Anomalies & Watchlist
What deserves human attention right now — investigation queues, not more distribution charts.
Filters (above) apply only to 5.2 Battlegrounds (function excluded — breadth is measured across functions, so filtering by function would collapse it). 5.1 Compensation outliers, 5.3 High signal-density postings, and 5.4 Alerts stay full-snapshot: the first two need per-job titles the row-level index doesn't ship, and Alerts are warehouse-wide data-quality diagnostics, not a market question. Summary above always describes the full snapshot.
5.1 Compensation outliers
Stage 2 — Statistical outliers (verify)
| Company | Title | Band | Midpoint USD | Fence | × fence | vs list max | Action |
|---|
Stage 1 — Pre-normalization artifacts
| Company | Title | Min | Max | Currency | Period | Reason | Action |
|---|
Stage 2: Tukey fence (1.5×IQR) within seniority band on comp_normalized midpoints;
list capped by dollar excess over the fence, ordered here by fence multiple. Bar = midpoint vs
the largest shown; hairline = that band's fence. Stage 1: disclosed roles excluded from
Section 3 (unprocessable currency/period).
5.2 Cross-function battleground companies
| Company | Tier | Roles | Functions | Breadth | Function coverage |
|---|
Hiring Breadth Index = distinct functions / 8. Floor ≥0.5 (4+ functions). Catalog-complete — no enrichment required.
5.3 High signal-density postings
Enriched subset only. Density = count of distinct controlled behavioral signals (max 6). Not a quality judgment. Ties at the 6-signal maximum are listed alphabetically by company — a sample of the tied set, not a ranking within it.