Do People Leave Managers, or Companies? What the Evidence Actually Says
Somebody said it to you within a week of the resignation. Your HR head, or a consultant, or a director over the phone that evening. People don't leave companies, sir. They leave managers.
It landed, because it explained something you could not otherwise explain and it did so cleanly. One person, one relationship, one thing to fix. You started re-examining the reporting manager. You may have already decided to move him.
Before you do, it is worth knowing that the sentence is a compression of something more complicated, that the research it is attributed to does not state it in that form, and that two separate bodies of evidence say the compressed version is wrong in a specific and expensive way.
This is not the usual correction where a claim turns out to be half-true and therefore harmless. Acting on this one sends you to the wrong intervention, and you can spend two years there. The pillar this sits under, why your best people leave, sets out the sequence a resignation actually follows. This page deals with the single sentence that stops most promoters from seeing that sequence at all.
Where the sentence came from
The maxim is popularly traced to Gallup's research as told in First, Break All the Rules, the 1999 book by Marcus Buckingham and Curt Coffman, drawing on roughly 80,000 manager interviews conducted between 1975 and 2000.
That is a serious body of work, and the book's actual argument is a reasonable one: that the immediate manager shapes an employee's experience of work more than most people assume, and that great managers do not manage the way the textbook says.
What it is not is a book containing the sentence you were quoted. The compressed slogan is a later simplification rather than a verbatim finding. It was distilled by other people, repeated for twenty-five years, and hardened into something that gets used as a law. That distinction matters, because everything the compression drops — the qualifications, the range, the other variables — is precisely the part you need.
What two independent lines of evidence say
The first is a review rather than a single study, which is the right instrument for a question like this. Jafni Bin Johari Jiken and Dr Asrif Yusoff of Durham University Business School, publishing in the International Journal of Organizational Analysis, examined 39 papers from the past ten years and reached a conclusion that is deliberately two-sided: people leave managers, but they also leave organisations.
Workload, pay, scheduling and progression clarity matter comparably. And the sentence in that review that should stay with you: even a good manager cannot retain people when the workload is high and the resources are inadequate. The manager is not a substitute for the conditions he is managing inside.
The second line is quantitative. Culture Amp analysed 175 teams of at least eight people each, across hundreds of fast-growing companies, and attributed departures as follows: development opportunities 52%, leadership 28%, the direct manager 12%, pay 11%.
Read that carefully rather than triumphantly. The direct manager at 12% is not zero — it is a real share, and on a base of every departure in your business it is a real number of people. But it is the third-smallest of the four, and the largest by a distance is the absence of anywhere to go. Pay at 11% is the figure most promoters would have put first, which is a separate correction and one we take apart in do employees really leave for money.
Culture Amp calls the maxim "the biggest lie in HR", and also observes something worth noting plainly: the maxim's biggest proponent sells management training.
That observation does not make the claim false. People who sell a solution are sometimes entirely right about the problem. It does mean the claim deserves the scrutiny you would apply to any other sales argument — because a proposition that routes every retention problem, whatever its origin, to the purchase of manager training is a proposition with an unusually convenient shape.
What the correct statement actually is
Here is the part most rebuttals get wrong in the opposite direction.
The correction is not "the manager doesn't matter." The manager clearly matters. Gallup's State of the American Manager work found that managers account for at least 70% of the variance in engagement scores across business units — and stated precisely, that means 70% of the difference between teams inside the same company, not that the manager causes 70% of engagement. It is a strong statement about why one team feels different from the team next door, and a silent one about everything the company does equally to both.
The correct statement is narrower and less quotable. The compressed maxim assigns to one variable what the evidence spreads across several. Manager, organisation, workload, pay, progression clarity — all of them load, in proportions that vary by business.
And the reason to care is not accuracy for its own sake. It is that the compression sends you to the wrong intervention, confidently, with a clear conscience.
The Attribution Error
This is the framework, and it is a description of what a promoter does after he accepts the maxim. Three things follow, and all three are wrong.
He replaces or coaches the manager, and the attrition continues
The most common version. A team is losing people, the maxim identifies the cause, and the manager is either sent for training or quietly moved. Six months later somebody else in that team resigns, and then somebody else.
This confuses promoters, because the intervention was decisive and the outcome did not change. The constraint was usually never the manager. It was that the roles in that team had no next step attached to them — no larger scope, no title that meant anything, no decision anyone could own. A new manager arrives into the same structure and delivers the same nothing, more pleasantly.
You have then also lost a manager who may have been perfectly competent, and taught the rest of your leadership layer that when a team has a problem, the person in charge of it gets removed. Which is its own retention problem, and the subject of why your managers leave.
