AI search for your store

AI search is a way of looking up information in which a language model retrieves pages from an index or from the web, writes an answer from them and cites some as sources, instead of just returning a list of links.

AI Overviews is the summary written by a Gemini model that Google shows above the search results for some queries, with links to the pages that back up the information.

Google AI Mode is a search experience inside Google in which a Gemini model answers long, multi-part or follow-up questions with a written response and links to the web pages that back it up.

ChatGPT Search is the ChatGPT feature that searches the web to answer with current information and links to the sources it uses, instead of replying only from what the model learned in training.

Perplexity is an AI search engine that answers each question with text composed from web pages it retrieves and links as sources, instead of returning a list of results.

Microsoft Copilot is Microsoft's artificial intelligence assistant for the general public, the one that replaced Bing Chat in November 2023.

Gemini is Google's family of artificial intelligence models and the name of the conversational assistant that uses them, the app that replaced Bard in February 2024.

A zero-click search is a search that ends without the person clicking any result: they abandon it, reformulate it or settle for what they see on the results page itself, such as an AI-generated summary, a featured snippet or a business listing.

An LLM (large language model) is an artificial intelligence model trained on enormous amounts of text to predict, piece by piece, how a sequence of words continues, which lets it write, summarize or answer questions.

Search intent is the goal a person has when they type a query into a search engine: what they want to get out of the results, beyond the words they use.

A conversational query is a search typed or spoken in natural language, the way you would ask a person: a full sentence with its context and its conditions, often part of a conversation in which each question builds on the last.

Query fan-out (also called query decomposition) is the technique an AI search engine uses to split a question into subtopics and run several related searches at once, gathering more results before it writes the answer.

Semantic search is a search technique that interprets the meaning and intent of a query, not just its exact words, to return relevant content even when it does not share the same terms: it understands synonyms, relationships between concepts and context.

Embeddings are numerical representations of a text, an image or other content: lists of numbers, called vectors, that a model calculates so that similar contents end up close to each other.

RAG (retrieval-augmented generation) is an artificial intelligence technique in which a system, before a language model answers, retrieves relevant documents from an index or a knowledge base and hands them to the model to read, so the answer rests on them and not only on what the model learned in training.

Grounding is the technique that connects a language model to external information at the moment of answering, such as a search engine's index or a company's documents, so the answer rests on those sources and not only on what the model learned in training.

An AI hallucination is an answer from a language model that sounds plausible but is false or not supported by the source it uses: a fact, a quote or a product feature the model states with confidence without it being true.

E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) is the framework Google uses to describe the quality of a page: the creator's first-hand experience, their knowledge of the topic, their authority as a source and, above all, their trustworthiness.

Helpful content is content that Google describes as created primarily to help people and not to manipulate search engine rankings: original, complete, written by someone who knows the topic and that does not send the reader off to search again.

Scaled content abuse is a Google spam policy that prohibits generating many pages mainly to manipulate search rankings rather than to help the people who read them.

Information gain is, in SEO, the idea that a page is worth more when it adds facts the reader has not already seen on other pages about the same topic.

Topical authority is, in SEO, the idea that a site gains visibility on a topic when it covers it in depth and is recognized as a reference source on it, beyond the value of each individual page.

An answer engine is a search system that, given a question, returns the answer directly instead of a list of links for the user to dig through.

AI SEO is the work of preparing a website so that AI search engines and assistants, such as Google's AI Mode, ChatGPT or Perplexity, find its pages, understand them and cite them in their answers.

GEO (Generative Engine Optimization) is the practice of optimizing a website and its content so that generative engines, such as ChatGPT, Perplexity or Google's AI Mode, use its pages and cite them as sources in the answers they compose, instead of just ranking them in a list of links.

AEO (Answer Engine Optimization) is the practice of preparing a website's content so that an answer engine, such as a search engine's featured snippet, a voice assistant or an AI chat, picks it as the direct answer to a question instead of just listing it among the results.

SEO vs AEO and GEO for an online store: what each one wins, what it asks of your product pages, how it's measured, and what Google and Bing say.

LLMO (Large Language Model Optimization) is the work of influencing what large language models, like those behind ChatGPT or Gemini, say about a brand or a product, both from what they learned in training and from what they retrieve from the web when they answer.

SGE (Search Generative Experience) was Google's experiment, opened in Search Labs in May 2023, that answered some searches with an AI-generated summary and links to dig deeper.

A generative summary is the text an AI search engine writes in answer to a query, drawing on several pages it retrieves at that moment, with citations or links to the sources that back it, and shows above the list of results or in place of it.

A search prompt is the text a person types or dictates to an AI assistant with web access, such as ChatGPT, Claude or Google's AI Mode, to get an answer based on search results.

Vector search is a retrieval technique that turns queries and documents into numeric vectors, called embeddings, and returns the documents whose vectors sit closest to the query's.

A token is the unit of text a language model works with: a short word, a piece of a word or a punctuation mark.

The context window is the maximum amount of text, measured in tokens, that a language model can take into account at once when generating an answer: the instructions, the conversation, the documents it is given to read and its own answer.

A generative engine is a search system that combines several language models with a search engine: it reformulates the query, retrieves pages, summarizes them and writes an answer of its own from them, with citations to the sources used.

Semantic SEO is the practice of creating and organizing a site's content around topics, entities and the searcher's intent, rather than isolated keywords, so that a search engine understands what each page is about and how it relates to the others.

This month's changes in AI search that affect an online store, each one dated and checked at the official source that announced it.