Signals That Appear Before the Resignation

By the time turnover and absence move, the condition has been present for a long time. Earlier signals exist and are observable without any instrument.

At team level

Leave not being taken, or taken in fragments and interrupted.

Rising short-term absence frequency, which precedes long-term absence.

Meetings that used to have discussion becoming silent.

Ideas and suggestions stopping. People who have concluded nothing changes stop proposing things, and this is an efficacy signal.

Quality complaints rising while output holds, which indicates people are keeping volume up by cutting corners.

Handover and documentation degrading, because there is no time for anything not immediately due.

Vacancies staying open while the work does not reduce.

At individual level

Changes from that person's own baseline, not comparisons between people.

Withdrawal. Less contribution, camera off, fewer voluntary contacts.

Detachment. Someone who visibly cared appearing not to.

Uncharacteristic irritability or flatness.

Working at unusual hours, or working while on leave.

Uncharacteristic errors.

Not taking breaks.

A drop in the quality of their thinking rather than their effort, which frequently comes first.

The important limits

This list can be misused, so the boundaries matter.

These signals are not diagnostic. They indicate that something may be worth asking about, and nothing more.

Do not build surveillance on them. Monitoring individuals' hours, messages or activity to infer wellbeing is intrusive, damaging to trust, and legally exposed in several jurisdictions.

Do not raise them as evidence. "I have noticed your camera is off more" is an accusation. "How are things going?" is a question.

Life outside work explains many of them, and it is not the employer's business unless the person chooses to share it.

Never infer a clinical condition. That is not a manager's judgement to make, in either direction.

What noticing entitles you to do

Ask, once, without pressing. Openly, privately, with time available.

Offer an adjustment you can actually make.

Signpost support clearly.

Leave the door open if the answer is that everything is fine, which it frequently will be the first time.

Follow up later, which is what distinguishes concern from a box-ticking exercise.

Aggregate signals worth tracking

At team level, tracked over time and not attributed to individuals.

Leave utilisation.

Out-of-hours activity, in aggregate.

Short-term absence frequency.

Vacancy duration.

Time from concern raised to response.

All five are already in systems you have, none requires individual monitoring, and together they give an earlier picture than any survey cycle.

Where the manager cannot see

Distributed teams hide most individual signals. The compensation is more frequent direct contact, not more monitoring.

Large spans of control. A manager with thirty reports cannot notice changes in any of them, which is a structural finding rather than a personal failing and belongs in the design of the organisation.

Aggregate rather than individual monitoring

The distinction that keeps this practice legitimate.

Aggregate, team-level, over time: leave utilisation, out-of-hours activity, absence frequency, vacancy duration, meeting load. Legitimate, useful, and it identifies where to look.

Individual monitoring of hours, messages, activity or location to infer someone's state: intrusive, damaging to trust, and legally exposed in several jurisdictions.

The test: would you be comfortable telling the whole workforce exactly what you collect and why? If not, do not collect it.

Say what you measure, in advance, in plain terms.

Individual attention comes from a manager who knows their people, not from a dashboard. Where spans of control make that impossible, the answer is smaller spans rather than better surveillance.

Baselines, not comparisons

The signals only mean something against the person's own normal.

Someone quiet who has always been quiet is not a signal.

Someone talkative who has gone quiet is.

Comparing people against each other produces false positives and unfair assumptions, frequently along personality and cultural lines.

Which requires knowing people, which requires regular contact, which requires a manageable span of control.

Where a manager does not know their reports well enough to spot a change from baseline, the structural problem is the span rather than the manager's attentiveness.