Amazon AI & AIO Glossary: Key Terms for Sellers

A+ Content

A+ Content is enhanced listing content available to brand-registered Amazon sellers, displayed below the standard listing fields. It supports rich text, comparison charts, and branded image modules. A+ Content is separate from the title, bullet points, and description that Alexa for Shopping (formerly Rufus) primarily reads when evaluating a listing for recommendation.

Agentic commerce

Agentic commerce describes shopping mediated by AI agents that discover, compare, evaluate, and sometimes initiate purchases on a shopper's behalf, rather than the shopper browsing and deciding entirely on their own. Alexa for Shopping on Amazon is a current live expression of agentic commerce in the e-commerce context. See the guide to agentic commerce on Amazon for a full explanation of what this shift means for FBA sellers.

AI-Native Performance Score (AIO Score)

The AI-Native Performance Score is Keoxs's 0–100 metric measuring how well an Amazon listing aligns with the intent-matching dimensions described in the published COSMO and SPN research frameworks. It is a Keoxs-developed methodology, not an official Amazon metric; Amazon has not published a seller-facing score for Alexa for Shopping eligibility. The score is available free at app.keoxs.com by entering any ASIN.

AIO (AI-Native Optimization)

AIO stands for AI-Native Optimization: the practice of optimizing an Amazon listing so AI shopping assistants — particularly Alexa for Shopping (formerly Rufus) — can read, understand, and recommend it to shoppers. AIO is distinct from traditional Amazon SEO, which focuses on keyword indexing and search ranking. A listing can rank on page one for its main keyword while still being invisible to Alexa for Shopping if it lacks the intent signals and entity clarity the AI looks for. See the full Alexa for Shopping listing optimization guide for the practical framework.

Alexa for Shopping (formerly Rufus)

Alexa for Shopping is Amazon's AI shopping assistant, rolled into the search experience in 2026. Instead of returning a long page of results, it reads listings and recommends a short set of products — typically fewer than ten — in response to a shopper's natural-language question or voice query. It was formerly named Rufus before Amazon's May 2026 rebrand. Alexa for Shopping is a confirmed, live Amazon feature; its exact internal recommendation mechanics are not publicly documented. See what Alexa for Shopping is and how it finds products.

ASIN

An ASIN (Amazon Standard Identification Number) is the unique 10-character alphanumeric identifier assigned to each product listing on Amazon.com. Every optimization task — audits, backend attribute lookups, AI score checks, and competitor analysis — starts with the ASIN. ASINs are assigned by Amazon and cannot be chosen by the seller.

Backend attributes / search terms

Backend attributes are hidden listing fields on Amazon — such as generic search terms, product attribute fields (material, target audience, color, size), and item type keywords — that are not displayed to shoppers but are used by Amazon to index and understand a product. Filling these fields correctly affects whether your product enters the candidate pool before Alexa for Shopping can even evaluate it for recommendation. See Amazon backend attributes for AI search for a full guide to which fields matter most.

BSA (Business Solutions Agreement)

The Business Solutions Agreement (BSA) is the contract that governs the relationship between Amazon and sellers, developers, and third-party solution providers. As of 2026, the BSA is reported to include an Agent Policy that governs how automated tools and AI agents may interact with Amazon's systems — requiring automated tools to identify themselves as such and prohibiting unauthorized data extraction, among other reported requirements. The SP-API is the BSA-compliant alternative to web scraping. See the Amazon BSA Agent Policy guide for what this means in practice.

COSMO

COSMO is a knowledge-graph framework described in Amazon's published research (SIGMOD 2024) for mapping products to buyer intents, use-cases, and contextual needs through a structured graph of typed relationships. The research describes 15 relationship types organized around four root concepts: usedFor, capableOf, isA, and cause. COSMO is published research from Amazon's engineering team, not a publicly confirmed description of Amazon's live ranking or recommendation system. Keoxs's COSMO Map tool and intent analysis are built on this research. See what Amazon COSMO is for a plain-English explanation.

Entity clarity

Entity clarity describes how precisely a listing states the attributes, use-cases, and audience that an AI shopping assistant needs to understand and recommend the product. A listing with high entity clarity allows the AI to confidently extract: what the product is, who it is for, in what context it is used, and what problem it solves — without ambiguity or guesswork. Low entity clarity is a common reason a listing ranks well on keywords but fails to appear in AI recommendations.

