Generative Engine OptimizationGEO

The work of getting an AI system to name a company when someone asks for a recommendation. GEO operates on two fronts: your own website, so a model can read and classify it, and the third-party pages the model draws on as sources. The second front is the larger one, and the one classic SEO does not cover.

What this means for youRanking seventh on Google still earns clicks. Missing from the three names in an AI answer means you do not exist for the person asking.

Where AI answers come from ›

Answer Engine OptimizationAEO

A second name for much the same thing as GEO. Some vendors split them: AEO for the answer box at the top of search, GEO for chat systems. In practice the measures overlap so heavily that the distinction rarely holds.

What this means for youWorth knowing only when a vendor tries to sell you both separately.

AI visibility

How often and how prominently a business appears in the answers AI systems give. Unlike a search ranking this is not a single position but a distribution: the same question returns different names depending on system, country and day. That is why you measure repeatedly rather than once.

What this means for youOne query tells you almost nothing. Ten questions across several weeks give you a picture.

Visibility score

A number between 0 and 100 summarising a company's measurement results: how often it gets named, in what position, across how many of the systems checked, and whether the mention carries a source. At TheCited the value sits in the customer dashboard and recalculates weekly.

What this means for youThe score on its own says little. Its direction over eight to twelve weeks says everything.

Mention

An AI system writes the company name into its answer. A mention is the minimum form of visibility and not yet evidence: a model can produce a name from memory without giving any source for it.

What this means for youBeing named brings customers. Being named with a source brings trust as well.

Citationsource

The AI system does more than give a name: it links the page the information came from. That linked page is the source. It usually belongs to a directory, a best-of list, a forum or a press article rather than to the company being named.

What this means for youTo understand why a competitor gets named, look at their sources. The answer sits there more often than on their website.

Where AI answers come from ›

Share of answersshare of voice

The percentage of checked questions where a company gets named. Ten questions, asked three times, across five systems produce 150 answers. A name appearing in 40 of them gives a share of 27 percent.

What this means for youThis number compares cleanly against competitors because it does not depend on how many questions you ask.

Presence

Whether a company appears on a page that AI systems read as a source. A directory can be cited a hundred times: if your name is not on it, none of that reaches you. Presence is the gap between the sources that count and the sources you appear on.

What this means for youThe fastest lever there is. An entry in a frequently cited directory takes effect within weeks.

AI Overview

The answer text Google generates above the classic search results. It names companies and links sources, so it behaves like an AI answer while appearing inside search. Not every query triggers one.

What this means for youFor local services it shows up often and pushes the blue links further down.

Promptquestion

The question a person puts to an AI system. In a measurement the prompt is the controlled variable: same wording, same country, repeated several times. Only then can you compare one week against another.

What this means for youChange the wording and you change the result. A frozen question set is the basis of any tracking over time.

Query fan-out

An AI system breaks one question into several searches, runs them in parallel and assembles the answer from the hits. "Which wedding photographer on Lake Garda" becomes a dozen internal searches for places, styles and prices.

What this means for youThis is why optimising for a single keyword helps little. You need to be findable across many neighbouring questions.

Do you need to rank to get cited? ›

llms.txt

A proposed text file in a website's root directory meant to tell AI systems what content lives there. The proposal dates from 2024. No major provider has publicly confirmed that it reads the file.

What this means for youQuick to add, harmless, and no substitute for a single one of the measures that demonstrably work.

Does llms.txt do anything? ›

Structured dataSchema.org, JSON-LD

A machine-readable block in a page's source code that sorts facts into fixed fields: company name, address, opening hours, prices, reviews. The usual format is JSON-LD, the vocabulary comes from Schema.org.

What this means for youA model then does not have to guess the name out of running text. Most effective for address, phone number and opening hours.

Does schema get you cited? ›

NAPname, address, phone

Company name, address and phone number written identically on your own website, in your Google Business Profile and across every directory. A difference as small as "St." against "Street" is enough for systems to treat two entries as two companies.

What this means for youThe dullest item on any roadmap and one of the most effective for local businesses.

AI crawlerGPTBot, ClaudeBot, PerplexityBot

The programs AI providers use to fetch web pages. Each provider runs its own under its own name: GPTBot and OAI-SearchBot for OpenAI, ClaudeBot for Anthropic, PerplexityBot for Perplexity, Google-Extended for Google. A site's robots.txt can block them individually.

What this means for youMany website builders block these bots by default. Blocked means you cannot be cited.

Live search and indexretrieval

Two routes by which a model reaches facts. With live search it runs a web search for your question and reads the hits. With an index it falls back on a stored collection refreshed at longer intervals. Perplexity searches mostly live, ChatGPT leans more on stored data.

What this means for youThis explains the different response times: Perplexity shows changes after two to eight weeks, ChatGPT takes months.

Entity

A thing a system recognises as distinct: a company, a person, a place. A business is clean as an entity when its name, location and activity are described the same way everywhere and the entries point at each other.

What this means for youBlur the entity and your visibility splits across two half-companies, both too weak to get named.

Hallucination

An AI system produces information that is wrong and that it has no source for: an invented opening time, a wrong price, a company in the wrong place. Hallucinations occur more often where little solid material about a subject exists online.

What this means for youAnother reason to publish clear facts. What a model cannot find, it tends to invent.