# Hiring assessment H-01

Folio Alder Bookkeeping | October 30, 2026

Decision: ranking stays paused. Applicants get a person-led review against written job criteria. No automatic shortlist or rejection. No applicant data enters ranking.

Affected: applicants for bookkeeping and accounting roles and staff reviewing applications. Scope: resume and job-description matching, without video, voice or personality analysis.

| Harm | Mechanism | Check | State |
| --- | --- | --- | --- |
| Qualified applicants missed | Gaps in employment history or unfamiliar job titles could lower a ranking. | Compare rankings with a blind, person-led review against the same written job criteria. Review every disagreement. | Not tested |
| Proxy discrimination | Location and school history could stand in for protected characteristics. | Obtain the input dictionary. Exclude zip codes from ranking inputs and test paired invented applications that differ only in location. | Vendor evidence missing |
| False precision | A recruiter could treat a rank as a fact and overlook the original application. | Keep ranking off. Require a written human decision reason tied to the job criteria, with no automatic rejection. | Manual route set |
| Applicant data exposed | Resumes could be retained, reused or shared beyond the agreed purpose. | Confirm deletion, training exclusions and subprocessors before any applicant data reaches the ranking feature. Use invented test records. | Terms incomplete |

Hiring lead runs and signs comparison findings. Office lead verifies settings. Employment counsel reviews the workflow and notice. Managing partner alone may release the hold once all gaps are resolved. Next check November 13, 2026. No test result or resume approval is recorded.