Agentic commerce glossary

Agentic commerce

Commerce in which AI agents and systems help consumers discover, compare, decide and perform commerce actions, potentially including checkout.

AI shopping assistant

An AI experience that helps people research or choose products through natural-language conversation.

AI product recommendation

A product an AI system surfaces as a suitable option in response to a shopper's stated or inferred intent.

Agentic checkout

A checkout journey started or orchestrated within an AI or agent experience, rather than a conventional browse-to-cart flow on the retailer's website.

Merchant of record

The business legally responsible for the sale, including payment, tax, fulfilment and customer obligations, depending on the arrangement.

ACP (Agentic Commerce Protocol)

OpenAI's open commerce protocol, co-developed with Stripe, for connecting merchants and AI commerce experiences, including product discovery and commerce actions.

UCP (Universal Commerce Protocol)

An open standard Google uses for agentic commerce across surfaces including AI Mode and Gemini, covering discovery and checkout, with post-purchase on its roadmap.

MCP (Model Context Protocol)

An open protocol, introduced by Anthropic, for connecting AI applications to external tools, systems and data sources. It is broader than ecommerce and is not a product feed.

Google Shopping Graph

Google's product-information system behind shopping experiences across Google's surfaces.

Conversational attributes

Optional Google Merchant Center attributes that give AI and conversational systems more product nuance, including question-and-answer, related-product and variant information.

Product feed

Structured product information sent to a commerce, search or advertising destination.

Supplemental data source

An additional product-data source or feed used to add or update attributes without replacing the primary source.

Product data enrichment

Adding or improving accurate attributes, context or product facts to raise data quality and usefulness.

Structured data schema

Machine-readable markup on a web page that describes its visible entities and properties. Useful for search understanding and features, but not a guaranteed AI recommendation lever.

AI visibility

A defined measure of whether a brand, product or source appears within a controlled set of AI and search experiences. Always state the platform, prompt set and methodology.

AEO (Answer Engine Optimisation)

A broad industry term for raising the likelihood that a brand or its content is used or surfaced in answer engines.

GEO (Generative Engine Optimisation)

A broad industry term for raising visibility and citation in generative AI systems. It often overlaps substantially with AEO and SEO.

Recommendation rate

The share of a defined prompt or test set in which a retailer or product is actually recommended. The methodology must state what counts as a recommendation.

Citation rate

The share of a defined AI-answer test set in which a source or domain is cited. Citation and product recommendation are different metrics.

Agentic-sourced sale

A sale attributed to a defined agentic or AI source under a stated methodology. Do not use the label without documenting the attribution rule.

1. Start from the question most of these terms come back to: what an AI system can read about each product.
2. Run the free AI visibility audit, which answers it for up to 50 of your product pages.
3. Read why a product is missing from ChatGPT Shopping to see the terms in use.
4. Then read what decides its rank once it is in.

Most of these terms come back to one question: what an AI system can read about each product. The free AI visibility audit answers it for up to 50 of your product pages. Sighted then enriches the catalogue at scale for conversational discovery, the same technology that powers SO-ME, and the two ChatGPT guides show the terms in use: why a product is missing from ChatGPT Shopping and what decides its rank once it is in.