# Data quality audit

Overall status: **PASS**

## Validation checks

| Check | Status | Detail |
|---|---|---|
| row_count:bls_us_monthly.csv | PASS | 811 rows; minimum 700 |
| row_count:eurostat_vacancy_rates.csv | PASS | 557 rows; minimum 450 |
| row_count:eurostat_employment.csv | PASS | 81 rows; minimum 75 |
| row_count:eurostat_unemployment.csv | PASS | 19 rows; minimum 15 |
| row_count:ons_uk_vacancies.csv | PASS | 302 rows; minimum 250 |
| row_count:linkedin_freeze_vectors.csv | PASS | 9 rows; minimum 9 |
| row_count:linkedin_first_rung.csv | PASS | 5 rows; minimum 5 |
| row_count:linkedin_remote_mismatch.csv | PASS | 7 rows; minimum 7 |
| us_common_month | PASS | 2026-06-01 |
| sector_coverage | WARN | 17 country-sector cells |

## Headline audit values

US, June 2026 (December 2019 = 100): employment 104.7, openings 109.8, hires 89.9, quits 92.7.
European employment, 2026 Q2 (2019 Q4 = 100): Germany 100.6, France 106.8, Netherlands 107.8.

## Guardrails

- LinkedIn measures describe activity observed on the LinkedIn platform and are not population estimates.
- The hero trajectory compares February with May 2026; it is not a 2021–2026 country trail.
- Bubble area represents employment stock; x and y positions are year-on-year changes, not levels.
- Cross-country vacancy paths are normalized to each country’s own 2021–2026 peak because source measures differ.
- The analysis is descriptive. It does not identify AI, fear, policy or any single cause of the slowdown.