The signs someone is about to quit, and what to do that week

The short answer

People do not usually leave in a burst of drama. They withdraw in small increments over a few months, each one individually explainable, and then one day the message arrives and in hindsight all of it was obvious.

The difficulty is not that the signs are subtle. It is that every single one of them has an innocent explanation, and most of the time the innocent explanation is the right one. Which is why guessing about individuals is a bad strategy, and asking is a good one.

The famous signs are the late ones

Search for this and you will get the same list everywhere: unusual leave, calls taken outside, LinkedIn suddenly updated, dressing differently, logging off on time.

They are not wrong. They are just the end of the process. Someone taking a Thursday afternoon off for an interview has already applied, already been shortlisted, and in most Indian hiring processes is three or four weeks from an offer. You are not catching anything. You are watching a countdown.

There is also a cost to building your early-warning system out of these signs: it turns managers into people who notice what their reports are wearing. That is a worse relationship than the one you were trying to protect, and it does not even work.

The signs that come first

These are quieter, they come months earlier, and almost all of them are absences rather than events.

• They stop disagreeing. The clearest single signal. Someone who used to push back in a design review and now says “sure, either way” has not become agreeable. Caring has become expensive and they have stopped spending.
• Their horizon shortens. They engage fully with this sprint and vaguely with anything past the next quarter. Watch who stops having opinions about the roadmap.
• They stop volunteering. Not refusing work, just no longer picking up the unowned thing, the interview panel, the new joiner.
• Their answers get shorter. “Fine.” “All good.” If your weekly check-in used to get three lines and now gets one word, that is data, and it is the kind that shows up in a tool before it shows up in a meeting.
• They go quiet in groups and stay normal one-to-one. Often the first visible change, and often misread as a confidence problem.
• Recovery stops. They take leave and come back looking the same as when they left. This one is worth taking seriously on its own.

Every item on that list has a dozen harmless explanations: a bad quarter, something at home, a project they dislike. None of them is evidence. All of them are a reason to ask.

What the team shows before the person does

Here is the part that changes how you look for this. Individuals are hard to read, easy to misjudge, and uncomfortable to watch. Teams are not.

A department whose average mood has drifted down for six weeks, whose meeting load has climbed, and where fewer people are bothering to answer the weekly question at all, is a department where somebody is working up to a resignation. You do not know who. You do not need to know who, because the useful responses at that point are things you would do for the whole team anyway: cut the meeting load, fix the on-call rota, move the deadline, make the expert sessions genuinely easy to book.

That is why Wanderfly reports at team level and hides any group under five people, and why no leader can see an individual’s mood, stress score or bookings. It is not only a privacy decision. Naming people would make the measurement worse: the moment an answer is attributable, it stops being honest, and you are left measuring how safe people feel being candid rather than how they are.

The conversation to have

A stay interview is half an hour with someone who is not leaving, about what would keep them and what might not.

It sounds obvious and almost nobody does it. Gallup’s 2019 research found that 51% of people who left said nobody had talked to them about their satisfaction or their future in the three months before they resigned. Over half of all voluntary exits happened in silence, and 52% of people who left said their manager or organisation could have done something.

How to run one so it is not theatre:

• Their manager runs it, not HR. If HR runs it, it is a process. If the manager runs it, it is a relationship.
• Twice a year, scheduled in advance. A stay interview arranged at short notice reads as a response to something, and they will spend it working out what.
• Not attached to the appraisal. Nobody tells you what would make them leave in the same month you decide their increment.
• No form. Write three lines afterwards for yourself. The moment it produces a record the person has not seen, it stops working.

Questions that get an honest answer

The failure mode is asking something so general that “yeah, good” is a complete answer. Ask about the recent and the specific.

1. What is the best week you have had here in the last three months, and what made it that one?
2. What is the thing that most often makes a Monday harder than it needs to be?
3. Which part of your week would you delete if you could, with no consequences?
4. What are you learning at the moment? If the answer is nothing, that is the conversation.
5. If you were somewhere else in a year, what would it most likely have been about?
6. What have I got wrong recently that you did not tell me about?

Question six is the one that earns the rest. Asking it, and then not getting defensive about the answer, is most of what makes the other five work the next time.

Do not ask “are you happy here”. In a company where someone is on a 60- or 90-day notice period and depends on a reference, the only safe answer is yes.

What to do with what you hear

One rule: change one thing within two weeks, and say which thing and why.

It does not have to be the big thing. Someone who says their Mondays are ruined by a 9am status meeting and finds it moved to Tuesday has learnt something more important than the meeting slot: that saying things here causes things to happen. That is the entire mechanism. A stay interview that produces sympathy and no change is worse than not having one, because it answers the question the person was really asking.

If what you hear is about pay and you cannot move on pay, say so plainly and early. People leave over money much less often than they leave over being told a vague maybe for two quarters.

When you are wrong

You will read someone as a flight risk who is simply having a hard month, and you will miss someone who resigns cheerfully on a Tuesday. Both will happen.

That is the honest reason to treat individual signals as prompts to ask rather than conclusions to act on, and to put the measurement at team level where it is both more reliable and less intrusive. The leading-indicator argument sets out what that measurement has to do to be worth anything, and the 90-day plan covers what to change once you have it.

And if someone does resign: ask what would have had to be different, and when. The answer is almost never about the last month.

Questions people ask