In one sentence: This page defines, in one plain sentence each, every term used across the explainer pages, so that a reader (or an AI) can look up a word and get an answer without a sales pitch attached.
Terms are grouped by topic rather than alphabetized, because the words make more sense next to their neighbors. Each definition stands on its own. Where a number appears, its source is at the bottom.
How a search engine works
Search engine. A system that stores a copy of web pages in an index and returns a ranked list of them when someone types a query. Google, Bing and DuckDuckGo are search engines.
Crawler (also crawling, Googlebot). An automated program that visits web pages by following links and reads what it finds so it can be added to the index. Google’s main crawler is called Googlebot.
Index (also indexing). The search engine’s stored database of pages it has crawled and decided to keep; Google states that indexing is not guaranteed for every page it processes.[1]
Serving. Google’s word for the step where it pulls matching pages from the index and orders them in response to a query.[1]
Ranking. The order in which results appear; position one is the first result under the ads.
Algorithm. The set of rules and systems the engine uses to decide ranking; Google says it involves hundreds of factors and does not accept payment for position.[1]
Organic result. A result that earned its position through ranking rather than payment.
Paid result (ad). A result that appears because the advertiser won an auction; labeled as an ad and separate from organic ranking.
Query. The words someone types or says to a search engine or an AI.
Click-through rate (CTR). The share of people who saw a result and clicked it; the number-one organic result received 27.6% of clicks and all of page two combined 0.63% in one 12.2-million-query study.[2]
Zero-click search. A search that ends without a click to any website; 68% of US Google searches ended that way in early 2026, up from 58.5% two years earlier.[3]
Featured snippet. A 2015-era Google feature that quoted one passage from one page in a box above results; a forerunner of AI Overviews.
Local search
SEO (search engine optimization). The work of making a website easy for a search engine to find, understand and trust, so it ranks well for relevant queries.
Local SEO. The subset of SEO focused on searches with a place in them and on the map results; its main inputs are the business profile, reviews, and consistent name, address and phone data.
Google Business Profile (GBP, formerly Google My Business). The free listing that represents a business on Google Search and Maps, holding its name, category, services, hours, photos, reviews and service area; the single most cited source in local AI answers at 28.63% of citations.[4]
Map pack (local pack, 3-pack). The block of three businesses with a map that Google shows for searches with local intent.
NAP. Name, address, phone: the three facts that must match everywhere a business is listed.
Consistency. The degree to which a business’s NAP, hours and services agree across its profile, website, reviews and directories; the main way an engine decides a fact is true.
Service-area business. A business that travels to customers rather than receiving them at a storefront, such as a plumber; recommended by AI at a similar rate to storefront businesses (29.4% versus 33.6%).[4]
Directory. A third-party site that lists businesses, such as Yelp, Facebook, Angi, Healthgrades or a manufacturer’s dealer locator; Yelp was 9.53% of local AI citations.[4]
Reviews. Customer-written ratings and comments on Google, Yelp and elsewhere; read by AI systems as evidence for a business’s claims, and read by 97% of consumers for local businesses.[5]
Review signals. The count, recency, rating and wording of reviews, rated among the top three local ranking factors by practitioners.[6]
Schema (structured data, JSON-LD). A machine-readable way of labeling facts on a web page, such as business name, address and hours; carried by only 3.97% of mobile pages in the LocalBusiness form.[7]
llms.txt. A proposed text file meant to guide AI crawlers; in one check of 137,210 domains, 97% of such files received no requests.[8]
Answer engines and generative engines
Answer engine. Any system that replies to a question with a composed answer rather than a list of links; AI Overviews, AI Mode, ChatGPT and Perplexity are answer engines.
Generative engine. An AI system that writes its reply fresh each time rather than retrieving a stored one; the same systems seen from the machine’s side.
Large language model (LLM). The AI component that understands and writes language; the engine’s “brain,” which for local questions has no memory of your business and relies on retrieval.
Retrieval. The step where a generative engine fetches current sources (profiles, websites, reviews, directories) to write from, at the moment of the question.
Query fan-out. Google’s term for AI Mode breaking one question into several sub-queries and running them at once.
Citation. The source link an answer engine attaches to a claim; the currency of a generative engine, with 10% to 11% of URLs collecting 37% of citations in one million-citation analysis.[9]
Entity. A single recognizable thing, such as one business with one name, address, phone and set of services, that the engine can form from agreeing sources.