He reads the departure as a personality clash, and therefore as unpreventable
This is the quieter cost and the more damaging one.
If people leave managers, then a departure is fundamentally about a relationship between two individuals — chemistry, temperament, ego, style. And relationships between individuals are treated, in most businesses, as weather. Unfortunate. Not really anybody's fault. Nothing structural to look at.
That reading excuses you from the examination. Nobody asks whether the role had a ceiling, whether the person had asked for something a year ago and never received an answer, whether three people in that function have now left at the same seniority for the same kind of job elsewhere. The pattern stays invisible because each case was filed as a clash of personalities, and clashes do not aggregate into evidence.
He treats retention as a management-skills problem rather than a design problem
Follow the maxim consistently and every retention intervention becomes a training intervention. A workshop. A coach for the sales head. A leadership module.
None of these touches how roles are defined, whether progression exists, who is allowed to decide what, or where the ceiling sits in each function. Nothing about the design changes, and the design is what produced the departures.
The finding that should settle the argument
Culture Amp's analysis contains one result that is stronger than anything else in this article, and it deserves to be stated on its own.
Where development opportunities were poor, management quality made no measurable difference to retention.
Not a reduced difference. No measurable difference. In teams with nowhere to go, a good manager and a poor one produced the same outcome — the people left either way.
This is what the maxim cannot survive. If the manager were the primary driver, management quality would show up everywhere, most of all in the difficult environments where it should matter most. Instead it disappears exactly where the structural constraint is binding. That is the signature of a variable that transmits an effect rather than one that causes it.
The manager is the transmission, not the source
Which is the reframe worth keeping.
A good manager can deliver clarity, scope, progression and consequence — but only if the organisation has those things to give him. He can pass on what exists. He cannot manufacture a next role that the structure does not contain, or authority the promoter has not released, or a decision he is not permitted to make.
Where the organisation has none of it, the manager becomes a well-liked person delivering nothing. His people like him. They say so. Then they leave anyway — politely, with a good word for him on the way out, often genuinely meaning it.
This is also why exit interviews so reliably produce the combination that confuses promoters most: no complaints about my manager, alongside a resignation letter. Both statements are true. Neither is the reason. We deal with why that instrument produces this result in why exit interviews don't work.
Why the maxim collapses entirely in a promoter-led business
There is an Indian-specific problem with the sentence that nobody writing it in America has to consider.
In a great many businesses of your size, the manager and the company are the same person. If your second line reports to you, or reports to a family member who reports to you, then "he left his manager, not the company" is not a distinction. There is nothing on either side of the comma. The maxim asks you to separate two things that, in your structure, are one thing — and it hands you a comfortable answer for a question it cannot address.
The second case is the one you will recognise faster. The long-tenured manager who has been with you fourteen years, who everybody likes, who is unfailingly decent to his team — and who has no authority whatsoever. He cannot approve a hire, set a number, or promise anyone anything, because those decisions have never left your cabin.
The maxim predicts his team should be stable. It is not. People join under him, spend two years, and go somewhere they can be given something. He is a good manager in every sense the maxim measures, and he has nothing to transmit. That layer, hollow and well-liked, is examined in why middle management fails.
How to test it in your own business
You do not have to take any of this on trust. The test takes an hour and uses data you can get.
Take your last five departures at any level that mattered. For each one, establish a single fact: was the role they went to a step up in scope or title? Not a step up in salary — scope or title. A bigger team, a wider remit, a function they now own, a designation that is genuinely senior to the one they held with you.
You can find this. It is on their public profile, the recruiter who placed them knows, or you can ask them directly a year on, when nothing is at stake and most people answer honestly.
Then count. If three or more of five moved up, the constraint in your business was progression, not management — and no amount of work on your managers would have held them. If most moved sideways for similar work, the maxim has a stronger case in your firm and you should look harder at specific reporting lines.
One caution, because this is a real test and should be described honestly: it establishes a pattern, not a cause. A person can take a bigger title and still have left over a manager. But it is directional evidence drawn from what people actually did, which makes it better than anything an exit interview will give you, and it points at your structure rather than at a personality.
The single next action
Run the five-departure test this week, on paper, with names.
Then take whichever answer it gives you and do the thing that answer implies — not the thing the maxim implies. If progression was the constraint, the work is on roles and next steps, and it starts with the two or three people you cannot afford to lose next, which is where how to retain your key people picks up.
If you want the structural view rather than the last-five view — where in your organisation roles run out of next steps, and which of your managers hold a title but no decisions — the Business Pulse diagnostic maps it. It is the difference between fixing a manager and fixing the thing he was standing in front of.