Structured Data for AI Agents

Implement comprehensive Schema.org Product markup with every machine-critical attribute: name, description, SKU, GTIN, brand, colour, material, dimensions, weight, and condition. Machine customers evaluate products by constraint satisfaction against structured attributes - not by reading marketing copy.

Embed Offer schema with real-time pricing data: price, priceCurrency, availability (InStock/OutOfStock/PreOrder), priceValidUntil, and eligibleRegion. AI agents performing zero-click commerce require unambiguous pricing that can be compared programmatically across merchants.

Publish AggregateRating and Review schema with itemReviewed references - agents use review aggregates as trust signals when comparing products. Include reviewCount, ratingValue, and bestRating to enable normalised comparison across different rating scales.

Deploy Organization schema with verifiable identity attributes: legalName, taxID, duns, iso6523Code, and foundingDate. Trust architecture for agent commerce begins with machine-verifiable merchant identity - agents must confirm who they are transacting with before executing delegated authority.

Validate all structured data with Google's Rich Results Test and Schema.org's validator, then test with at least three AI shopping agents (Perplexity Shopping, Google AI Shopping, Amazon Rufus) to verify that agents extract the data you intend. The gap between what you publish and what agents consume is the signal clarity gap.

Implement hasCredential and hasCertification properties on your Organization schema to declare verifiable trust credentials - ISO certifications, industry accreditations, payment processor verifications, and regulatory compliance status. Agents performing agentic commerce use these credentials as pre-transaction trust filters.

Publish machine-readable return and refund policies using Schema.org MerchantReturnPolicy: returnPolicyCategory, merchantReturnDays, returnMethod, and returnFees. Agents operating under delegation design constraints must verify that merchant policies fall within the authority boundaries their human principal has specified.

Encode warranty information using Schema.org WarrantyPromise with durationOfWarranty and warrantyScope - agents comparing products across merchants need structured warranty data to evaluate total cost of ownership, not just purchase price.

Declare your shipping policies using OfferShippingDetails with shippingRate, deliveryTime, and shippingDestination. Structure these as machine-parseable constraints rather than prose descriptions - an agent evaluating whether a merchant meets its principal's delivery requirements needs exact data, not 'fast shipping available'.

Implement trust calibration metadata in your structured data: publish your dispute resolution process, customer service availability, and escalation procedures in machine-readable format. Agents need to assess not just pre-purchase trust, but post-purchase recovery capability.

Publish an OpenAPI 3.1 specification at a well-known endpoint (/.well-known/openapi.json) that describes your commerce API surface. Agents performing B2B agentic commerce discover and consume APIs programmatically - if your API is not self-describing, agents cannot integrate with your systems.

Implement JSON-LD WebAPI schema that declares your API endpoints, authentication requirements, and rate limits. Include potentialAction properties that describe what agents can do: SearchAction for product queries, BuyAction for purchases, TrackAction for order status.

Design your API responses to include structured provenance metadata: data freshness timestamps, source system identifiers, and confidence indicators. Agent legibility requires that agents can assess the reliability of the data they receive, not just the data itself.

Implement content negotiation so agents can request data in their preferred format (JSON-LD, Turtle, N-Triples) via Accept headers. Different agent frameworks consume different serialisation formats - supporting multiple formats maximises your agent-accessible surface.

Publish a machine-readable API documentation manifest that includes authentication flows, error response schemas, pagination patterns, and webhook specifications. Agents that cannot understand your API's error handling will fail silently - structured error documentation prevents cascading failures in autonomous agent workflows.

Publish a comprehensive JSON-LD SiteNavigationElement that declares your site's information architecture - every major section, its purpose, and its relationship to other sections. Agents use navigation schemas to build internal maps of your content, enabling targeted retrieval rather than exhaustive crawling. This is the intelligence layer that transforms a website into an agent-navigable knowledge structure.

Implement ItemList and CollectionPage schemas on index pages that declare all items in a collection with their URLs, positions, and summary descriptions. Agents querying 'what products do you sell in category X' can answer from your structured data without loading every product page.

Deploy BreadcrumbList schema on every page to declare the page's position in your site hierarchy. Agents use breadcrumb data to understand content relationships and resolve entities across pages - a product page's breadcrumb tells an agent which category, brand, and department the product belongs to.

Publish a machine-readable content inventory using Dataset or DataCatalog schema that declares all your structured data endpoints, their update frequencies, and their coverage scope. Agents that know where your data lives and how often it changes can optimise their retrieval strategies.

Implement llms.txt and llms-full.txt files at your site root following the emerging standard for agent-observable content declaration. These files provide AI systems with a curated summary of your site's purpose, key content, and citation preferences - functioning as a structured introduction that agents read before crawling.