Data quality scorecard

Daily versioned baseline. Cite this in every research release.

As of

6 September 2026 (v1)

Composite score

8.1/10

Six audit dimensions

Source SLA

9.1/10

Collector liveness and source freshness

Measuredtoday

Engineering

7.5/10

Pipeline reliability, observability

Partialtoday

Quality

8.5/10

Coverage, accuracy, dedup

Measuredtoday

Research substrate

6.3/10

Papers, citations, methodology

Assertedtoday

UI/UX surface

8.8/10

Public discoverability of substrate

Measuredtoday

Security

9.8/10

Access control, secrets, GDPR

Measuredtoday

Volume

52,210

Jobs total

4,576

Jobs last 30d

54,782

Intelligence corpus

5,312

Intel last 30d

121,322

Research papers

948

Cross-source verified

547

News events

2,050

Salary benchmark rows

Coverage — full corpus

Jobs · salary

27.7%

Jobs · description

81.4%

Jobs · country

89.5%

Jobs · H3 (geocoded)

64.1%

Intel · geom

88.1%

Intel · salary

27.6%

Intel · Tier A (≥80)

45.6%

Intel · Tier B (60-79)

39.9%

Coverage — research-grade subset(42,884 rows · API + ATS only)

Excludes Kaggle datasets, US DoL H1B records, newsletter snippets, and manual drops — sources that arrive thin by design and have no canonical URL to fetch from. The composite quality score (D3) is computed against this surface so the rating reflects actual pipeline operations, not data origin.

Research · description

98.1%

Research · geom

98.7%

Research · salary

32.7%

Research · Tier A (≥80)

54.6%

Research · Tier B (60-79)

39.8%

Source SLA breaches

b_lab · never_succeeded (monthly) · never_succeeded
job_source_text · never_succeeded (weekly) · never_succeeded
sbti · never_succeeded (monthly) · never_succeeded
uk_gender_pay_gap · breached (monthly) · 2485 hours since success · breached

Trend (last 30 days)

Date	Composite	Jobs	Intel	Papers
2026-09-06	8.1	52,210	54,782	121,322
2026-09-05	8.1	51,942	54,671	120,122
2026-09-04	8.1	51,658	54,403	119,434
2026-09-03	8.1	51,446	54,119	117,749
2026-09-02	8.1	51,223	53,907	116,560

Methodology: scripts/data-quality-snapshot.py runs daily, queries volume + coverage + freshness across jobs, intel_jobs, intel_papers, intel_news_events and intel_salary_benchmarks, then writes the result into the data_quality_snapshot table with snapshot_date + snapshot_version as the unique key. The composite score is a weighted average of D1-D6 (D3 quality 25%, D4 substrate 25%, D5 UI 20%, D1 source SLA 15%, D2 engineering 10%, D6 security 5%). Source code is committed to the repository.