The GEO and AEO glossary

GEO (Generative Engine Optimization)

The practice of getting your brand mentioned in the answers of generative engines like ChatGPT and Gemini. SEO aims to climb the results list; GEO aims to enter the answer itself. Related page →

AEO (Answer Engine Optimization)

Preparing content for surfaces that answer questions directly: AI summaries, voice assistants, chat engines. It overlaps heavily with GEO; the two usually describe the same work. Related page →

AI visibility

How often and in what way your brand appears in AI engine answers. It is a measurable metric: in which questions are you mentioned, at which position, with which framing? Related page →

LLM (Large Language Model)

The model type behind ChatGPT, Gemini and Claude. Trained on text, it produces probabilistic answers — which is why the same question does not always get the identical answer.

Prompt tracking

Asking the questions your customers ask to the engines on a schedule and recording the brands in the answers. This is the core method of AI visibility measurement. Related page →

Visibility score

Signals like mention rate, position and share of voice reduced to a single 0-100 number. How rankzupAI computes it is written plainly on the methodology page. Related page →

Mention rate

The share of tracked runs where your brand appears in the answer. Mentioned in 8 of 20 runs means a 40 percent rate.

Position

When an answer lists multiple brands, where yours comes in the list. First position in an AI answer is the equivalent of the first result in classic search.

Citation

The source link an engine bases its answer on. Perplexity and AI Overviews show citations openly; the sites in that list shape the answer's content. Related page →

Grounding

A model basing its answer on current web sources. Gemini's search-backed answers are the classic example; grounding makes the sources it reads measurable. Related page →

AI Overviews

The AI summary at the top of Google search, compiled from multiple sources. A large share of users read only this summary before deciding. Related page →

Google AI Mode

Google search turned into a conversation: follow-up after follow-up, often without ever opening the results list. Strongest on English queries for now. Related page →

Share of voice

Your share of all brand mentions across the answers. The most honest summary of the competitive picture: what percent of the market conversation is yours? Related page →

Sentiment

The framing an engine uses when mentioning you: positive, neutral or negative. Being mentioned is not enough; how you are mentioned moves the purchase decision.

Hallucination

A model stating something that does not exist as if it did. Wrong prices, invented features or false claims about your brand fall here; watching for them is reputation work.

Zero-click search

The user getting the answer on the search page or inside the assistant and never visiting a site. AI answer surfaces amplify this behavior; visibility is now won before the click. Related page →

RAG (Retrieval-Augmented Generation)

The model fetching information from external sources before answering and building the answer on it. Perplexity and every search-backed answer works on this architecture.

AI bots and crawlability

Whether crawlers like GPTBot, Google-Extended and PerplexityBot can read your site. A misconfigured robots.txt means the engines never read you — and your chance of entering answers closes at the door. Related page →

Confidence band

Showing a low-data score as a range instead of a single number. LLM answers are probabilistic; honest measurement shows the uncertainty instead of hiding it. Related page →

Brand-free query

A question without any brand name in it: 'best protein powder' is brand-free, 'is X trustworthy' is branded. Fair competitor comparison uses brand-free questions only; branded ones inflate their owner artificially. Related page →

AI searchability

How readable and quotable your site is for AI engines. It differs from classic crawlability in its goal: not getting indexed, but becoming the source an answer is built from. Bot access, clear answer structure and extractable content decide it. Related page →