Glossary
Explore LLM terminology with GEO diagrams and measurement examples ›
Basics
One AI assistant GeckoCheck reads, such as ChatGPT, Gemini or Google AI Overview.
How often and how well AI assistants name a brand when people ask them about its category.
What an AI model replies when it is asked a tracked query.
A query that names the brand.
A page an AI answer links to as its source.
Generative engine optimization: the work that makes a brand easier for AI assistants to find, trust and name in their answers.
A country and language the brand is measured in, read as its own analysis.
An AI answer that names the brand in its text.
A kind of buyer the brand serves. A query can belong to one persona, several, or all of them.
One period of measurement. New readings appear in the next report.
A recurring idea in how the AI answers describe brands, such as price, quality or delivery.
A subject the tracked queries are about, and where the brand is missing from it.
A question people ask AI assistants about the brand's category, measured in every report.
A query about the category, where the models choose who to mention.
The change between two percentages. A move from 20% to 25% is 5pp.
Measures
How strongly the brand appears in AI answers: the share of tracked queries where a model names it, weighted by prominence.
How strongly and positively AI answers present the brand, less open risks, graded against peers on each model.
Of every page the AI models cited in this scope, how many are the brand's own.
How much more often the brand is expected to show up in category answers once a piece of work is live.
When most of a figure rests on one AI model, so losing that model would move it a lot.
The tracked queries that have at least one read answer from a model in scope in this report.
The tracked queries where at least one model in scope gave a sentiment reading in this report.
The brand's slice of the AI conversation against tracked competitors, across citations, mentions and sources.
Queries where the brand appears on exactly one of the AI models in scope.
The AI model that speaks least positively of the brand.
Citations, mentions and sources counted by their strength, not equally.
Customer journey
The journey stage where a buyer compares a short list of options.
A step a buyer goes through, from discovery to referral, each with its own queries.
The journey stage where a buyer first learns what options exist in a category.
How much of the customer journey is healthy on each AI model: present, holding steady and well spoken of.
The journey stage where a buyer is ready to choose and asks where and how to buy.
The steepest fall in the brand's presence from one journey stage to the next.
The journey stage after buying, when a customer asks how to use, fix or return what they bought.
The tracked queries where at least one AI model names the brand.
The journey stage where a customer recommends the brand to others.
The journey stage where a customer decides whether to buy again or stay.
Journey stages that need attention because the brand is weak, slipping or spoken of negatively there.
Competition
Which brand leads each category of the market, by share of voice.
Recurring themes where competitors win or lose against the brand in AI answers.
A brand the AI models name in the same answers, tracked beside the brand.
How far the brand sits from the #1 brand, on the model where the contest is closest.
The brand against one competitor at a time, side by side on presence, share and sentiment.
How far ahead of the second brand the brand sits where it leads.
The average of every tracked brand on the same measure, scope and model.
The average of the brand's tracked peers on the same measure and model.
Tracked queries where a competitor leads the AI answers instead of the brand.
Where the brand places against the tracked market on each AI model.
Sentiment
The share of AI answers that name the brand and describe it negatively.
An AI answer that names the brand without praising or criticizing it.
The share of AI answers that name the brand and describe it positively.
Sources
Partner and reseller pages that sell or promote the brand.
A competitor's own site, cited in answers on the brand's queries.
A website address, such as example.com, that groups the pages the models cite.
Cited domains whose pages mention the brand.
Publishers and independent sites that write about the brand.
Citations of outside pages that mention the brand.
Pages the brand publishes itself.
A site that publishes content the AI models cite, such as a magazine, review site or forum.
A page the AI models used as background without citing it in the answer.
Platforms the brand can post to, such as Reddit, YouTube or LinkedIn.
A page or publisher the AI models draw on when they answer.
What kind of site a source is: owned, earned, affiliate, social or competitor.
Technical audit
A bot an AI company sends to read web pages, for answers, for training or on a user's behalf.
Whether AI models can reach and understand the brand's pages.
A group of related audit checks, such as AI access and guidance, structured data or performance.
The word beside an audit score: Excellent, Good, Fair, Needs attention or Critical.
One score out of 100 for the brand's site: how well AI models can reach and read it, and how sound it is.
How much a failed check matters: Blocks AI access, Hygiene, Recommendation or Info.
How fast, secure and sound the brand's pages are.
Code on a page that states plainly what it is about, so AI engines read it as a product, an organization or an article.
A check of the brand's own site the way AI models meet it: whether they can reach, read and quote it.
A proposed file that lists a site's key pages for AI models. No major AI provider has committed to reading it.
AI shopping
AI assistants helping people choose, compare and buy products, and sending them to stores.
How often AI shopping answers send buyers to the brand's own store.
The details shopping agents compare: price, identifiers, availability, shipping and returns.
An AI assistant that can browse a store and buy on a person's behalf.
Work and Studio
One piece of work inside a workstream, from a page rewrite to a post or an outreach email.
The queue of work for the brand: projects, the workstreams inside them and the actions in each, from finding to done.
The facts, voice and look Studio writes from, and the company record AI assistants read about the brand.
A written plan for work someone else will do, such as a page for a writer or a pitch for an outlet.
A piece of work Studio has written, waiting for review before it goes live.
The answers, queries and cited pages behind a finding.
A short summary of where the brand stands in the report and what changed.
Something the report shows that is worth acting on, such as a gap, a fall or a negative theme.
GeckoCheck's assistant. It answers questions about any page, decides what to do about a finding, and drafts in Studio.
The few actions most worth taking now, drawn from the report's findings.
The person who holds a workstream or action and moves it to done.
The work for one goal: the pages, evidence and actions that serve it.
The list of work waiting in the Action Hub, in the order to do it.
Where the work gets drafted: page rewrites, new pages, briefs, posts and presentations, each reviewed before it goes live.
A group of actions inside a project that go together, such as one page to write or one outlet to reach.