AI Visibility · Monitoring Methodology
How We Measure AI Visibility
The methodology behind every score on the Leaderboard: what we ask, what we capture, and how it becomes a number you can act on.
Why We Score
AI answers are the new first opinion.
Patients and HCPs increasingly ask an AI system before they search, click, or call. Whatever that answer says, accurate or not, with your brand or without it, shapes the decision that follows. The AI doctor is already in the room: people arrive at appointments having consulted AI about symptoms, treatments, and brands, and clinicians use the same tools to summarize evidence.
The AI Visibility score measures how a brand shows up at that moment: whether it is present, cited, accurately represented, and consistently described across the AI engines people actually use. It turns an unmonitored risk into something a regulated brand can see, track, and improve.
What goes into every monitoring run
Real questions, not keywords
Every brand and condition is monitored against a curated universe of real questions, grounded in actual search demand and organized by audience (patients, HCPs, caregivers) and intent. Branded questions cover awareness, evaluation, access and affordability, usage, and side effects. Unbranded disease-state questions cover awareness, prevention, symptoms, diagnosis, treatment options, and living with a condition.
The engines people actually use
Each question is asked across the leading AI answer engines: Google AI Overviews, Google AI Mode, ChatGPT, Claude, and Perplexity. The roster expands as new assistants earn real usage. Every full answer and every cited source is captured, not just a pass/fail check.
On a schedule, over time
AI answers vary by platform, prompt, time, and retrieval behavior, so a single screenshot is not a benchmark. Monitoring runs on a recurring cadence, and every run is stored as a dated snapshot, building a longitudinal record of how each engine's answers change.
The Signals
What we measure in every answer
For each question, on each engine, on each run, the full answer and its cited sources are captured and evaluated across six dimensions. Together they describe not just whether a brand appears, but how it is trusted, positioned, and represented.
Inclusion
Does the brand appear in the answer at all? Presence is tracked per engine, per question, on every run.
Citations and source mix
Is the brand's own site among the cited sources, and what mix of sources does the engine trust? Brand-owned, government, and third-party citations are identified for every answer.
Prominence and share of voice
Is the brand a primary source or a passing mention? For disease-state questions, every brand named in the answer is captured in order of appearance, producing a competitive share of voice.
Accuracy and message fidelity
Do the engines agree on the verifiable facts, and does the answer reflect the evidence? Factual consistency is evaluated across engines. For regulated brands, visibility and accuracy must be measured together.
Tone
Health answers should be consistent, cautious where caution is warranted, and free of alarmist or overpromising language. Sentiment is evaluated across engines alongside the facts.
Change over time
Each run is compared to the last: citations gained and lost, brands added and dropped, and how every signal is trending.
The Score
From thousands of answers to one number.
Each question receives a score from 0 to 100, computed as a weighted blend of the signals above, with the heaviest weight on whether engines represent the brand accurately and consistently, not merely on whether it gets mentioned. Question scores roll up through topic clusters to a single headline score for each brand and condition, alongside plain-language mention and citation rates per engine.
Because every run is a dated snapshot, each score carries its trend: month-over-month movement, longer-horizon deltas, and a momentum measure that shows whether visibility is being earned or eroded.
We publish the dimensions, the scale, and the bands, but not the exact weights. The weighting evolves as the engines evolve, and keeping it private keeps the score honest: it has to be earned with genuinely accurate, citable content, not gamed.
Score bands
Built for What’s Next
The answer layer is spreading beyond the chat box.
Voice
Ask a smart speaker or a car about symptoms, dosing, or a brand and one answer comes back: no list of links, no second chance to be found.
Agents
AI agents increasingly research, compare, and shortlist on a person's behalf. Brands that are consistently cited and accurately represented are the ones agents carry forward.
Robotics and ambient AI
As AI moves into pharmacies, clinics, and the home, the same answer layer travels with it. The work of being the trusted, well-cited source compounds across every new surface.
What is truly important doesn’t change: be the accurate, well-cited, consistently represented source, and every new surface inherits it.
See how brands score today.
Explore the LeaderboardConnect With Us
Know exactly how AI sees your brand.
Whether the opportunity begins with visibility or nonprescription access, we’ll help you find the clearest next move across Search, AI and ACNU.
contact@idea-evolver.com