ComplyLens — LL144 Guide

LL144 Audit Data Requirements: What to Export from Your ATS

Before you commission an LL144 bias audit, you need the right data. Here is exactly what the audit needs and how to export it from common ATS platforms.

The minimal dataset

candidate_id, category, selected
cand_001, male, 1
cand_002, female, 0
...

candidate_id: unique applicant identifier (de-identified is fine).
category: protected category — sex, ethnicity, or their intersection (e.g. "female_asian").
selected: outcome — 1 = selected/advanced, 0 = not (or a continuous score, see below).

Score-based tools

If your tool scores candidates on a continuous scale instead of a binary outcome, export the score column instead — the audit computes median-threshold scoring rates and score-based impact ratios.

candidate_id, category, score
cand_001, male, 87.5
cand_002, female, 63.2

Exporting from common ATS platforms

Platform	Export path	Notes
Greenhouse	Reports → Hiring reports → Candidate demographics export	Include all candidates screened by the tool, not just hires
Lever	Reports → Diversity export (or custom CSV)	Filter by the position(s) and time window the tool was used
Workday	Report Writer → custom candidate report	May require report-writer permission; ask HRIS admin
CSV (any)	Direct export	Ensure consistent category labels across rows

Data quality checklist

Include every candidate the tool evaluated, not only finalists — the audit measures the tool's full screening funnel.
Cover the full audit period (typically 12 months or the tool's deployment period).
Use consistent category labels (normalize "M", "Male", "male" → "male").
De-identify names/emails if you prefer — only the category and outcome matter.
Aim for 100+ records per category for statistically meaningful impact ratios (fewer is still analysable but less reliable).

Next step

Once your export is ready, the statistical analysis takes 72 hours. Toolkits from $29, analysis from $299.

Start with the Readiness Toolkit — $299