Absorption (GEO)
The process by which a generative engine uses a source inside its answer, with or without a visible citation.
SFT · 6 fiches
Réponses directes sur Structure, Flow, Trust et les cas d’usage de la méthode.
SFT découpe toute stratégie de visibilité en trois couches : Structure (architecture technique et sémantique), Flow (circulation de l'autorité et de l'attention), Trust (signaux de fiabilité pour les moteurs et les humains). Chaque recommandation Google ou GEO se range dans l'une de ces trois couches.
Les moteurs modernes raisonnent en entités reliées dans un graphe de connaissance, pas en chaînes de mots-clés. Exister comme entité identifiable donne une base stable pour être compris et cité, indépendamment des fluctuations d'algorithme.
Site jeune ou technique : commence par Structure. Contenu correct mais peu de conversions : regarde Flow. Site propre techniquement mais absent des réponses IA : le problème est presque toujours Trust.
Structure : menu balisé en schema Menu/MenuItem et fiche GBP complète. Flow : maillage entre pages plats signature et page réservation. Trust : avis récents et cohérence NAP sur tous les annuaires locaux.
Structure : architecture claire entre fonctionnalités produit et ressources RH. Flow : maillage entre glossaire RH et pages produit. Trust : études de cas clients et conformité réglementaire affichée clairement.
Sur un produit traitant des données sensibles, Trust domine largement Structure et Flow : politique de confidentialité claire, certifications affichées, et transparence totale sur l'usage des données avant toute optimisation de conversion.
The process by which a generative engine uses a source inside its answer, with or without a visible citation.
Optimization for direct answers, featured snippets and answer engines, where the goal is extraction rather than only ranking.
Google Search experience that provides AI-generated answers and follow-up interactions inside the search journey.
Google Search feature that summarizes answers with generative AI and may cite supporting sources.
A link from an external website to another website, historically used by search engines as a signal of authority.
HTML meta tag or HTTP header telling search engines not to index a page.
HTML link element indicating the preferred URL when similar or duplicate pages exist.
The act of being named or referenced as a source inside a generative engine answer.
Core Web Vitals metric measuring unexpected layout movement during page loading.
A set of related pages organized around a central topic or pillar page.
The process of removing, merging or improving weak content to protect overall site quality.
Google performance metrics measuring loading speed, visual stability and interaction responsiveness.
The amount of crawling search engines can allocate to a website during a given period.
Google tool used to ask the search engine to ignore specific backlinks in exceptional cases.
Google quality concept covering Experience, Expertise, Authoritativeness and Trustworthiness.
SEO approach that optimizes people, brands, concepts and organizations as identifiable entities rather than isolated keywords.
Schema.org type used to describe visible question and answer content on a page.
Discipline focused on structuring content so it can be extracted and cited by generative answer engines. Formalized in 2023 by Princeton and IIT Delhi research.
Crawler or control identifiers used by AI companies and search engines to access or govern content usage.
Google system targeting content created mainly for search engines instead of real users, integrated into core systems.
HTML attribute indicating language or regional page variants to search engines.
Core Web Vitals metric measuring page responsiveness to user interactions.
A database of entities and relationships used by search engines to understand people, organizations, places and concepts.
Core Web Vitals metric measuring when the largest visible content element finishes loading.
Proposed file format intended to summarize a site for language models. It is not an official ranking requirement.
Quebec law requiring French to receive equivalent or stronger treatment in many commercial contexts.
A brand or entity mention inside an AI answer without necessarily providing a clickable link.
Google indexing approach where the mobile version is the primary version used for indexing and ranking.
Google algorithm update targeting thin, duplicate and low-quality content.
Network of websites built primarily to manipulate link authority.
Google algorithm update targeting manipulative links and keyword stuffing.
Large-scale page generation from data and templates to cover many query combinations.
Process where an AI system retrieves external sources to ground or support its generated answer.
Root file giving crawlers instructions about which areas of a site they may access.
Schema.org property linking an entity to its official external profiles and identifiers.
Shared vocabulary and JSON-LD format used to describe entities, relationships and content for machines.
XML file listing URLs to help search engines discover and crawl a website.
Page that returns HTTP 200 while behaving like a not-found page.
Approach combining search optimization and user experience quality.
Perceived authority of a site or entity on a subject, built through depth, consistency and reliability.
Server response metric measuring the delay before the first byte is received.
Emerging protocol concept for AI agents to interact with structured commerce data and trusted merchant identity.
Google category for high-stakes topics such as health, finance, safety and law.
Search result where the user gets the answer without clicking through to a website.