Shopping Models
The AI shopping surfaces the platform runs — products, prices, sources, and what each surface returns
Shopping analyses run against consumer AI shopping surfaces rather than provider APIs. Instead of a text answer, each surface returns a ranked list of products — with prices, merchants, ratings, and images — which feed the shopping visibility and product-tracking metrics.
- Products & pricing — the surface returns structured products (title, price, merchant, rating, image; with optional brand/SKU when available), not just prose.
- Sources & citations — the surface attributes the web pages behind its answer; these feed shopping citation metrics.
- Search queries — the intermediate queries the surface generated while answering. Only some surfaces expose them.
Available surfaces
What a product carries
Every product parsed from a surface is normalised to the same shape before it lands in your analyses, so a card looks the same whichever surface it came from:
- Identity — title, brand, and retailer SKU when present.
- Pricing — current price, original (struck-through) price, currency, and discount.
- Merchant — the selling merchant's name and URL, plus every other merchant offering the same product.
- Signals — average rating, rating count, in-stock, and free-shipping flags.
- Image — the product image, backfilled from the product page when the surface omits it.
Not every surface fills every field — availability depends on what the surface itself exposes. Fields a surface does not provide are simply left empty.
How surfaces differ
- ChatGPT OpenAI — the richest surface: structured products, the source pages behind them, and the search queries it ran. This is the default surface for a new shopping query group.
- Alexa Shopping Amazon — Amazon's shopping assistant. Returns products with prices and images; source pages appear only when the answer attributes them, and it does not expose search queries.
Which surfaces run on your analyses depends on your subscription and the surfaces enabled for a domain. Manage the selection per query group in the app under Shopping.
The general model catalogue — interface and native API models for standard (non-shopping) analyses — lives on Model Capabilities.