How it works
Public signals in, decisions out
Four steps, and no point at which you are asked to interpret a raw feed.
- 01
Confirm your identity
Tell us which clinic, which locations, and which public profiles are actually yours. Misattributed profiles are the single biggest source of wrong reputation data, so this is a confirmation step, not a guess.
- 02
Choose markets and competitors
Set the patient-market languages you care about and name the competitors you actually compete with. Your plan sets the limits.
- 03
Baseline
ClinicRadar collects your public history, classifies it, computes the first Trust Health Score with its Coverage Confidence, and produces your first findings.
- 04
Monitor and act
From then on it watches continuously, raises risks as they emerge, refreshes competitor and AI-visibility comparisons, and keeps a prioritized action list current.
Collection
What is read
Only public sources, and only the ones that matter to your vertical and your markets.
- Public review platforms relevant to your vertical and your patient markets.
- Public social posts and mentions that reference the clinic.
- Public search results and the sources that rank for your treatments.
- Public answer-engine responses to prompts you define.
Analysis
Where the AI is, and where it is not
A language model classifies each public item into themes from a reviewed clinic taxonomy, detects sentiment and risk signals, and writes plain-language summaries. That is the whole of its role.
The Trust Health Score is computed afterwards by a deterministic, versioned formula over those classifications. The model never produces the number, never adjusts it, and never overrides the coverage rules. This is why the same period always yields the same score, and why the methodology can be published in full.
When the available public evidence is too thin to support a comparable number, Coverage Confidence reports Insufficient and the numeric score is suppressed rather than shown with a footnote.
Guarantees
What holds at every step
- Nothing behind a login is collected.
- No patient data is ingested, stored, or inferred.
- Raw provider snapshots are stored privately, scoped to your workspace, and never publicly addressable.
- Every derived finding retains a pointer to the public item and the date it was collected.