The sentence arrives in most exit conversations in some form, and it has been repeated at conferences for so long that it now functions as common sense. A resigning employee says the new role is a better opportunity. The manager says the person was ambitious. Somebody in HR says, afterwards, that people do not leave jobs, they leave managers. Everyone nods and the file closes.
The line is worth taking seriously precisely because it is so often repeated. It is also worth testing, because a claim that explains everything explains nothing, and a manager who hears it as an accusation learns nothing they can use on Monday.
What the claim gets right
There is a real mechanism underneath the slogan. Most of what determines whether a job is tolerable is decided locally: how work is assigned, whether effort is noticed, whether standards are consistent, and whether a person can raise a problem without paying for it. Those are managerial decisions, not organisational ones. A good salary does not compensate for a manager who changes their mind weekly.
So the claim is directionally right, but the word “leave” implies a single dramatic event when the decision is usually accumulated. People do not resign on the day of the bad meeting. They resign months later, after deciding that nothing will change, and cite the offer in hand because it is easier to explain.
By the time someone resigns, they have usually been leaving for a year.
That gap is where the work sits. Spot the pattern while it is forming and the resignation becomes preventable.
The behaviours that actually push people out
Ask people who have left why, a year later, and the answers cluster less around temperament than around a small set of behaviours. Very few involve a manager being unpleasant; most involve a manager being unpredictable.
- Uneven standards. One person is pulled up for lateness, another is not. The team reads this accurately and stops believing that performance is what gets rewarded.
- Information withheld. Decisions that affect someone’s work are announced after the fact, or not at all. Being the last to know is read as being unimportant.
- No cover. When a project fails, the manager names the person who made the call rather than owning the decision they approved. This is the fastest way to lose a good senior person, because they can see the same thing happening to them.
- Work assigned by habit. The same person gets the interesting project, the same person gets the reporting. After two years, the second has evidence that their development is on nobody’s list.
- Correction-only feedback. A manager who speaks only to correct creates a team that avoids them.
Notice what is absent from that list: charisma, warmth, being liked. Some of the managers people stay with longest are blunt and hard to please. Their common quality is predictability. You know where you stand and what earns criticism, and the rules do not change mid-quarter.
| Behaviour | What it looks like in an ordinary week | What it costs over a year |
|---|---|---|
| Uneven standards | One late arrival mentioned, another ignored | Discretionary effort drops to the minimum |
| Withheld information | A reorganisation known to two people out of eight | Rumour fills the gap, usually worse than the fact |
| No cover | A client escalation pinned on the analyst | Senior people start hedging in writing |
| Habitual assignment | Same person on the visible project again | The other person updates their CV quietly |
| Correction-only feedback | First conversation in six weeks is a complaint | The team stops raising problems early |
Why exit interviews do not catch it
Exit interviews are conducted by someone with an interest in a clean record, at a moment when the employee has already decided to be polite. The person leaving has a new job to start and no reason to spend an hour describing a manager’s behaviour to a company they are leaving. So the form records “better opportunity” and “career growth”, which are true but incomplete.
Three sources are more honest, and all are available before anyone resigns.
Internal moves. Where people ask to transfer out of a team, and how often. A pattern of sideways moves out of one department is the clearest signal available, and it is usually recorded somewhere nobody reviews.
The questions people stop asking. A team that has stopped raising issues in review meetings has not become harmonious. It has concluded that the meeting is not the place.
Skip-level conversations. The manager above hears things the direct manager never will, not because the team is disloyal but because some things are only sayable upwards.
What an organisation can do, short of replacing every manager
Replacing every manager is not an option, and the replacements would be promoted on the same criteria that produced the current group: technical performance and tenure. The change has to come from what the role demands and how it is judged.
- Write down what the role is for. A one-page manager standard, agreed by the leadership team, listing four or five observable behaviours. Not values on a poster: behaviours a person could be observed doing or failing to do. Make retention a manager metric, with context. Team attrition, internal transfer requests and engagement scores belong in the review, alongside a conversation about why the numbers look as they do. Without context the metric gets gamed; as a prompt for a conversation it changes attention. Run upward feedback managers cannot opt out of: two questions, twice a year, results shared with the manager and their own manager. Uncomfortable for one quarter, ordinary after that.
- Give managers somewhere to take the hard cases. A person who has just been promoted and is managing a former peer, or a long-serving underperformer, needs a conversation with someone experienced, not a course. This is where coaching earns its cost, because the issue is specific.
- Make the transfer route visible. If a good person is unhappy with a manager, an internal move is cheaper than a resignation. Many organisations have the mechanism and never mention it.
Assessment helps at the point of selection: a well-built assessment centre can identify people ready to manage before they are promoted into it. It cannot fix a manager who is already in place and behaving badly, and organisations that buy it expecting that tend to be disappointed.
The four questions worth asking this quarter
To find out whether the claim is operating in your organisation, skip the exit files and ask these.
- Which teams have had more than one internal transfer request out in the last twelve months?
- Which managers have had no upward feedback at all in the last two years?
- When a project fails, who is named in the review, and at what level does the account stop?
- Can a manager in this company describe, in one sentence, what the company expects of them beyond hitting targets?
None of this needs a survey. What is missing is not data but the decision to look. For managers new to the role, the practical work of getting these behaviours right is covered in our note on management skills for first-time managers.