Citation Integrity™ Glossary

A product or service that generates answers for users using AI. CiteAbility™ Outcomes currently evaluates Perplexity, ChatGPT, Gemini, Claude and Grok while in Beta, each through the provider's API with web search switched on.

Also known as: Answer engine

An automated agent used by an AI, search or retrieval service to request and read web content.

Also known as: Bot, User agent

A technical permission state showing that a crawler is not blocked from requesting a website or page. Permission does not prove that a visit occurred.

Also known as: Crawler permission

Identifying the source, organisation, author or evidence that supports a statement.

A comparative view of Citation Integrity™ results across a qualifying dataset or defined peer group. A benchmark is not a universal ranking.

A reference to a source used to support or attribute information in an answer or published statement.

How consistently published information is clear, attributable, supported and suitable for responsible reliance when an AI system considers using it as a source.

The degree to which published information can be responsibly relied upon, evidenced, attributed and used as a source for AI-generated answers.

CiteAbility’s free initial diagnostic that assesses a website’s public information for evidence, attribution, accessibility and other public Citation Integrity™ dimensions within the defined audit scope. It gives a company its first Citation Integrity™ Score, from up to 10 pages.

Also known as: Audit, Free Audit

A badge included with Starter, Growth and Enterprise, and for Agency client workspaces while the Agency plan is active, earned by passing the certification checks and never sold. There is no fee and no renewal, and a subscription does not guarantee it. Certification does not guarantee citation, retrieval, ranking, recommendation or mention by any AI system.

Also known as: Certified, Certification

The eight public dimensions used to describe what CiteAbility™ assesses: Source Authority; Entity Authority; Organizational Probity; Evidence & Citations; First-Hand Experience; Content Quality; Technical Accessibility; and Integrity Analysis. The dimensions do not disclose the proprietary scoring methodology.

Also known as: Eight dimensions, Framework

The CiteAbility capability that watches for changes or regressions in a company’s Citation Integrity position and relevant website, evidence and access conditions over time.

Also known as: Monitor

A company’s 0–100 measure of how dependable its published information is for AI systems to use and cite, at a defined assessment date. It is the result of the most recent completed assessment: the full assessment of up to 25 important pages on a paid plan, otherwise the free Audit of up to 10 pages. It changes only when a new assessment completes. It is not the average of individual items being worked on, and not a prediction or guarantee that any AI platform will cite the company.

Also known as: Score

A plain-language indication of whether the evidence observed in a defined test suggests that a company may be considered as a source. It is not a prediction or guarantee of citation.

A Beta that records observed answers, mentions, citations, recommendations and source use from Perplexity, ChatGPT, Gemini, Claude and Grok for selected questions. It does not prove that a website change caused an observed outcome.

Also known as: Outcomes

A source that is explicitly referenced or linked in an observed AI answer.

A claim is a factual statement about a business, product, service, person or event that an AI system may repeat. CiteAbility™ considers whether a claim is attributable, supported by relevant evidence, appropriately qualified and suitable for responsible reliance. Repetition or confident wording is not proof.

How much of the assessment could be fully checked and evidenced. It is separate from the score itself: a high score with low confidence means the pages assessed well, but some evidence behind them could not be completely verified — so the number should be treated with more caution, not as a worse result. Confidence is not a guarantee of truth or future AI behaviour.

One of the eight public Citation Integrity™ dimensions. It considers whether published information is clear, useful, coherent, appropriately detailed and suitable for responsible reliance.

Independent or additional evidence that supports the same factual statement or conclusion.

The plain-English conclusion of a CiteAbility™ assessment for one page, claim or finding. There are four: CITE means the information is dependable enough to be relied on as it stands; CITE WITH QUALIFICATION means it can be relied on if the stated caveat is kept with it; VERIFY means independent confirmation is needed before relying on it; DO NOT CITE means it is not currently suitable to rely on. A decision describes the published information, not the honesty of the organisation.

Also known as: CITE, VERIFY, DO NOT CITE, CITE WITH QUALIFICATION

One of the eight public Citation Integrity™ dimensions. It considers the strength and clarity of the identifiable person, organisation, product or entity associated with the information.

Information that supports, substantiates or helps verify a factual claim.

One of the eight public Citation Integrity™ dimensions. It considers whether important claims are supported by relevant evidence and appropriate source references.

One of the eight public Citation Integrity™ dimensions. It considers whether information demonstrates direct experience, original knowledge, observation, testing or other first-hand involvement where relevant.

One of the eight public Citation Integrity™ dimensions. It considers whether the available information, taken as a whole, is coherent, appropriately qualified and suitable for responsible reliance.

An observed instance where an AI answer names or refers to a company, product, service, person or other entity. A mention is not necessarily a citation or recommendation.

Evidence showing that a crawler or automated service actually requested a website or page. Observed access does not prove that the content was cited, mentioned or used in an answer.

Also known as: Crawler visit

One of the eight public Citation Integrity™ dimensions. It considers whether verified, attributable information about an organisation’s conduct, transparency or standing is materially relevant to reliance on its published information. Missing information is not treated as misconduct, allegations are not treated as established facts, and unverified information is not treated as adverse.

A recorded result showing what an AI answer environment found, understood, named or sourced during a defined test. An outcome does not prove that a website change caused the result.

Original material produced by a person or organisation with direct knowledge of the subject, event, data, product or activity being described.

Information about where content, evidence, data or a claim originated and how its source can be identified.

A status shown when an organisation-level check is still awaiting confirmation from an independent source. It is not an adverse or negative finding by itself, and it may change once independent confirmation is available.

Also known as: Provisional status

Context or wording that appropriately limits a claim so it does not overstate what the evidence supports.

An observed AI answer that positively suggests, selects or favours a company, product, service or option. A recommendation is distinct from a simple mention or citation.

The process by which an AI or search system finds candidate information or sources before producing an answer. CiteAbility™ does not claim control over an external provider’s retrieval process.

A page, document, record, person, organisation, dataset or other identifiable origin of information.

One of the eight public Citation Integrity™ dimensions. It considers whether a source is identifiable, attributable, relevant and sufficiently dependable for the information it provides.

An observed indication that an AI answer relied on, referenced, quoted, linked or otherwise drew from a source. The exact evidence available depends on the answer environment being observed.

One of the eight public Citation Integrity™ dimensions. It considers whether relevant public information can be accessed and interpreted reliably by ordinary web and machine-reading systems.

The average assessment of the pages, claims and findings currently being tracked in Fix It. These items are often selected because they need work, so this is not your company’s Citation Integrity™ Score.

Also known as: Tracked average

The act of checking whether information, identity, evidence or source details are supported sufficiently for the defined assessment purpose.

A source whose identity, origin or supporting evidence has been checked to the extent defined by the relevant CiteAbility™ product or assessment. “Verified” does not mean infallible and does not guarantee use by an AI system.