Published October 27, 2026.
Agree on the definition first
Regrettable attrition sounds obvious until two leaders apply it to the same departure. One counts anyone the manager would rehire; another counts only high performers in hard-to-fill roles. Write the definition before the quarter begins, including who applies it and when.
A workable definition names three inputs: role criticality, documented performance evidence, and replacement difficulty. It also names what does not count, such as a departure during a documented restructuring. Publish the rule so the review is not a negotiation after each exit.
Use evidence available at departure
The review should rely on records that existed before the resignation: the most recent performance documentation, the role's vacancy history, the staffing plan, and the skills inventory. A rating created after the employee announces a departure is hindsight, not evidence.
Where the record is thin, mark the case as unclassified rather than guessing. Unclassified cases are useful because they reveal where the organization lacks basic workforce evidence. Report them as a separate count so they do not inflate either the regrettable or non-regrettable total.
Separate role criticality from performance rating
A solid performer in a role with no backup can be more damaging to lose than a high performer in a deep bench. Score criticality on replacement time, customer or production dependency, and unique system knowledge. Score performance from the documented review record.
Keep the two scores visible beside each other instead of collapsing them into one number. A departure can be high criticality and average performance, which calls for a knowledge-transfer response rather than a compensation review. The matrix prevents the conversation from becoming a general judgment about the person.
Build the review packet
For each departure, assemble the resignation date, last day, role, tenure, manager group, documented performance summary, vacancy history, replacement plan, and exit reason code. Add the employee's stated reason and whether a counteroffer was made, without editorializing.
The packet should fit on one page plus links to source records. Reviewers need to reach the underlying review or requisition, not a paraphrase. Assign a preparer and a reviewer, and record the date the packet was checked against the source systems.
Run the review without naming individuals
Aggregate the packets by role family, tenure band, and manager group. Share counts, rates, and themes rather than a list of names. In small teams, suppress cells below a minimum size or combine categories so a reader cannot identify a single departure.
Invite the operations leader and the finance partner who can speak to replacement cost and service impact. Keep the discussion focused on patterns the organization can change: scheduling, span of control, onboarding, or pay review timing. Do not use the meeting to reopen an individual's performance rating.
Test the definition against last quarter
Apply the new definition to the previous quarter before publishing the current result. If the count changes substantially, explain whether the change comes from the rule or from the workforce. Readers should be able to compare periods without quietly switching definitions.
Record any definition change with an effective date and a reason. Keep the prior classification available for audit. A trend line built on three different definitions is not a trend; it is three separate snapshots wearing the same label.
Publish actions and owners
Close the review with no more than five actions. Each action needs an owner, a due date, a measure, and a review point. Examples include adding a backup for a single-point role, moving a pay review earlier, or changing a schedule posting rule.
Report progress at the next review using the same counts. If an action did not move the measure, say so and revise it rather than carrying it forward unchanged. The purpose of the review is to reduce avoidable loss, not to produce a polished quarterly deck.