Define the scope first
The brand, public identity, market context, and buyer questions are configured before results are collected.
Evidence first. Scores second.
How AI Brand Lens Measures AI Visibility
AI Brand Lens measures whether AI systems mention, position, cite, or omit a brand across the buyer questions that matter—while preserving the provider evidence behind every result.
What AI visibility means
A timestamped view of whether a brand appears in AI answers.
AI visibility describes whether and how a brand appears when an AI provider answers a relevant buyer question. A useful measurement looks beyond a single mention to the brand’s position, competitors in the answer, returned citations, and differences between providers.
AI Brand Lens treats each result as an observation—not a permanent ranking. The AI Visibility platform connects those observations to the original answers so teams can review what happened and compare change over time.
Questions and providers
Measurement starts with the questions buyers actually ask.
Monitoring questions define the scope of a visibility result. A narrow local-service question and a broad category question test different buying situations, so question selection must remain visible when results are interpreted.
Keep questions comparable
Configured providers receive the same substantive question. AIBL does not quietly rewrite the test to improve one provider’s result.
Compare major providers
Paid monitoring supports ChatGPT, Gemini, Claude, and xAI where included by the plan and currently available. The free snapshot compares ChatGPT and Gemini.
From question to measurement
A six-step, evidence-first workflow.
Define the target
A scan starts with a specific brand or entity, its public identity, relevant market context, and buyer questions selected before results are collected.
Run comparable questions
Configured providers receive the same substantive question so differences can be compared without quietly changing the test between systems.
Preserve provider evidence
The original response, provider and model context, timestamp, citations, completion status, and evaluation output are stored before summary conclusions are presented.
Evaluate observable signals
A separate evaluation identifies brand mentions, recommendation position, competitors, citations, and other structured facts. It does not rewrite the provider’s answer.
Aggregate conservatively
Completed response-level observations become question, scan, brand, competitor, and trend metrics. Missing or failed responses are distinguished from genuine omissions.
Compare over time
Recurring scans use a stable measurement approach so teams can inspect movement while retaining the evidence behind each historical result.
How the AI Visibility Score is calculated
Completed responses define the denominator.
Each completed provider response scores 100 when the tracked brand is mentioned and 0 when it is not. The AI Visibility Score is the average of those completed response scores. Failed or unusable responses are reported separately and do not count as brand omissions.
Position, competitors, citations, provider agreement, and confidence add context, but they do not directly change this mention-based visibility score.
Visibility
At response level, whether the tracked brand appeared in a completed answer. Aggregate visibility summarizes those completed observations.
Position
Where the brand appeared in a recommendation or ranked list when the provider returned an order that can be evaluated.
Competitors
Organizations surfaced as recommended alternatives or comparable providers, normalized while preserving provider-level evidence.
Citations
Sources returned with an answer. A citation is evidence for that response—not a permanent endorsement or guaranteed recommendation.
Agreement
The percentage of question groups where all completed providers agree on whether the tracked brand appeared.
Confidence
A separate conservative composite of response completion, provider agreement, and mention signal strength—not a grade for the brand.
Illustrative example
How one answer becomes structured observations.
This fictional example shows the shape of an evaluation. It is not real provider or customer data.
Monitoring question
“Which patio design companies should I consider near Raleigh?”
Short fictional response
“Consider Oak & Stone Outdoors first, followed by Northstar Patio Co. and Greenline Patios. Northstar is known for custom pergolas.”
- Tracked brand mentioned
- Yes · response score 100
- Recommendation position
- 2
- Competitors observed
- 2
- Citations returned
- None
This response contributes a 100 to the visibility average because the tracked brand appeared. Its position and competitor observations remain separate evidence for deeper analysis.
Evidence and interpretation
Provider output, AIBL evaluation, and owned evidence stay distinct.
Keeping these layers separate makes a result inspectable and prevents an interpretation from being presented as though the provider said it.
01 · Provider output
The original answer, question, provider/model context, completion state, timestamp, and available citations are preserved before aggregation.
02 · AIBL evaluation
A separate structured evaluation identifies mentions, rank, competitors, citations, and evaluation confidence without rewriting the answer.
03 · Owned evidence
Public pages, schema, sitemap, robots rules, and optional llms.txt guidance are assessed as supporting proof—not as guarantees of a recommendation.
Learn how to strengthen that public proof in AI Readiness or explore the AI visibility guides.
Variability and limitations
Why AI visibility changes between providers and runs.
Point-in-time observation
A scan records what configured providers returned at a particular time. It does not establish a permanent or universal ranking.
Provider independence
AI Brand Lens does not control provider models, retrieval systems, indexes, citations, or availability. Provider behavior can change without notice.
Answers can vary
AI systems are nondeterministic, and providers use different models, retrieval systems, indexes, and source-selection methods. Disagreement is evidence rather than an error to hide.
Recurring measurement adds context
Repeated scans make it possible to compare observations over time. Provider, model, question, and methodology context still matter before movement is treated as meaningful change.
Evaluation can be wrong
Structured evaluation reduces manual inconsistency but can still misidentify brands, positions, competitors, or claims. Important conclusions should be checked against the preserved response.
Methodology evolves
Provider coverage, scoring, evaluation, and confidence methods may change as the product and AI ecosystem mature. Material methodology changes should be versioned and communicated.
Customer inputs
Brand details, questions, competitors, and configuration supplied by the workspace.
Provider evidence
Responses, citations, completion information, provider/model context, and timestamps created during scans.
Public evidence
Publicly accessible pages and signals evaluated for brand identity, readiness, citations, or supporting proof.
For more answers about provider coverage, confidence, data, and plans, read the FAQ. For personal information, vendors, retention, choices, and requests, read the Privacy Policy. Product outputs are informational and should be verified before important decisions.