Measurement methodology
Last updated: 10 September 2026
EntityRise measures how a brand appears across a confirmed set of buyer questions tested across multiple AI providers, then evaluates the public competitive information environment around the brand.
Each audit is a point-in-time Snapshot. Snapshot-specific execution details - including the confirmed questions, configured providers, and observation coverage - live in Audit Integrity. The full methodology is maintained on this page. AI visibility itself is defined on our What is AI visibility page.
Business understanding
EntityRise researches the public presence of the business and creates a Business Profile for the customer to confirm.
Buyer topics
From the confirmed profile, EntityRise generates commercially relevant buyer topics. The customer reviews and confirms them.
Buyer-question set
EntityRise generates buyer questions from the confirmed context. The customer confirms the test set. The standard audit uses 20 questions: 17 discovery and 3 brand-aware.
Provider execution
Each confirmed question is tested independently against each configured AI provider.
Competitive and public-evidence research
Commercially substitutable competitors are selected from the confirmed Business Profile and researched across the same public-evidence dimensions as the brand.
Deterministic measurement and comparison
Visibility outcomes are scored from stored observations. Benchmark comparison direction is produced with consistent scoring rules.
Strategic priorities and report
EntityRise identifies the highest-priority conditions that need to improve and publishes the Snapshot report.
Every EntityRise audit is built on buyer questions generated from the confirmed Business Profile and confirmed buyer topics. The customer reviews and can edit those questions before the audit runs. The standard confirmed test set is exactly 20 buyer questions: 17 discovery questions and 3 brand-aware questions. Snapshot-specific counts are recorded in Audit Integrity.
Discovery questions do not name the brand. They measure organic discovery visibility: whether the brand appears when a buyer asks a category, comparison, or purchase-intent question without naming the company. Mention rate, active recommendation rate, provider visibility, topic visibility, and the AI Visibility Index are measured from these observations.
Brand-aware questions explicitly name the brand. They measure recognition when the brand is asked about directly, and are shown as recognition evidence. They are not organic discovery visibility.
Each confirmed question is tested independently against each configured AI provider. EntityRise does not rely on repeated runs of the same question to produce a measurement. The current standard configuration uses five providers - OpenAI, Claude, Gemini, Grok and Perplexity. Model versions used in a given Snapshot are recorded in Audit Integrity, not on this page.
For the standard current structure, 17 discovery questions × 5 providers = 85 discovery observations, and 3 brand-aware questions × 5 providers = 15 brand-aware observations - 100 question-provider observations in total. Provider count is configuration-driven; Audit Integrity records what a Snapshot actually executed.
The brand appears in the response.
The brand is explicitly presented as a recommended solution or choice.
The brand appears as an alternative option but does not meet the active-recommendation classification.
The observed ordinal position when an ordered recommendation exists.
Qualifying citation support exists under the production methodology.
How accurately the brand or product is described, when that component is determinate.
Observable rates - such as mention rate and active recommendation rate - are reported separately. The AI Visibility Index is a secondary composite summary. It does not replace the underlying measurements. It summarizes measured visibility across applicable discovery observations.
When published, the Index uses these canonical component weights:
The Index is calculated from scoreable unbranded discovery observations. Unavailable observations are excluded from denominators when a component is not determinate; they are not automatically treated as zero.
The Competitive Benchmark describes how the brand's public information environment compares with commercially substitutable competitors.
Benchmark competitors come from the confirmed Business Profile and the primary commercial decision the business serves. They are selected for commercial substitutability, not simply because an AI provider mentioned them. Up to three benchmark competitors are selected.
The client and the selected benchmark competitors are researched across the same public-evidence dimensions.
The six research dimensions are:
Whether public evidence associates the brand with the relevant commercial category.
Independent sources that discuss, review, or otherwise evidence the brand.
Public traces of the brand being recommended or compared as a choice.
Whether distinctive product or service attributes are associated with the brand in public evidence.
Whether claims can be checked against public proof.
Public reputation evidence relevant to buyer trust.
Each dimension receives a comparison state:
A Benchmark stronger result describes a difference in the public information environment. It does not mean that difference caused an AI recommendation, and it does not mean the competitor will always be recommended.
Technical and diagnostic readiness is separate from live AI visibility. Strong technical foundations do not guarantee recommendation visibility. Diagnostic evidence describes supporting conditions around the brand. Those conditions can make brand and product information easier to discover and verify. They are not presented as proven AI ranking factors.
The current diagnostic areas are:
01
Technical Accessibility
02
Brand & Product Understanding
03
Buyer Information & Proof
04
External Trust & Authority
Priority AI Visibility Fixes - also called Strategic Priorities - identify the highest-priority conditions that need to improve. EntityRise synthesizes measured AI visibility, the deterministic benchmark comparison, bounded evidence interpretation, and diagnostic evidence.
Confidence reflects the strength of the supporting record for a conclusion: evidence breadth, research coverage, benchmark determinacy, counter-evidence, and interpretation limits. Incomplete or unavailable observations are disclosed. Unsupported claims do not become strategic conclusions. Insufficient evidence remains insufficient rather than being forced directional.
See what this looks like in a real report - view the sample report →
Three rules protect the audit from unsupported conclusions:
01
Competitors are validated. Commercial substitutes must be supported by evidence and the confirmed business context.
02
Strategic priorities require evidence. Priority AI Visibility Fixes must be grounded in validated audit facts, benchmark evidence, or diagnostic conditions. Fixes describe WHAT needs to improve and WHY it matters.
03
No guaranteed rankings. AI outputs change across providers and time. EntityRise measures observed visibility; it does not promise placement.
AI is used for business understanding, classification, and report writing. Measurement, provider execution, evidence storage, competitor validation, scoring rules, and benchmark comparison direction remain controlled by EntityRise.
Results apply to the confirmed questions, providers, market, language, and period tested in that Snapshot.
Equivalent queries may produce different responses over time or across environments.
Provider APIs and search-grounded environments may differ from consumer-facing product interfaces.
Competitive or diagnostic differences do not establish why a provider produced a result.
Failed or unavailable observations are disclosed and excluded from denominators when not determinate. They are not automatically treated as brand absence.
Failure to retain public evidence does not prove that evidence does not exist.
First-party statements are treated differently from independent public evidence.
Diagnostic readiness and measured AI visibility are separate concepts.
Audits analyse public website content and public AI answers. Private information supplied by clients is handled separately and is never shared with competitors or used in other companies' audits. Details are in the Privacy Policy.