Definitions
AI visibility terminology for law firms
AI legal discovery is developing faster than its vocabulary. FirmRanker uses explicit definitions so research does not collapse different observations into the same metric.
Terms
- AI visibility
- Whether, where and how a law firm appears when people use AI systems to ask questions relevant to its services, lawyers, markets or expertise. Why it matters: It is a pattern across observations, not a single result. Not to be confused with: A ranking or a single screenshot.
- Appearance
- A firm entering the substantive body of an AI response. Why it matters: Distinguishes being part of the answer from appearing only in sources. Not to be confused with: Source-only presence.
- Mention
- The firm appears substantively but is not placed into an actionable consideration set. Why it matters: A mention is not an endorsement. Example: “Firm A handled a well-known case in 2019.” Not to be confused with: Recommendation.
- Suggestion
- The firm is presented as one the user may consider. Why it matters: Most AI list answers are suggestions, not recommendations. Example: “Firms in Sydney include Firm A, Firm B and Firm C.”
- Recommendation
- The AI system explicitly recommends or endorses the firm. Why it matters: The strongest form of response treatment. Example: “I would recommend starting with Firm A.” Not to be confused with: Being listed first.
- Source-only
- The firm or its domain appears in source material but is not substantively surfaced in the answer. Why it matters: Prevents inflating visibility with background citations.
- Negative / cautionary
- The firm appears in a materially adverse or cautionary context. Why it matters: Not every appearance is favourable.
- Prompt
- The exact text submitted to an AI system in an observation. Why it matters: A prompt is part of the experimental condition and is never rewritten after collection.
- Prompt archetype
- A category of prompt representing a type of client question, such as direct recommendation, comparison or problem-based. Why it matters: Allows results to be compared across consistent question types.
- Prompt intent
- The kind of response the user's question appears to request. Why it matters: Must be stored separately from what the answer actually did. Not to be confused with: Response treatment.
- Response treatment
- How an AI answer actually treats a firm: mentioned, suggested, recommended, source-only or cautionary. Why it matters: The core classification behind FirmRanker measures. Not to be confused with: Prompt intent.
- Study
- A defined research programme with a specific question, population and methodology. Why it matters: Findings belong to studies, within their stated scope.
- Experiment
- A particular combination of prompt, market, practice area and AI platform/model. Why it matters: Groups observations made under the same conditions.
- Observation
- The saved AI response from a specific run. Why it matters: The unit of evidence; preserved unchanged. Not to be confused with: A ranking.
- Entity
- A distinct real-world organisation or person, such as a law firm or lawyer. Why it matters: Measurement counts entities, not text strings.
- Entity resolution
- Matching the different names used for a firm to one canonical entity while preserving the original text. Why it matters: Prevents splitting one firm's visibility or merging different firms.
- Canonical firm
- The single reference record a firm's name variants are resolved to. Why it matters: Provides a stable unit for analysis.
- Appearance frequency
- How often a firm appears across repeated relevant observations. Why it matters: Requires repetition; one answer cannot produce it.
- Recommendation frequency
- How often a firm receives recommendation treatment across repeated relevant observations. Why it matters: Separates being named from being put forward.
- Prompt coverage
- The range of client question types for which a firm appears. Why it matters: A firm visible for one question may be absent for others.
- Model coverage
- The AI systems in which a firm appears. Why it matters: Systems differ; visibility in one does not imply visibility in others.
- Recommendation stability
- How consistent the recommended set is across repeated runs of the same prompt. Why it matters: Shows whether a result is typical or incidental.
- Cross-model consensus
- The degree to which different AI systems surface the same firms for the same need. Why it matters: Indicates whether visibility is system-specific.
- Source landscape
- The pages, domains and source types retrieved or cited around AI answers. Why it matters: Describes the information environment — not what caused an answer.
- Retrieved source
- A source returned during provider search or retrieval. Why it matters: Retrieval is not citation. Not to be confused with: Cited source.
- Cited source
- A source explicitly cited or linked in an AI answer. Why it matters: Citation is not causation. Not to be confused with: A cause of the recommendation.
- Resolved source
- The actual publisher page reached after resolving any provider redirect. Why it matters: Some providers route citations through redirect URLs.
- AEO
- Answer engine optimisation — efforts to influence how AI systems answer questions. Why it matters: FirmRanker measures; it does not assume which tactics work.
- GEO
- Generative engine optimisation — a related term for influencing generative AI outputs. Why it matters: Used inconsistently across the industry; FirmRanker does not attempt to settle it.