELSVERBAND.swiss, citing a Carpathia analysis, 12 March 2026. Switzerland, top 30 B2C online shops |
| Galaxus reported CHF 3.764 billion in platform sales for 2025, up 17%, with 10 million products on sale. | Digitec Galaxus AG, 7 January 2026. Switzerland and EU, company figures |
| ChatGPT reads structured metadata such as price and product description, and ranks merchants selling the same item on availability, price, quality and whether they are the maker or primary seller. | OpenAI Help Center, Shopping with ChatGPT Search, fetched 16 October 2026. ChatGPT shopping results |
| In March 2026, visits referred by AI tools to US retail sites converted 42% better than other traffic, a year after converting 38% worse. | Adobe Analytics, reported by TechCrunch, 16 April 2026. United States retail |
Where does an agent purchase break for multi-brand retailers?
We audit multi-brand retailers stage by stage, because each stage fails for a different reason. The rows below show the failure we look for first at each stage and what changes once it is fixed, in the order a shopper's request meets them, from the first question to the reorder.
-
Discover
Where it breaks
Category pages are thin lists of products with no answers
What changes once it is fixed
Buying guides per category that assistants can cite
-
Understand
Where it breaks
Identifiers are missing for part of the catalogue
What changes once it is fixed
Every product matched to the right model, with missing identifiers measured and filled by template
-
Choose
Where it breaks
Delivery and return terms differ from what the markup says
What changes once it is fixed
The agent reads the same delivery and return terms your customers see, declared once and applied to every offer
-
Buy
Where it breaks
Store stock is visible to people but not to agents
What changes once it is fixed
The agent can check store stock and reserve for collection, through tools exposed over MCP
-
Retain
Where it breaks
Agent orders land in analytics as direct traffic
What changes once it is fixed
You see which orders came from agents, attributed in GA4 through agentic checkout order events
Where should retailers start?
We usually start this kind of engagement with the Buyer Journey Check. Comparison missions show which competitor wins your categories and on which term. From there, the order of the remaining work follows the stage table above, so each step fixes the next place where agents drop your products.
Comparison missions show which competitor wins your categories and on which term.
What should retailers read next?
These guides and worked examples take the questions on this page further for multi-brand retailers. Each link names the topic it covers and says in one line why it matters for a shop like yours, so you can open the one that answers your next question rather than reading all of them in order.
- How AI shopping agents compare products, traced on a gravel bike
A worked comparison between shops selling similar bikes, and the fields that decided the shortlist.
- Do you need a GTIN for AI shopping agents
How agents match the same product across retailers, and what happens to offers without an identifier.
- Agentic commerce in Switzerland: TWINT, CHF and VAT
The Swiss market context for retailers: marketplaces, TWINT, CHF prices and VAT as an agent reads them.
- What is agentic commerce
The definition and examples to share with colleagues who are new to agent-led buying.
What do retailers ask about agents?
These are the questions teams in multi-brand retailers ask us most often about AI shopping agents, answered in full. Each answer stands on its own, so you can forward one to a colleague, and an assistant asked the same question can quote it as it is.
How can a Swiss shop compete with Galaxus when agents compare prices?
On the terms an agent can verify where a specialist often beats a marketplace: stock in the size or colour asked for, delivery time to the postcode, store pickup, return terms and advice. Declare each of them on every offer, matched to the product by GTIN. Galaxus lists 10 million products, so a smaller shop rarely wins on range and wins instead on a complete, accurate offer.
Should I sell on Galaxus or invest in my own shop for AI agents?
Often both, for different jobs. A marketplace listing puts your offers where many comparisons already start, while your own shop is where you control the data an agent reads, the return terms and the customer relationship. Agents compare offers by GTIN across both, so the data has to be complete and identical wherever the product appears, or one listing undercuts the other.
How do AI agents choose between retailers selling the same product?
They match the product by identifier, then compare the offers on price, availability, delivery time, return terms and ratings. A retailer wins by declaring all five accurately, since an offer with a missing field is often dropped before the comparison starts.
Can AI agents use in-store stock?
They can when store availability is declared per location or exposed through a tool the agent can call. That opens prompts such as "available today in Geneva", which a shop with online stock only cannot answer.
How do you fix structured data across a large catalogue?
By template rather than by page. We measure coverage across the catalogue, fix the templates that generate the markup, and validate after each catalogue change so new products ship with complete data.
Get your retail catalogue ready for agents
Bring one product page from your catalogue to the first conversation, and we will show you which of the fields above an agent can read today and which one it is missing.