Employee Flight Risk Indicators You Can Read Without HR Software

Three months before he resigned, your operations head stopped fighting you about the vendor.

You remember the meeting because it went unusually smoothly. He had opposed that supplier for a year — the pricing, the delivery record, the fact that the proprietor was known to the family. That Thursday he listened, said theek hai sir, let us try it for one quarter, and moved to the next item. You drove home thinking he had finally learned to pick his battles.

He had not learned anything. He had stopped bothering. The argument was not worth the friction because he was not going to be there for the consequences.

Search this question and you will get a dozen lists of warning signs. Look at who published them. Almost every one is a workforce-analytics or HR-software company, and every article lands in the same place — connect your data, score your people, watch the dashboard turn amber. The lists are not wrong. They are a sales sequence with an article on top.

You do not need any of it. You have between four and forty people whose departure would genuinely hurt, you see most of them every week, and the signals that predict a resignation are behavioural, personally visible and free. They require attention pointed in the right direction, and for most promoters it is pointed at exactly the wrong thing.

The inversion this article is built on

Here is the claim, and everything below follows from it: the signals that matter are withdrawals, not complaints.

A person who is complaining is still invested. The complaint costs him something — goodwill, political capital, an awkward twenty minutes outside your cabin — and nobody spends that on a place he has already decided to leave. Complaint is an act of belonging. It says I expect this to be better and I expect to be here when it is.

The dangerous state is silence, and silence is what almost every promoter reads as contentment.

That is why resignations feel like ambushes. You were watching the man who argues, who sends long messages at eleven at night, whose face you can read across a room. He is not going anywhere. The one who left was the one who had become easy to manage.

This sits inside a larger sequence, which the pillar piece on why your best people leave sets out as the Exit Clock: five stages, nine to fourteen months, resignation at stage five. What follows are the six signals that appear at stage three — the last stage where holding on to someone is still cheap.

The Six Withdrawals

One — they stop arguing

This is the single most reliable indicator and the most commonly misread.

The person who used to push back in the review meeting now agrees quickly. Not sullenly. Pleasantly. He accepts a target he would have contested last year, nods at a decision he thinks is wrong, and does not follow up with the note he always used to send.

Promoters do not experience this as a warning. They experience it as relief, then as maturity — he has finally settled down, he understands the pressures now. In a family-influenced business where deference is already the default, it is easier still to miss, because agreeable behaviour looks like correct behaviour.

Disagreement is expensive. People only pay for it when they expect to live with the outcome. Ask yourself when a senior person last changed your mind about something that mattered. If you cannot place it inside the last quarter, and could have last year, that is your signal.

Two — they stop asking for things

No more requests. Not for a headcount, not for budget, not for a tool, not for scope, not for the training programme he raised twice in 2025.

An ask is an investment in a future inside your business. It is a person spending effort today on a condition that only pays off if he is still here in a year. When the asks stop, the future has been relocated. He is no longer building anything that requires your approval.

This is the quietest of the six because nothing appears to have happened. Nobody notices the absence of a request. There is no meeting where it fails to occur. The unanswered ask is where the clock starts — a real request neither delivered nor refused, just allowed to go quiet — and the stage after it is when he stops asking at all.

Most promoters, asked when a key person last requested something substantial, cannot answer, and then read their own failure to remember as evidence that everything is fine.

Three — they withdraw from anything without an immediate deliverable

The work continues, which is what makes this hard to see. Reports land, numbers get met, customers are handled.

What stops is everything adjacent to the job. He drops out of the process improvement group. He does not want the interview panel any more. The new-office committee meets without him. He has a reason each time, and each reason is true — the month-end, the audit, a customer escalation.

Discretionary effort goes first because it is the only part of the job nobody can hold him to. You cannot put "did not join the committee" in an appraisal. He knows that, and he is withdrawing along the exact line where withdrawal is unpunishable. That is not cynicism. It is accuracy.

This is where the behaviour starts resembling the wider pattern in quiet quitting in India — the internal resignation that precedes the written one, sometimes by years.

Four — they stop building relationships beyond their function

Watch the edges of the organisation, not the centre of the role.

A new person joins finance and your production head does not walk over. The informal lunches with the sales team stop. He no longer takes the plant visit he used to take because he liked knowing the supervisors. He answers in the cross-functional WhatsApp group only when tagged, and then only with what is required.

Relationships across functions are how a person builds standing over a five-year horizon. They cost time now and pay later. A man consolidating into his own function is not being territorial — he is being efficient about a shorter horizon than the one you have in mind for him, protecting what he must still deliver and letting go of what only mattered if he stayed.

Five — their time horizon shortens

This one is the most precise of the six, and almost nobody writes it down.

