Retention rules

Settings → Organization → Compliance → Retention flags candidacies that no longer have a lawful basis to keep, and can delete them. Every match lands in a review queue first, unless you turn on auto-execute for that rule. The same page sets how long saved searches and agent chats are kept. Only organization owners and admins can change it. The flagging itself runs once a day across every workspace in the organization.

​

The four rules

Each rule has its own Enabled toggle and a period in days (1 to 3,650). All four ship off by default: nothing is flagged until you turn at least one on. The rules can flag rejected, withdrawn and hired applications. Hired and offer-stage candidates are still skipped while Never flag hired or offer-stage candidates is on.

​

Enabled vs auto-execute

Enabled turns a rule on: matching candidacies get flagged and land in the review queue. A separate auto-execute toggle, off by default, skips the review step: once the review window passes, a flagged candidacy proceeds straight to deletion with nobody clicking anything. If a candidacy matches several rules and any one of them has auto-execute on, the whole flag proceeds automatically.

​

The review queue

Flagged candidacies wait here before anything happens, whether or not auto-execute is on. For each one you can:

• Send consent email: manually resend the retention notice, if the automatic one didn’t go out.
• Cancel deletion…: extends retention (365 days by default, 1 to 3,650) so tomorrow’s check doesn’t re-flag it. Requires a reason, which goes on the audit trail.
• Delete now…: runs the deletion immediately instead of waiting. This can’t be undone.

​

Exclusions

Two toggles, both on by default, protecting the candidate rather than requiring you to opt in:

• Never flag hired or offer-stage candidates: employment records have their own retention duty.
• Never flag candidates in the talent pool: adding someone to the talent pool is treated as a deliberate decision to keep the record.

​

Windows

• Review window (hours, default 72, up to 720): how long a flag waits before an auto-executing rule proceeds. Rules without auto-execute ignore this. Nothing fires on its own for them.
• Email candidates when flagged (on by default): sends the candidate a heads-up with a link to renew consent, on the same window as the review. Not legally required, but a courtesy.

​

The GDPR Art. 17 grace window

Separate from the four rules above, and always active: when a candidate lodges a right-to-erasure request, Staffer holds it for this many days before running the deletion. Default 30 days, and you can only shorten it, never extend it beyond 30. See Candidate data requests for what happens to a request while it waits, and afterward.

​

Data lifetimes

Saved searches and agent chats have a lifetime that’s always on. Once something is older than its period, a daily check deletes it, with no review queue. You can change each period, from 1 to 3,650 days, but you can’t turn the deletion off. Deleted searches and chats can’t be restored. Deleting a search doesn’t remove anyone you shortlisted or promoted from it. Only organization owners and admins can change these periods.

​

What deletion does

Deleting a candidacy, whether by auto-execute or by clicking Delete now…, runs Staffer’s erasure cascade: the application is removed, and if it was the person’s only remaining application anywhere in your organization, their whole profile is anonymized rather than deleted outright, so the historical record survives without identifying them. See Candidate data requests for exactly what’s deleted and what’s kept.

​

Interview recordings and transcripts

Interview recordings, transcripts and assessments are deleted together with the application, for both AI and human interviews. That happens when a retention rule removes the application, or when someone removes the candidate from the job. Our interview and notetaker providers delete their copies within a day. Moving a candidate to another job keeps their interview until the application they moved into is removed. See AI interviews and Interviews and scheduling.

​

Common questions