# Describe domain

> What you are, what you offer and on what terms, written so a machine reads it correctly and a person would recognize it.

Source: https://agentrelations.org/framework/domains/describe

## The outcome and why it matters

Once an agent reaches you, it must work out what you are and what you offer. A person skims and fills gaps with common sense. An agent extracts facts and guesses the rest, and its guesses get repeated. Structured, consistent facts are how you stop being misdescribed.

An analogy

Think of a product label. A good label states what is inside, the size, the price and the rules for use, in the same place every time, so anyone can read it correctly, including someone who has never seen the product.

Led by: Marketing and brand, Engineering and product. See the [responsibility matrix](https://agentrelations.org/framework/personas).

## Capabilities

5 things an organization does in this domain. Standards named here are explained in the Standards Map.

Capability 1

### A single source of truth for facts

One maintained record of what is true about the organization: names, locations, hours, people, products, prices, policies and claims.

What good looks like

Every public channel is produced from that record. A change is made once and appears everywhere. Each fact has an owner and a review date.

Typical standards and techniques

- A fact register or knowledge base owned by a named team
- Controlled vocabularies for your sector
- Wikidata identifiers where they exist

Capability 2

### Structured data on the website

Machine-readable markup that states what a page is about, next to the text a person reads.

What good looks like

Every key page carries correct markup that matches its visible content. Markup is validated on each release and never claims what the page does not show.

Typical standards and techniques

- schema.org vocabulary in JSON-LD (Organization, LocalBusiness, Product, Offer, Event, FAQPage and others)
- Google’s structured data guidelines for eligibility
- Open Graph tags

Capability 3

### Machine-readable offers and catalogs

Products, services, prices, stock and policies in forms that agents and shopping systems can read in bulk.

What good looks like

The catalog feed is complete, refreshed as fast as stock changes and consistent with the site. Return, shipping and warranty terms are marked up, not buried in a PDF.

Typical standards and techniques

- Product feeds (for example Google Merchant Center and similar feed formats)
- schema.org Offer, MerchantReturnPolicy and ShippingDetails
- GS1 identifiers (GTIN)

Capability 4

### Documentation written for agents

For services with an API, descriptions precise enough that an agent can use the service without a human reading the manual.

What good looks like

Every public endpoint has a published description with examples, limits and error meanings. Names and summaries say what an operation does in plain words.

Typical standards and techniques

- OpenAPI descriptions
- JSON Schema
- MCP tool descriptions
- A2A Agent Card skill descriptions

Capability 5

### Terms and policies in plain language

The rules agents and their people need before acting: pricing conditions, eligibility, refunds, privacy and what automated access is allowed.

What good looks like

Policies are short, current, versioned and published at stable addresses. There is a published statement on automated access and agents.

Typical standards and techniques

- Plain-language policy pages with dates and version numbers
- Terms of use that address automated agents
- IETF AIPREF usage preferences (in development)

See the [Standards Map](https://agentrelations.org/standards) for who owns each standard and how mature it is. To learn the skills, see the [Specialist certification](https://agentrelations.org/learn/certification/specialist) (Marketing, Engineering and Commerce).

## KPIs

Defined precisely, so two organizations count the same way.

Fact consistency

The share of facts in your register that appear with the same value on your site, in your structured data and in your top external sources, checked by sample each month.

Markup validity

The share of key pages whose structured data validates without errors and matches visible content.

Feed freshness

The median time between a change in stock or price at the source and the same change in the published feed.

## What each maturity level looks like here

1. 1
Level 0: Closed
Facts live in scattered documents, images and PDFs. Agents contradict each other about you.
1. 2
Level 1: Visible
Facts are on the site as text, but vary from page to page and channel to channel.
1. 3
Level 2: Described
A fact register feeds the site. Key pages carry validated schema.org markup, and offers are in a feed.
1. 4
Level 3: Conversational
Documentation, catalogs and policies are all machine-readable and checked for consistency across channels.
1. 5
Level 4: Callable
Descriptions are produced from the source of truth by automation and tested on every release.
1. 6
Level 5: Trusted
Agents describe you correctly in sampled checks, and a mistaken description is corrected within days.

[See all domains side by side](https://agentrelations.org/framework/maturity) or take the [self-assessment](https://agentrelations.org/framework/maturity#self-assessment).
