Answer engine optimization · Reference
The AI search glossary for local business owners (20 terms, no marketing degree required)
Every industry sprouts its own jargon, and AI search built one overnight. Most articles on this stuff are written for marketers who already know the code. This one's for you, the owner who's wondering why ChatGPT recommended the place down the street instead of yours. Twenty terms, explained like we're sitting across from each other with coffee.
The engines themselves
Answer engine
The catch-all for ChatGPT, Perplexity, Gemini, and Claude when people treat them like they used to treat Google: typing a question and getting an answer right there instead of a page of blue links to sift through.
AI Overview
Google's homegrown answer engine: the AI-written summary now parked at the top of plenty of search results, above the traditional listings. It grabs from multiple sources and names them, so it acts like ChatGPT even though it lives inside Google's walls.
LLM (Large Language Model)
The engine under the hood of ChatGPT, Claude, and Gemini. It's software trained on mountains of text that guesses which words should follow next. You don't need to grasp the mechanics to build your business around it, the same way you never needed to understand PageRank to care where you ranked on Google.
Zero-click search
When someone gets their answer straight from the AI and never touches a website. It's spreading fast, and it's flipping the old goal on its head: "get them to click my link" becomes "get my name into the answer they never leave."
Getting named (or not)
GEO (Generative Engine Optimization)
The work of influencing how your business appears when an AI answer engine talks about your field or your town. It's SEO's direct heir, except you're not chasing a results page, you're chasing a paragraph an AI writes on the spot. JPMorgan called this out by name in its 2026 technology trends report, which tells you it's crossed into mainstream territory.
AEO (Answer Engine Optimization)
Mostly swapped in for GEO. If there's a real difference, AEO tilts toward building content that answers a precise question, while GEO covers the wider set of signals (reviews, listings, mentions) that determine whether an engine trusts you enough to say your name at all.
Citation
That moment when an AI answer actually names your business, the thing every audit on this blog is designed to track. Landing citations is the whole point now. You can rank beautifully on Google and still collect zero citations from ChatGPT, because the two systems pick sources for entirely different reasons.
AI citation share
Among everyone battling for a given question in your city, what slice of answers actually names you versus names a competitor versus names nobody. We've run this across hundreds of queries in med spa, day spa, and wellness, and the division's almost never fair. A few businesses hoard most of the share, and most get shut out completely.
Reverse-citation audit
Starting from an AI's answer and tracing backward to see exactly where it pulled from, then checking whether you show up clean on those same sources. If an engine relies on Google Maps, Yelp, and local press for your category, that's precisely where your listing needs to be bulletproof, rather than on channels the engine never touches.
Trust threshold
The pattern where AI engines don't slowly grade businesses by quality, they seem to draw a hard line and only mention those above it. Our data and outside studies both flag review count and star rating as one version of this cutoff: drop below roughly 4.3 stars or too thin a review base, and some models exclude you entirely rather than ranking you lower.
How the machine decides what to trust
Hallucination
When an AI states something false with total confidence: a business name, address, or phone number that's wrong or invented. It happens less with straightforward local questions than open-ended ones, but it's real enough that keeping your facts straight everywhere online actually matters.
RAG (Retrieval-Augmented Generation)
The tech behind why ChatGPT can talk about a place that opened last week. Instead of leaning only on what it memorized during training, RAG lets it search the live web first, then craft an answer grounded in what it just found. This is why fresh listings, recent reviews, and current pages suddenly carry weight.
Crawler (or bot)
An automated program that collects the contents of websites. Googlebot has been doing exactly that for decades. AI companies run their own, and the important part is that each vendor runs two kinds. Search crawlers like OAI-SearchBot and Claude-SearchBot build the index the assistant reads when it answers, so blocking one of those makes you invisible to that engine. Training crawlers like GPTBot and ClaudeBot only collect text for model training, and blocking them costs you nothing in citations. A lot of robots.txt templates end up blocking the wrong one. We keep a full reference of which name does what.
Structured data (schema markup)
Extra code on your webpage that lays out facts in a format machines can read directly (your hours, address, services, reviews) instead of forcing a crawler to parse sentences and guess. It doesn't promise a citation, but it erases ambiguity about basics, and we've confirmed AI engines do use it when it's present and right.
Knowledge graph / entity
How search engines and AI models store facts about a specific "thing," your business, as a connected object rather than just text on a page. Once you're recognized as a distinct entity, tied to a consistent name, address, and category across the web, engines have an easier time matching you to the right question.
llms.txt
A newer, still-optional file some sites are parking at their root, robots.txt-style, meant to give AI crawlers a tidy summary of what's on the site and where. Adoption's early and patchy across engines, so treat it as a small bonus signal rather than a replacement for the fundamentals above.
Where the old and new rules meet
E-E-A-T
Google's framework for sizing up content: Experience, Expertise, Authoritativeness, Trust. It came before the AI search wave, but the same thinking carries over to how answer engines weigh sources, especially for anything involving health or money, where a bad recommendation has real consequences.
Directory
A third-party listing site like Yelp, RealSelf, or Healthgrades that AI engines often cite instead of your own website. In our data, directories make up the majority of sources for broad questions, which is why showing up correctly on the right directories for your category rivals your own site in importance.
Prompt
The actual words a customer types or speaks into an AI tool. Grasping the real prompts people use in your category, "best day spa near me" versus "where can I get a facial in [city] this weekend," beats guessing at keywords, because the AI answers the specific question asked rather than a general topic.
Training data
The massive body of text a model learned from before launch day. This sits apart from what RAG pulls live, and it's why a model can be confidently wrong about anything that changed after its knowledge cutoff, one more reason live, current information about your business matters more than it used to.
If half these terms were foreign to you an hour ago, that was the point. You don't need to become a GEO expert to run your shop. You need enough vocabulary to ask the right questions, starting with whether your business gets mentioned at all. The free check below answers exactly that.