Why Your Managers Leave — The Ceiling Test

He was the third in four years, and he was the best of the three.

The first left for a company nearer his home town. The second left because his wife was posted to Pune. This one is going to a competitor for what he calls a bigger role, though the designation on the letter looks a great deal like the one he already has. Three men, three reasons, no connection between them, and you accepted each explanation as it came because each was entirely plausible.

Now put the dates side by side. The first lasted twenty-two months. The second, twenty-six. This one is going in his twenty-fourth.

The reasons vary. The timing does not.

When the stated reasons differ every time and the timing holds steady, you are not looking at three decisions. You are looking at one structure producing the same result three times. Something in that seat has a limit built into it, and every capable occupant reaches that limit at roughly the same point — which is why the better the man, the faster he gets there.

This is the most expensive attrition in a promoter-led business and the least examined, because it always arrives disguised as an individual event. Three juniors leaving the same desk is read as a problem. Three managers leaving the same role over four years is read as bad luck, three times.

The sequence underneath any single one of those exits is the one set out in why your best people leave — the decision made nine to fourteen months before the resignation. This article is about why the sequence starts at the same month for every person you put in that chair.

The layer that is falling fastest

Start with what is happening to the population your second line is drawn from and belongs to.

Gallup's India figures show manager engagement falling from 39% to 30% in a single year, while individual contributors fell from 24% to 19%. Managers remain more engaged than the people below them, which is what you would expect. But the manager layer is deteriorating faster than the layer it supervises.

That matters for you specifically because in a business of your size this layer is thin — six people, perhaps ten — and it is the only thing standing between you and every operational decision in the company. When the layer below erodes you lose output. When this one erodes you lose the ability to be absent.

It also means the replacement you are about to hire comes from a pool that is getting worse, not better.

The Ceiling Test

Four ceilings. Most seats have one, some have two, and the seat that has emptied three times usually has three. Each has a distinguishing question you can answer from memory in two minutes.

The authority ceiling

He is held responsible for a number he cannot make the decisions to produce.

He owns the plant target but cannot authorise overtime past a figure set four years ago. He owns the sales number but every discount above 5% comes to you. He owns delivery but the vendor was chosen by you and cannot be changed by him. None of this is written anywhere. It emerged, decision by decision, and it is now the shape of the job.

The distinguishing question: name the three largest decisions he made alone last quarter that you did not review. Not decisions he recommended and you approved — decisions he made, that you heard about afterwards, and did not overturn.

If you cannot name three, this is his ceiling, and he found it long before you did. A layer that carries accountability without decision rights is a layer doing coordination and calling it management, which is the mechanism in why middle management fails.

The succession ceiling

The seats above him are not available. Usually because they belong to family, or because you are sitting in them and have no plan to move.

Be clear about this: the constraint is legitimate. It is your company, nobody is owed your chair, and a promoter who keeps the commercial function to himself for twenty years is not committing an offence. Some of the best businesses in India are run exactly this way, deliberately.

The failure is not the constraint. The failure is that it is never said out loud. So a capable man spends two years reading the room — which meetings he is not in, which decisions still route to your brother-in-law, what happened when the last person asked about the COO title — and reaches an accurate conclusion by inference, in private, without ever having had the conversation. By then he has also concluded that no honest answer was available, so he is not negotiating a different kind of future here. He is interviewing.

The distinguishing question: has anyone told him, in words, what the top of this business looks like for a non-family professional? If your answer is that he must have understood by now, this is the ceiling.

The learning ceiling

He has stopped learning from you specifically.

In a promoter-led firm you are usually the manager's only real source of development. No structured training, no peer group at his level, no boss above him except you. For the first eighteen months this is an extraordinary arrangement — he is taught the business by the man who built it, and he knows it is worth more than an increment.

Then the transfer completes. He has watched you price, negotiate, handle the difficult customer, decide what to do when the money is short. He can predict your view before you give it, which is exactly the point at which you stop being a source of learning and become a source of approval. Nothing has gone wrong. The teaching simply finished.

Now he has two options: grow sideways into something you have not built, or leave. Most businesses of your size have nothing sideways.

This is the least recognised of the four, and it takes your strongest people first, for the reason in why high performers leave — the better he is, the sooner the transfer completes.

The credibility ceiling

His own team routes around him to you.

They come to your cabin directly. They copy you on things he should close. They wait for your view before acting on his. He cannot hire without your sign-off, cannot remove a non-performer at all, cannot approve a ₹40,000 spend, and everyone reporting to him knows where the real decision sits — often through the WhatsApp group he is not in.

