The AI SEO glossary: SEO, AIO, AEO, GEO and every term around them.

Search optimization for the engines that answer questions instead of listing links — Google AI Overviews and AI Mode, ChatGPT, Gemini, Perplexity and Copilot. It keeps everything classic local SEO does and adds AIO, AEO and GEO on top.

The category this company is filed under. The four disciplines, in the order they depend on each other, are SEO, AIO, AEO and GEO. The full definition · the service.

Search engine optimization (SEO) is the work of earning a position in Google’s organic results and local map pack for the searches a business wants — through technical health, content built around real search demand, a complete Google Business Profile, consistent citations and reviews.

The foundation. A business that has never done this has nothing for the AI-era disciplines to cite. Local SEO for contractors · the service.

AI optimization (AIO) is the work of making a business legible to AI systems — one consistent entity with the same name, services, area and facts on every page and profile, expressed in structured data and plain-text files a model can parse without guessing.

The signals: Schema.org structured data, entity consistency, llms.txt, crawl access for AI agents, clean HTML. See entity and structured data.

Answer engine optimization (AEO) is the work of getting a page selected as the answer to a question — in Google AI Overviews, featured snippets, voice assistants and chat interfaces — by opening with a direct, self-contained answer, using question-shaped headings and marking FAQs up with FAQPage schema.

The writing discipline: question-shaped headings, a self-contained answer first, specifics after, a FAQ block marked up as FAQPage. The AEO guide.

Generative engine optimization (GEO) is the work of getting a business cited and recommended inside the answers generative engines write — ChatGPT, Gemini, Perplexity, Copilot and Google AI Mode — by publishing citation-worthy specifics, being corroborated by independent sources, and keeping the entity consistent and current.

Named by a 2023 research paper, which found that specifics — sources, quotations, concrete figures — improved visibility in generated answers and that keyword stuffing did not. The GEO guide.

The part of search engine optimization concerned with searches that have a place in them — “plumber near me,” “roofer in Franklin” — and the results that answer them: the map pack, the localized organic results and, now, the local AI answer. It runs on the Google Business Profile, location-specific pages, consistent listings and reviews.

The checklist.

The work of making a site crawlable, indexable, fast and unambiguous: canonical URLs, a sitemap, a correct robots.txt, structured data, Core Web Vitals, mobile rendering, HTTPS, redirects that preserve old URLs. The part of SEO a customer never sees and every engine reads first.

The free scan checks forty-two of these on any page.

Any system that returns an answer rather than a list of links: Google’s featured snippets and AI Overviews, AI Mode, voice assistants, and the chat engines — ChatGPT, Gemini, Perplexity, Copilot. They differ in how they choose, but they all lift a passage from a source and present it as the answer.

Optimizing for them is AEO.

An answer engine that writes its answer with a language model from sources it retrieves, and cites them — ChatGPT, Gemini, Perplexity, Copilot, Claude and Google’s AI Mode. The mechanism is retrieval-augmented generation: retrieve, synthesize, cite.

Optimizing for them is GEO. The engines this site names, every time, are Google AI Overviews, Google AI Mode, ChatGPT, Gemini, Perplexity, Microsoft Copilot, Claude.

The generated summary Google places above the ordinary results for some searches, with links to the pages it drew on. Rolled out across the United States in May 2024. For local questions it leans on Google Business Profile and review data.

How a local business shows up in it.

Google’s conversational search: a separate tab that breaks a question into several searches, reads across them, writes a longer answer, keeps the thread going, and for local questions shows businesses with ratings and a map. Available to everyone in the United States since 2025.

The free AI answer check asks it your question live.

The kind of model that powers ChatGPT, Gemini, Claude and the rest: a system trained on a very large body of text that writes fluent answers. On its own it answers from what it learned in training; connected to search, it answers from what it retrieves now.

Which is why a business’s record on the web has to be consistent both today and over time.

The method a generative engine uses to answer with current information: retrieve relevant pages and records, give them to the language model, and have it write an answer grounded in them, with citations. Each step sets a requirement — be retrievable, be extractable, be worth citing.

See generative engine.

Tying a generated answer to retrieved sources so that its claims can be checked against them. A grounded answer cites; an ungrounded one is the model’s recollection. The sources a grounded answer cites are the sources a business needs to be on.

A confident, fluent statement by a language model that is not true — an invented business, a wrong phone number, a service you do not offer. Consistent, specific, corroborated facts across the web are the practical defense: they give the model something to retrieve instead of something to guess.

A bot an AI company runs to fetch web pages: GPTBot, OAI-SearchBot and ChatGPT-User (OpenAI), ClaudeBot (Anthropic), PerplexityBot, Google-Extended (Google’s AI training crawler), Applebot-Extended and others. A site that blocks them, deliberately or by a security plugin’s default, is absent from the answers they build.

Named and allowed in this site’s robots.txt, with the private paths disallowed in every group.

A thing an engine can identify unambiguously — this business, as distinct from every other business with a similar name — with a stable set of facts attached: name, address, phone, category, services, area, hours, people. An engine recommends entities it understands and other sources agree about.

