Trust & transparency
How we build our data
Catalitium combines indexed catalog listings, location-based salary references, and original research. This page explains how numbers on jobs, salary tools, and reports are produced.
Last reviewed:
Salary estimates
Pay figures come from three sources, in this order of preference:
- The range the employer published on the listing itself. Shown as disclosed pay, never relabelled as an estimate.
- Role × location bands we compute from disclosed ranges across the catalog. We group postings by role family, place and currency, trim the top and bottom 5% of each group, and publish the 25th/50th/75th percentiles only when a group holds at least 8 postings. Currencies are never blended. Current coverage: 99 published bands built from 3,974 employer-disclosed postings across 26 role families.
- Reference medians — a hand-maintained location and title table used only where the catalog has too few disclosed ranges to compute a band. These are references, not measurements from our data.
Where a band is shown we state how many postings it is based on. Below 8 postings we publish no band rather than imply a precision the sample cannot support. The posting count is the confidence signal — there is no separate confidence-score UI on Salary Tools, and we do not blend a score out of signals we cannot validate.
Known coverage limit: disclosed pay ranges are concentrated in the United States, the United Kingdom and Ireland, because pay transparency laws there require employers to publish ranges. We currently have no computed bands for Switzerland, Germany or Austria — DACH figures on Salary Tools are reference medians, not measurements from our catalog. Catalog pulse: DACH snapshot 2026. Category tables use CHF for DACH comparability; regional cards use local currency.
Catalog job selection
We focus on tech and AI/ML roles with signal over noise: salary hints where available, structured locations, and explainable filter heuristics for search queries. Catalog freshness depends on ingest cadence and listing dates — we do not claim an automated quality-based ranking of sources.
Ghost-job detection
Roles posted more than 30 days ago are flagged “May be filled” so candidates do not waste applications on likely-stale listings. This is a heuristic, not a guarantee—always confirm status with the employer. Live catalog share: Ghost jobs in tech 2026.
Market research reports
Reports are produced independently from catalog listings and aggregated benchmarks, not press-release recycling. Each report documents sources and limitations in its methodology section. Premium unlocks full report bodies; where tables support it, on-page CSV export is available.
Disclaimer
All figures are estimates for planning and negotiation—verify compensation and role status directly with employers. Catalog coverage is not a continuous live ATS dump. Questions? Contact us.