Terms of AI visibility, briefly explained

A glossary for beginners and decision-makers: the terms from measuring, questions, answer surfaces, pages and services, each explained in two sentences and linked to the article that deepens it with a fictional example. No numbers here, denominators yes.

For referenceUpdated 5 Sep 202630 terms

Term chain · from prompt to findingterms only, no numbers

Search promptone fixed question
Answer surfacewhere the answer appears
Observationpoint in time and path
Mention, position, sourcethree values, each with a denominator
Runsthe same prompts, a new point in time
Findingpatterns and open points

Every term leads to its definition; the articles deepen it with a fictional example.

What accompanies every value

AI search engine
Model version
Measurement path
Point in time
Denominator

If one detail is missing, the value is not comparable.

The chain shows how the terms relate. The glossary carries no numbers; every article carries its own fictional example.

In short

Seven articles, one language.

All terms follow the same rule: every value states its denominator, its answer surface, its measurement path and its point in time. Whoever looks up a word here finds the article that shows it with a fictional example, and the methodology that describes the measurement paths.

Group 1 · Measuring8 terms

What is counted.

AI visibility: The documented presence of a brand or domain in answers of AI search engines to a fixed set of search prompts. It only becomes meaningful with denominator, answer surface, measurement path and point in time. What is AI visibility?
Observation: An answer to one prompt on one answer surface at one point in time, documented and reviewed. It is the smallest unit of every measurement; a finding only emerges from many. Deep Audit, step 3
Mention: The brand or domain appears recognisably in the answer. The mention rate relates answers with a mention to all valid observations. Three values
Observed position: The place at which the brand is named within a single answer, for instance as the second of three named providers. It applies to this answer only and is not a ranking. Three values
Source of the mention: The link the AI search engine visibly supplies for the mention, with domain and page type. It proves which address was supplied, not why. Sources in AI answers
Denominator: The number of valid observations a value refers to. Without a denominator a percentage cannot be checked. Measurement frame
Run: A complete repetition of all prompts on all answer surfaces at a new point in time. Two runs show which observation repeats. Deep Audit explained
Finding: The interpretation of several observations with numerator, denominator and evidence rows. A finding names patterns and open points, not causes and no overall score. Deep Audit, step 4

Group 2 · Pages8 terms

What makes a page quotable.

SEO (search engine optimisation): The work of making pages findable, understandable and classifiable in classic search systems. It creates the foundation that AI answers need as well. SEO and GEO
GEO (Generative Engine Optimisation): The work of making content picked up, correctly classified and evidenced in generated answers of AI search engines. The term is not yet used consistently. Path 2, GEO
AEO (Answer Engine Optimisation): A neighbouring term of GEO that targets direct answers, such as answer boxes or voice assistants. What is measured is the same in both cases: whether and how a brand appears in answers. The bridge
AIO: An abbreviation with two meanings: Google AI Overviews or “AI optimisation”. peaksly does not use the short form and names the answer surface instead. Six answer surfaces
Quotable sentence: A sentence that stands on its own, carries a number with a unit and knows its date. AI answers pick up single sentences, not paragraphs. Sentence and evidence
Evidence: The source right at the sentence: data sheet, standard or the company’s own measurement. Without evidence a quotable sentence remains a claim. A sentence gets its evidence
Canonical: The statement in the head of a page which address is the main address. It assigns filter, print and legacy variants to the same page. Uniqueness
Zero-click: An answer that names a brand without anyone visiting the page. The mention is then visible but not measurable as a visit. Visibility without clicks

Group 3 · Surfaces and paths6 terms

Where an answer appears and how it is measured.

AI search engine: A system that delivers a generated answer to a question instead of a results list, such as Google AI Overviews, ChatGPT or Perplexity. One AI search engine can have several answer surfaces. Six answer surfaces
Answer surface: The interface on which an AI answer becomes visible. Every surface is observed separately and never offset against another. Two families
Visible interface: The consumer interface people actually use, with order, sources and overlays as they appear there. The Deep Audit reads it by hand. Interface and API
Measurement path: The path on which an observation comes about: visible interface, public guest interface or direct API. The path is stated at every value and never mixed. Methodology
API measurement path: The technical interface of an AI search engine as an agreed, repeatable measurement path for ongoing monitoring. An API value is not a visible consumer interface. Path per surface
Crawler: The program with which an AI search engine fetches pages before it can quote from them. Whoever locks out crawlers wholesale locks out the answer. Access

Group 4 · Questions5 terms

What is measured with.

Search prompt: The fixed question put to an AI search engine. In the knowledge area it is called a search prompt at first mention and a prompt afterwards. Choosing prompts
Prompt set: The versioned set of all prompts used to measure AI visibility. It mixes search intents and only changes with a new version. From conversation to set
Search intent: What a person wants to achieve with a question: orientation, comparison, purchase or checking a brand. Every prompt has exactly one search intent. Four intents
Brand prompt: A prompt that already names the company’s own brand. It tests the echo of the name, not findability, and therefore stays deliberately rare in the set. Four intents
Version of a set: The recorded state of wording, language and market of all prompts. If the wording changes, a new measurement series begins. Versions

Group 5 · Services3 terms

What peaksly makes of it.

Monitoring: The managed, ongoing measurement of a prompt set via agreed measurement paths at a fixed cadence. peaksly runs measurement, review, interpretation and reporting. Product and workflow
Deep Audit: The manual snapshot: fixed prompts on agreed visible answer surfaces, twice each, every answer reviewed by hand. It complements monitoring and does not replace it. Deep Audit explained
Reviewed by hand: The review note that a person has read the observation and confirmed mention, position and sources. Only then does the row count. Review

All articles in the series. Deep Audit explained · What is AI visibility? · SEO and GEO · Choosing prompts · Sources in AI answers · Why not cited · Six answer surfaces

From the term to the measurement.

The fundamentals article shows what stands behind the term AI visibility; the methodology describes the measurement paths and their limits.

AI visibility explained Read the methodology →