Agent-commerce readiness starts with observable merchant surfaces.
The planned retailer view is designed to organize merchant-side evidence from product, policy, request, cart, and checkout surfaces. It does not infer what an external model understood, preferred, recommended, or decided.
What the merchant view would show
- Requested pages and products.
- Identity evidence, where verification is supported.
- Recorded shopping milestones, where connected data supports them.
Product preview: this describes the intended pilot scope, not shipped integrations.
Watch the Observatory tour · Evidence definitions · How a pilot would work
What's at stake for retailers
Merchant logs may contain request-level evidence for automated and AI-referred traffic that standard on-site analytics may not preserve or expose with the same request-level provenance. Cartograph is being built to organize evidence from server, CDN, origin, product, policy, cart, and checkout surfaces. It cannot observe decisions made off property.
Where retail storefronts break
- Variant identifiers or attributes missing across named product surfaces.
- Availability values that materially conflict across a feed, page, and structured data within the observation window.
- Shipping or return policies that the defined scanner or collector cannot fetch or parse.
- A recorded bot-control challenge or block on a request carrying agent-related evidence, without treating that evidence as proof of legitimate AI-agent traffic.
- Analytics views that omit log-origin request evidence for automated and AI-referred traffic.
How the planned retailer view is organized
The same Readable / Consistent / Transactable framework from the homepage, applied to your storefront. This is planned; no merchant is measured or scored today.
- Readable — can the defined scanner or collector fetch and parse your product pages, structured data, sitemaps, and policies? Not whether an external model understood, cited, or recommended them.
- Consistent — do your feed, page content, and structured data materially agree within a defined observation window and comparison policy? Not whether an external agent trusted, preferred, or rejected you.
- Transactable — do merchant-observed events, or a separately authorized synthetic diagnostic, show progression through the exact cart or checkout milestones in scope? Not proof of a complete journey or an order.
Who owns this on a retail team
- Ecommerce leaders — prepare to review the last merchant-observed milestone before cart or checkout.
- Product data and merchandising — prepare to compare named surfaces for missing attributes, ambiguous variants, inaccessible policies, and material conflicts.
- Analytics — prepare to evaluate the five v0.1 evidence labels and the planned request-level Unidentified Traffic metric when complete log coverage is available.