He is not underperforming. He is unable to perform. The distinction is invisible from your chair, because the visible facts are identical: slow decisions, deferred conflicts, a man who does not seem to grip his function.

This is the ceiling most often misread as a capability problem, and the misreading is expensive, because it leads straight to the conclusion that you need a stronger person.

Why the stronger replacement fails faster

So you hire better. More experience, a bigger company on the CV, a salary to match.

He arrives and does what strong managers do: he starts deciding. He removes a supplier. He tells a fifteen-year employee that something will now be done differently. He commits to a customer without checking. Each of those is what you hired him for, and each crosses a line that exists in your head and appears on no document.

Then it happens in a meeting. He states a decision and you correct it in front of his team — not harshly, possibly gently, possibly with a joke. It takes eleven seconds and it is finished. Everyone in that room has learned where authority actually lives, and no private assurance afterwards reverses it.

He lasts a quarter, sometimes two, then leaves faster than the man he replaced — and you conclude that senior hires do not work in your business. The pattern has its own mechanics, in why executive hires fail.

The conclusion to hold on to is this: the seat sets the outcome, not the person. That is why the third manager leaves at the same point as the first two, and why the strongest of the three leaves soonest.

What it costs, properly counted

Gallup puts the cost of replacing an employee at one-half to two times their annual salary. For a manager that understates it, in three ways.

The first is time to authority. A successor can be productive in eight weeks. He cannot decide anything for six to nine months, because nobody below him acts on his word until they have watched you accept it. For that period you are back in every operational conversation you spent two years climbing out of.

The second is the relationships he held. His team's connection to your company ran through him — the supervisor who joined because he asked, the accounts man who stayed through a bad year after a conversation in the car park. Those do not transfer. They lapse.

The third is second-order, and promoters see it too late. Attrition inside his team rises after he goes, and two of the people who leave were his own hires. Their reasons will again sound individual and unconnected. Do not read that as proof that people leave managers rather than companies — the maxim is weaker than its popularity suggests, and we take it apart in do people leave managers or companies. Read it as the same ceiling, now visible one level down. The ones who stay but stop investing are a further cost with no line item at all, described in quiet quitting in India.

How to test it

Ten minutes, on paper, for every manager who reports to you. Four blanks each.

• What can he decide alone? Write the actual decisions, not the job description. If the honest list is "which of two people works Saturday", write that.
• What can he spend alone? One rupee figure. If you cannot state it because it depends on what the spend is for, that is the answer — there is no limit, there is only your mood, and he has learned to ask.
• Can he remove someone from his own team? Yes or no. Not "he would discuss it with me first" — can he.
• What is his next role here called? A name. If the sentence needs the words depends or eventually, it is blank.

Four blanks and you have found the ceiling. Two blanks and you have found which one it is. If you can fill all four for a manager who resigned anyway, his departure genuinely was individual — also worth knowing.

Run the same four for the seat that has emptied three times, and you will usually find the answers did not change across any of the three occupants.

What to do when the ceiling cannot be removed

Some you can lift this quarter. Give him a spending limit in writing. Give him the hiring decision for his own team. Stop taking the meeting his people should be having with him.

Some you cannot. The commercial function may genuinely be reserved for your son, who is twenty-six and returns in three years. That is a structural decision, and no amount of engagement work changes it.

Where the ceiling cannot be removed, the choice is not between keeping him and losing him. It is between him learning the truth at month six and learning it at month twenty-four.

A stated boundary with a stated timeline beats an implied promise, every time. Told plainly at month six — this is the top of this role, here is what it will carry, here is what will never be available, here is what I will pay and teach you in exchange — a capable manager will often stay three more years. He can plan. He can decide, with full information, that the trade is good. Many do.

The man who works the same facts out by inference at month twenty-four always leaves, and leaves believing he was misled — which costs you a reference in a market where your next three managers will check.

Where to start

Take the seat that has emptied more than once and fill in the four blanks for it, before you write the job description for the fourth person. If three come back empty, this is not a hiring problem, and hiring harder will produce the same result faster.

What holds a manager once the ceiling is honest is in how to retain your key people. The quiet signals that appear months before he tells you anything are in employee flight risk indicators. If you are building this layer for the first time rather than replacing it, start at build your first leadership team.

If you want the structural version — which roles carry accountability without decision rights, and which have no next step attached to them at all — the Business Pulse diagnostic maps it. It is the only work here that survives a change of personnel, because it fixes the seat rather than the man in it.