Network Benchmarks Methodology
How Clinelix builds privacy-first cross-clinic benchmarks.
Every percentile your dashboard shows is aggregated from real, consented practice billings — but with strict k-anonymity, no patient health information, and no clinic identifiers ever leaving the aggregation service.
k-anonymity
Cohorts with fewer than 5 distinct clinics are suppressed — your dashboard will simply show "Insufficient sample" rather than a value that could be reverse-engineered.
What we publish
p25, p50 (median), p75, and mean per dimension. Never raw values, never individual clinic results. Your own clinic is excluded from its own cohort.
No PHI
Patient health information is filtered out before billings ever reach the aggregation pipeline. Aggregations run server-side and produce only summary statistics.
Privacy controls in detail
- Cohorts with fewer than 5 distinct clinics are suppressed (k-anonymity).
- No clinic identifiers ever leave the aggregation service.
- Outputs are p25/p50/p75/mean only — never raw values from any single clinic.
- The requesting clinic is excluded from its own cohort to prevent self-leakage.
- All cohort queries are server-side; clients can only request their own pivot.
Lookback window
365 days
Refresh cadence
Daily incremental rollup; full recompute weekly.
Methodology version
2026-Q2
Dimensions benchmarked: Insurance AR health · Fee guide compliance · Procedure profitability · Chair utilization · Cash flow predictability