The entity sheet — the exact facts every profile must repeat — is the first deliverable of the branding layer.

Google’s database of entities and the relationships between them, assembled from the web, Maps, the Business Profile and structured data. A business resolved into it is one entity with known attributes; a business that is not is a set of pages Google is still guessing about.

Name, address, phone — the three facts that must be identical on the website, the Google Business Profile and every directory listing. Disagreement between them is the most common reason a local business is not trusted by Google or the AI engines.

Two meanings, both important. In local SEO, a citation is a listing of the business’s name, address and phone on a directory or another site, whether or not it links. In AI search, a citation is a source an engine names under its answer. The first kind is how you become the second kind.

Independent sources saying the same thing about a business: the profile, the directories, the reviews, the press, the association page. Engines treat agreement across sources they did not get from the business itself as trust, and they resolve ties with it.

The free listing that represents a business on Google Search and Maps — category, services, hours, photos, reviews, address or service area. The record Google and its AI answers trust first about a local business, and one of the public records every other engine reads.

The optimization checklist.

The block of three local businesses with a map that Google shows for searches with local intent. Ranked, by Google’s own account, on relevance, distance and prominence — most of which is set by the Business Profile and corroborated by the site, the listings and the reviews.

Also called the local pack or the three-pack.

Experience, expertise, authoritativeness and trustworthiness — the qualities Google’s search quality guidelines ask raters to look for, and a useful summary of what every engine wants: a named person accountable for the content, evidence they do the work, and a record other sources corroborate.

On this site it is a named founder, an author strip on every page, and a Person node in the schema.

Customer ratings and text on Google and other platforms. Their count, pace, rating and your replies set prominence; their text — what was done, where — is evidence the engines quote. Incentivized, gated or fabricated reviews violate every platform’s policy and can cost a business its profile.

Machine-readable statements about a page and a business, written in the Schema.org vocabulary and placed in the page as JSON-LD: this is a RoofingContractor named X, serving area Y, offering services Z, and this page answers these questions. The difference between an engine guessing and an engine knowing.

What to add and what to leave out.

The format Google recommends for structured data: a block of JSON in the head of the page rather than attributes scattered through the HTML. One graph per page, with stable @id values so the same Organization is the same entity on every page.

Structured data that lists the questions a page answers and their answers, so a machine reads them as questions and answers rather than paragraphs. Google limited the FAQ rich result in its listings in 2023; the markup remains the clearest statement of what a page answers, and the answers must also be in the visible HTML.

Every page on this site carries one where it has a FAQ.

The writing pattern AEO depends on: a heading shaped like the question as customers ask it, a complete answer in the first two or three sentences that would still be correct on its own, and then the specifics that earn it. The opposite of the warm-up paragraph.

The boxed answer Google shows at the top of some results pages, lifted from one page and linked to it. The original answer engine result; answer-first pages are what it selects.

A search that ends on the results page because the answer was there — in a snippet, a map pack, an AI Overview — and nobody clicked a result. A growing share of searches are zero-click, which is why being in the answer matters more than being in the list beneath it.

A Schema.org property that marks which parts of a page are suitable to be read aloud by a voice assistant — typically the headline and the direct answer. A small signal; the passage it points at is the large one.

A plain-text Markdown file at the root of a website that tells language models what the site is and where to read more. Proposed in September 2024; not a ranking signal, not read by any engine that has said so, and still worth an hour because it forces the one-paragraph statement of the entity every profile should match.

The guide · this site’s llms.txt and llms-full.txt.

The file at the root of a site that tells crawlers what they may and may not fetch. Where the AI crawlers are allowed or blocked, and where the private paths are disallowed — in every group, because a crawler named in its own group ignores the general one.

Google’s three measurements of how a page feels to load and use on a real device: how quickly the main content appears, how quickly the page responds to a tap, and how much the layout shifts while loading. A ranking factor and, more importantly, the difference between a homeowner on a phone waiting and leaving.

The one address a page declares as its official version, so that the same page reachable under several URLs is counted once. Every page on a well-built site declares one, and it matches the address in the sitemap.

The time between a customer’s inquiry and a business’s first real reply. In home services the first business to respond usually wins the job; an answer inside a minute, from a person or an assistant that says it is one, multiplies everything spent on being found.

The guide · the AI automation layer.

An inquiry that goes to one business and is never resold or shared with a competitor — as opposed to a marketplace lead sold to several contractors at once. On this site, leads are exclusive at the level of the lead and included in a flat monthly package rather than billed per call.

Lead generation.

Counting the actions that matter — phone taps, form submissions, booked calls — and attributing each to its source: the profile, the organic results, the ads, the AI referrals. What a monthly report is built on instead of a ranking chart.

Every site we host carries a private first-party traffic dashboard with AI search as its own channel.

A Google Ads keyword setting that lets an ad show for searches related to the keyword, not only the exact words. Paired with Smart Bidding and verified conversion tracking, one consolidated broad-match campaign outperforms a dozen tightly matched ones split by city — which is how we run every account.

Google Ads.