Entity trust. The engine’s confidence that an entity is real, specific and currently operating as claimed; built from consistency across sources.
Prompt. The question or instruction a person gives an AI; in visibility measurement, a fixed set of realistic prompts is run repeatedly.
Consistency (of answers). How often the same prompt returns the same businesses; 20% to 33% overlap between repeated runs in the largest local study.[4]
AI Overviews (AIO, the Google feature). The AI-written summary that appears above ranked results on many Google searches; rolled out to all US users on May 14, 2024.[10]
AI Mode. A separate tab in Google Search, launched in the US on May 20, 2025,[11] where the whole page is a conversation with follow-up questions; passed one billion monthly users in 2026.[12]
ChatGPT. OpenAI’s assistant; reached one billion weekly users in July 2026.[13] and was the most-used AI for local recommendations at 31% of AI users[14]
Perplexity, Gemini, Copilot, Claude. Other answer engines from Perplexity AI, Google, Microsoft and Anthropic respectively; same idea, different companies.
The optimization terms
AEO (answer engine optimization). The work of being the thing an AI says back when asked a question, instead of one of ten links; named for the output.
GEO (generative engine optimization). The work of being retrievable, consistent and cited when a generative engine composes its answer; named for the machine, and in practice the same job as AEO.
LLMO (large language model optimization). Usually a synonym for GEO; when used to mean influencing a model’s training rather than its retrieval, it does not apply to a local business.
AIO (AI optimization, the agency term). An umbrella label for AI-related marketing, not a distinct discipline; not to be confused with AI Overviews.
AI SEO. SEO adjusted for the fact that AI answers now sit on top of the ranked list; the same work as AEO and GEO under a reassuring name.
ASO (assistant search optimization). Being findable and bookable through voice and chat assistants; mostly the same as GEO, with extra weight on the sources assistants act from; not app store optimization.
AI visibility. How often and how consistently AI systems name a business when asked for a recommendation in its trade and area; a measurement, not a discipline.
AI visibility score. A single number, usually 0 to 100, that summarizes how often a business is named across a fixed set of prompts, platforms and repeated runs; there is no industry standard, so scores are comparable only within the same tool over time.
Recommendation rate. The share of relevant AI answers in which a tracked business is named; about a third across ChatGPT (33%), AI Mode (32%) and AI Overviews (38%), versus 66% on Google Maps.[4]
Businesses per answer. The number of businesses an AI names in one reply; averages of 4.1 (ChatGPT), 3.5 (AI Mode) and 2.5 (AI Overviews).[4]
Proximity. The distance between the searcher and the businesses an AI names; 71.7% to 78.2% of picks were within 10 km.[4]
The phone
AI receptionist. Software that answers a business phone in a natural voice, holds a conversation, collects what the business needs to know, and books the job or hands off to a person, on every call.
Voice agent. The general term for AI software that talks on the phone; an AI receptionist is a voice agent configured for one business’s front desk.
IVR (interactive voice response). The “press 1 for service” phone menu; a predecessor of the AI receptionist that reads keypad presses, not sentences.
Books the job. The standard that the caller hangs up with a confirmed time on the business’s real calendar and in its job software, rather than a promise of a callback.
Handoff (transfer). The AI receptionist passing a call to a person, warmly and with context, when the call needs one.
Speed to lead. The time between a customer’s first contact and the business’s response; the odds of qualifying a lead fall 21-fold between a 5-minute and a 30-minute response.[15]
Configuration (rules). The facts and instructions a business gives its AI receptionist, such as services, prices it may quote, what it may book, and what it must hand off.
Transcript. The written record of a call the AI receptionist produces, with a summary.
What this is not
This glossary is not a ranking-factor list, a how-to, or a vendor comparison. It defines words. It also does not attempt to cover general AI vocabulary (tokens, training, parameters) beyond what a business owner needs to read the other pages.
Where it is going
Forecasts, labelled as forecasts
A forecast, stated as one: within a few years, several of the optimization terms above (LLMO, AIO, AI SEO, conversational search optimization) will fall out of use, leaving SEO for the base layer, GEO or AEO for the answer layer, and AI visibility for the measurement. I would expect the phone terms to converge on “AI receptionist” as the plain name. New words will appear for assistants that act rather than answer. This page will be updated as the vocabulary settles.