Listen to the unit of time. Plans that used to be referenced in quarters start being referenced in weeks. A person who used to say by next Diwali we should have the second line running now says let me finish the audit and then we will see. The phrase "next year we should" disappears from his vocabulary entirely.

Watch what happens when you raise something eighteen months out — a new facility, a systems migration, the second generation entering the business. Two years ago he had views. Now he asks one clarifying question and waits for you to finish. He is not being difficult. He is simply not placing himself in that picture, and language follows belief faster than behaviour does. Nobody controls their own tense.

Six — the leave pattern changes

Single days. Short notice. Often mid-week, sometimes clustered — a Tuesday and a Thursday in the same fortnight, both booked the evening before, both for something domestic and unspecific.

Interviews happen on weekdays, and a senior process in India runs to four or five rounds across several weeks, each costing half a working day.

Handle this one carefully, because it is the only signal here that can damage you if you get it wrong. It is corroborating evidence, not proof. Parents get ill. School admissions happen. Property paperwork in this country eats whole days and nobody enjoys explaining it. Start treating leave applications as suspicious and people will notice inside a fortnight, and you will have manufactured the exact mistrust that produces the resignations you were trying to predict.

Use it only in combination. A changed leave pattern on its own means nothing. The same pattern in a man who has also stopped arguing and stopped asking is a different sentence entirely.

The signals that do not work

Half the value here is stopping you watching the wrong things. Most of the popular indicators are either too late or too noisy to act on.

• LinkedIn activity. A profile update, a burst of connections, a sudden appetite for commenting. By the time this appears the decision is usually made, and it is noisy both ways — plenty polish a profile out of vanity, and the ones actively interviewing often never touch it, knowing you can see it.
• Reduced output. The indicator every promoter trusts, and a lagging one. Performance drops after the internal decision, frequently after an offer is in hand. When you finally see the numbers slip you are already at stage four, where buying somebody back is the expensive proposition we set out in should you counter-offer a resigning employee.
• Visible unhappiness. The complainers mostly stay. Grumbling in the corridor, arguing about the appraisal, telling you the new system is a waste of money — that is investment wearing an unattractive expression. Gallup's job-search figures do show intent tracking engagement, at 74% of the actively disengaged, 55% of the not engaged and 30% of the engaged, but note the last one. Three in ten of your genuinely engaged people are looking too. Mood is not a filter.
• Long tenure as reassurance. The costliest misreading in Indian businesses. The fifteen-year employee is not your safest person; he is frequently your most exposed. His pay has fallen furthest behind the market, because increments compound off a small base and nobody rebenchmarks a loyal man. His role has changed least, because he is dependable exactly where he is. When he goes, he goes suddenly, with a completed decision and no handover document anywhere in the building.

Your strongest people leave for structural reasons rather than emotional ones, which is a related pattern we take apart in why high performers leave.

Why the timing of these six matters more than the list

All six appear at stage three of the Exit Clock — the passive-looking phase, before any offer exists. That is the reason to learn them. Nothing has been committed, no number written down, no spouse or parent told, and the request from stage one, granted now, will very often still work at a fraction of what the next two stages cost.

By the time you can see it in the output you are at stage four. There is a letter, a date, and a decision already announced at home, and buying it back runs to a 30 to 40% premium that rarely holds. Gallup found that 52% of voluntarily exiting employees say their manager or organisation could have done something to prevent them from leaving — and almost all of that something was available months before anybody typed a resignation letter. Asking afterwards does not recover it, which is why the instrument most businesses rely on is the wrong one; we deal with that in why exit interviews don't work.

How to test it, this week

Fifteen minutes, alone, on paper. No system and no conversation with anybody.

Write down your five most important people. Not the five senior-most — the five whose resignation on Friday would genuinely hurt. Then, for each, write three things:

• The last time they argued with you about something that mattered. A date or a month, not "he speaks up sometimes".
• The last thing they asked you for. Headcount, budget, scope, a tool, a title. Something that required your approval.
• The last commitment they made that extends beyond this quarter. A named outcome with a timeframe attached that they volunteered, not one you assigned.

Then look at the blanks. A blank row is your answer, and two blanks in one row is a man standing at stage three. Do not act on it by confronting him — that produces a denial and puts him on guard. Act on it by going back to the thing he last asked for and either delivering it or refusing it plainly, with a reason.

Where to start

Do the five names this week. Then take the row with the most blanks and, within seven days, reopen the ask that was never answered — not as a retention conversation, just as a decision you owe him. What holds a man who has options is a small set of specific things, set out in how to retain your key people.

If the blanks run across most of the five, the problem is not those men — it is that your business has stopped generating things worth asking for. The Business Pulse diagnostic maps where that happens: which roles have no next step attached, and which layer stopped being able to grow.