SEO, AEO and GEO glossary: the words, defined once.

A system that responds to a query with a composed answer rather than a list of links, drawing on retrieved web pages and citing some of them. Examples: Google AI Overviews and AI Mode, ChatGPT search, Perplexity, Gemini and Claude.

Why it matters: If your buyers ask an answer engine, the answer names a few brands. Being one of them is the whole point of AEO and GEO.

Structuring content so answer engines can extract, quote and attribute it directly: one clear question per section, a direct answer first, supporting evidence, and pages that are crawlable, indexable and eligible for snippets.

Why it matters: AEO is the on-page half of being quoted. It costs nothing but discipline and it also improves ordinary rankings. Read the guide.

Increasing how often, and how favourably, a brand or page is cited in AI-generated answers. The term was introduced in the paper "GEO: Generative Engine Optimization" (arXiv 2311.09735), first submitted on 16 November 2023.

Why it matters: GEO leans on presence beyond your own site: the review sites, communities and publications that AI engines quote. Read the guide.

Source: arxiv.org

Google's AI-generated summary shown at the top of some search results, with links to supporting pages. Google reported over 2.5 billion monthly active users at I/O on 19 May 2026, and states that no special optimisation, files or schema are required to appear.

Why it matters: When an Overview appears, fewer people click the ordinary results. Being cited inside it recovers some of that attention.

Source: blog.google

Google Search's conversational mode for deeper exploration and follow-up questions. It reached 1 billion monthly active users within a year of launch (Google I/O, 19 May 2026) and uses query fan-out over the Search index. It launched in India in July 2025, Google's first market outside the US.

Why it matters: Indian buyers see AI Mode by default. What it says about your brand is a question you can answer today.

Source: blog.google

A link or named source shown by an answer engine as the basis for part of its answer. Similarweb measured citations in 6.8% of ChatGPT answers in May 2026, up from 1.6% in June 2025.

Why it matters: Citation rate, mention rate and share of voice are the three numbers we report every month.

Source: aisearch.similarweb.com

A search that ends without the user visiting any website. Bain estimated in February 2025 that about 60% of searches end without progressing to another destination; Pew found in July 2025 that users clicked a result in 8% of visits with an AI summary versus 15% without.

Why it matters: Zero-click is why we report leads and revenue rather than traffic. Traffic is a shrinking proxy.

Source: www.pewresearch.org

A method where an AI system retrieves relevant documents, for example from a search index, and supplies them to the model at answer time so the answer can cite current pages. Google states that its generative AI search features use retrieval-augmented generation.

Why it matters: RAG is why being retrievable matters more than being in training data. If your page cannot be fetched at answer time, it cannot be cited.

Source: developers.google.com

The technique described in Google's AI optimization guide where one user question is expanded into multiple related sub-queries run against the index, and the results are combined into a single answer.

Why it matters: A page that answers a sub-question clearly can be pulled into an answer even if it never ranked for the original phrase.

Source: developers.google.com

A Markdown file at a site's root, proposed by Jeremy Howard on 3 September 2024, giving AI systems background and a curated list of links. Google states that it does not use the file for Search or its AI features.

Why it matters: We publish one because it costs ten minutes and harms nothing. We do not sell it as a tactic, and we say so. Read the guide.

Source: llmstxt.org

The user agents AI companies use to fetch web pages, split by purpose: training (GPTBot, ClaudeBot), search indexing (OAI-SearchBot, Claude-SearchBot, PerplexityBot) and user-initiated fetches (ChatGPT-User, Claude-User, Perplexity-User). Blocking a training bot does not remove you from that company's search answers; blocking a search bot does.

Why it matters: Many sites block AI crawlers by accident through a CDN default. It is the first thing we check.

Source: developers.openai.com

A robots.txt token that controls whether a site's content is used to train future Gemini models and for grounding in Gemini apps and Vertex AI. Per Google, it does not affect inclusion or ranking in Google Search.

Why it matters: Blocking it does not protect your rankings and may reduce your presence in Gemini. We leave it allowed.

Source: developers.google.com

Experience, Expertise, Authoritativeness and Trustworthiness: the qualities Google's ranking systems and quality raters look for in content and its creators. Google's guidance says trust is the most important, and asks whether it is self-evident to visitors who authored the content.

Why it matters: Named authors, real bios, dated sources and an honest "how we made this" note are how a small site earns trust.

Source: developers.google.com

Machine-readable annotations, usually JSON-LD using the schema.org vocabulary, describing a page's entities such as Article, Person, Organization, FAQPage and Product. It helps search engines understand content. Google says no special schema is required to appear in AI Overviews or AI Mode.

Why it matters: We add it as hygiene, to make the brand an unambiguous entity, not as an AI trick.

Source: developers.google.com

The number of pages an AI company's bots fetch for every visit they send back to websites, tracked by Cloudflare. For 1 to 7 August 2025 Cloudflare reported roughly Anthropic 50,000:1, OpenAI 887:1 and Perplexity 118:1. The figure moves monthly and should always be dated.

Why it matters: It shows why AI referral traffic is small for most sites, and why we track it as its own channel rather than expecting it to replace Google.

Source: blog.cloudflare.com

The percentage of AI answers within a fixed set of prompts that mention or cite a brand, compared with a named set of competitors on the same prompts. Because answers vary between runs, it is measured over repeated runs and reported with the prompt count, engines and date window.

Why it matters: This is the headline AI metric in our monthly report. Never a "ranking", always a rate.