GEO (Generative Engine Optimization)

GEO stands for Generative Engine Optimization: the practice of optimizing content so generative AI assistants — including ChatGPT, Gemini, Perplexity, and Amazon's Alexa for Shopping — can find, understand, and cite a product or brand when shoppers ask them for recommendations. For Amazon FBA sellers, GEO and AIO overlap significantly: clear, machine-readable listing content that satisfies Alexa for Shopping also makes a product more citable by external AI assistants. See the GEO for Amazon sellers guide for how the two connect.

Intent coverage

Intent coverage is the degree to which an Amazon listing addresses the full range of buyer intents — use-cases, audiences, occasions, and contexts — relevant to its product category. A listing with high intent coverage is matched to more shopper queries by the AI; a listing with low intent coverage may rank on primary keywords but miss the long tail of AI recommendation queries that competitors with broader coverage capture. Improving intent coverage is the core goal of Amazon listing optimization for Alexa for Shopping.

Knowledge graph

A knowledge graph is a structured map of entities (products, attributes, use-cases, audiences) and the typed relationships between them. The COSMO research (SIGMOD 2024) describes a shopping knowledge graph that Amazon's engineering team built to bridge the semantic gap between a shopper's natural-language query and relevant product listings — connecting, for example, "for outdoor camping" to specific product types and attributes rather than relying solely on keyword overlap.

Rufus

Rufus was the original name (2024–2026) of Amazon's AI shopping assistant; Amazon rebranded it as Alexa for Shopping in May 2026. References to "Rufus" in older articles, tools, and documentation refer to the same underlying feature now called Alexa for Shopping (formerly Rufus). Current content should use "Alexa for Shopping" rather than "Rufus" to refer to this feature.

Semantic intent / buyer intent

Semantic intent — also called buyer intent — is the underlying need or purpose behind a shopper's query, beyond the literal keywords typed. Examples include "for camping" (use-case intent), "as a gift for a runner" (occasion and audience intent), or "for someone with sensitive skin" (constraint intent). An AI shopping assistant such as Alexa for Shopping tries to match these intents to products; listings that explicitly address the relevant intents for their category are more likely to be selected as recommendations. Understanding buyer intent is the foundation of gifting and subjective-needs optimization.

SOAR (Share of Agentic Recommendations)

Share of Agentic Recommendations (SOAR) is an industry-coined metric for how often a brand's products are recommended by AI shopping assistants — such as Alexa for Shopping — relative to competitors in the same category. SOAR is an industry concept used by analysts and practitioners to frame the competitive stakes of AI shopping; it is not an official Amazon metric, and Amazon does not report recommendation frequency to sellers.

SP-API (Selling Partner API)

The Selling Partner API (SP-API) is Amazon's official programmatic interface for accessing seller data, listing data, and catalog information. It is the BSA-compliant alternative to scraping Amazon's website; tools that retrieve product data through SP-API comply with Amazon's Business Solutions Agreement, while web scraping does not. Keoxs uses SP-API exclusively to read listing and catalog data for audits. See the Amazon BSA Agent Policy guide for more on why the official API path matters.

SPN (Subjective Product Needs)

SPN refers to Amazon research published at WSDM 2025 describing how subjective shopper needs decompose into five facets: property (what the product is like), event (the occasion), activity (what the recipient does), goal (the intended benefit), and audience (who it is for). The paper presents an AI agent for gift-product recommendations. The "five dimensions" framework that circulates in the Amazon optimization community is an industry synthesis of this research, not an official Amazon seller-facing grid or evaluation framework. See gifting and subjective-needs optimization for how sellers apply these facets.

Two-stage retrieval / "5 not 50"

Two-stage retrieval is the analyst framing for how AI shopping systems are believed to work: first, a broad retrieval stage assembles a large pool of candidate products that match a query; then a re-ranking or selection stage narrows that pool down to a small recommended shortlist — often described as "five products, not fifty." This framing helps explain why keyword indexing (stage one) is necessary but not sufficient for AI visibility; listing quality and intent coverage influence stage two selection. This is industry interpretation of how AI recommendation systems generally work; Amazon has not publicly confirmed the exact internal mechanics of Alexa for Shopping's selection process. See Amazon AI: 5 products, not 50 for the full explanation.

Vine

Amazon Vine is an invitation-only review program that provides products to a network of trusted reviewers in exchange for honest published reviews, used by sellers to generate early review volume for new product launches. Vine is a confirmed Amazon program, separate from the incentivized or coordinated review practices prohibited by Amazon's policies. Reviews generated through Vine are clearly marked on the product listing.