Governance
Who changed the weights: the audit trail that makes a shortlist defensible
Published Reviewed by Hab Business Solutions
Every hiring dispute eventually becomes a question about configuration. If you cannot answer it in writing, the shortlist is an opinion.
Short answer
What should an AI hiring tool record about its own configuration?
The question always arrives later
Nobody asks how a shortlist was built on the day it is built. The question comes three months later, from a candidate, a regulator, a works council or an internal review, and by then the person who set the weights has moved teams.
Under the DPDP Act in India and the GDPR in Europe, a candidate can ask why they were evaluated the way they were. A recollection is not an answer.
Configuration is a decision, so log it like one
Weights, mandatory and optional requirements, boolean skill groups, alias matching and the workspace terminology catalogue all change the ranking. Each one is an editorial choice about what the company values in the role.
The audit trail records who changed each of them, when, and the reason given. That turns a configuration screen into a documented policy rather than a set of dials someone moved.
Decisions carry owners and reasons
Four explicit states exist: shortlist, select, reject and clear. Each carries a named owner and a written reason. Ranking reports export as PDF, standalone HTML, CSV and JSON, so the record leaves the tool intact.
Determinism ties it together. Because the same inputs reproduce the same score, a decision from three months ago can be re-run today and it will match. That is the property an auditor asks for and most tools cannot supply.
Frequently asked
Does the audit trail slow recruiters down?
It adds a reason field to a decision that was already being made. The cost is seconds; the alternative is reconstructing a hiring process from memory.
Can we export the record?
Yes, as PDF, standalone HTML, CSV or JSON.
Want this running on your next role?
Bring one job description and its applicants. You will see the ranked pool and the evidence behind it before the call ends.
No retainers to start · Pilot-first · Human-in-the-loop governance