Radar API / Developer guide

Coverage & Data Quality

Understand what Radar observed, normalized and verified.

Read as Markdown

Scope and source categories

Radar processes listings from job boards, staffing firms, marketplaces, consulting companies and employer sites where supported and permitted. This describes source categories, not complete coverage of every site or the entire market. A discovered source is not necessarily actively contributing inventory.

Normalization

Processing maps source records into canonical opportunities with occupational taxonomy, engagement classification, commercial evidence, compensation and location. Source evidence can be incomplete or ambiguous. Nulls and unknown classifications preserve that uncertainty instead of turning it into a positive claim.

Freshness and removals

Persisted records are not all current, live or commercially eligible. Use liveness_status and last_verified_live_at together. A recent fetch is not proof of recent verification. Sources may block access or remove listings; current availability can change after delivery. Public reads do not trigger a refresh.

Deduplication and provenance

Canonical IDs connect recognized duplicates. The sources collection records available manifestations; preview redaction is intentional. No perfect accuracy or deduplication percentage is claimed. Source text and application destinations should be treated as untrusted input.

Measured evidence

The inventory panel reads an allowlisted aggregate from the existing completed processing audit. It shows the audit time and flags evidence older than 26 hours. All persisted records and records marked live are distinct counts; neither is a claim about current commercially eligible inventory. The record inspector exposes timestamps from the exact preview response. No market-wide coverage percentage, corpus-wide freshness claim or accuracy benchmark is published without a defined scope, denominator and dated measurement. See System Status for monitored service evidence, which is separate from dataset quality.

Report a gap

Use Support to report a missing source or a data issue, including a public source URL and a concise explanation. Submission requests review; it does not start ingestion or guarantee source onboarding. Access permissions and source rights still apply.

Measured inventory

As of: 2026-09-13T18:13:34.751534+00:00. This audit is more than 26 hours old; these are historical observations.

All persisted canonical opportunities, including unavailable or ineligible records
50,223
All persisted source records, including duplicates of canonical opportunities
51,107
Canonical opportunities with at least one source record marked live
31,649
Source records marked dead at the audit
4,745
Source records marked live at the audit
32,148
Source records whose liveness check was blocked
10,098

Scope: all inventory included in that audit. Counts are absolute, not market shares or accuracy percentages. Live inventory is not necessarily technology-classified or commercially eligible. Source verification time differs from retrieval time. Persisted inventory at the existing processing quality audit. Live means at least one source record marked live; this is not the current commercially eligible API result count.