Behind a Company-Wide Turnover Rate of 4%: How an Entire Department Turned Over in Three Years

If you look only at a company’s annual turnover rate, everything can appear normal.

How reading tools work

Listen reads the article aloud. Speed read shows phrases in sequence at your chosen pace. Language practice compares available translations. Save keeps a bookmark in this browser; find it in the player’s bookmarks.

Share this article

Share on X (opens in a new tab)Share on Bluesky (opens in a new tab)

Not “employees without initiative,” but a workplace where taking initiative becomes dangerous

If you look only at a company’s annual turnover rate, everything can appear normal.

Suppose the company-wide turnover rate is around 4%. Judging from that number alone, the company might be described as “a place where few people quit” or “a relatively stable employer.”

But once the figure is broken down by department, direct supervisor, age, hiring route, and tenure, a completely different picture may emerge.

In one small department, every person who had belonged to the team a few years earlier eventually left through resignation or transfer. New hires were added along the way, so the headcount itself remained roughly intact. Yet the people still there were all employees who had joined within the previous one or two years.

There were still people.
The department name still existed.
The work was still being done, at least on paper.

But experience did not remain. The contents of the department were being replaced every few years.

That is not ordinary generational turnover.

It is a cycle in which the department hires newcomers, loses people once they understand the work, and fills the holes with more newcomers instead of developing and retaining accumulated experience.

Several other patterns were occurring at the same time.

Mid-career permanent employees around their thirties left the department.
Mid-career hires who had joined only recently requested transfers because they felt they could not continue in the current workplace.
Experienced employees who had learned how to run the work resigned.
Even long-serving employees transferred through a consultation channel and higher management without going through their direct supervisor.
Employees who raised problems were answered not with workload adjustments but with comments about their “attitude,” “fit,” or “softness.”
Cases stalled when decisions were sent to the supervisor, but moved forward when the employee asked the actual practitioner or related department directly.
New work was added without removing old work, and subordinates were expected to fill in undefined parts.
Internal events were forcibly linked to numerical claims about operational efficiency, while the design of the measurement itself was pushed back onto the staff.

Seen one by one, each case can be explained away as “a bad fit,” “the person had personal circumstances,” or “the mid-career hire did not suit the workplace.”

But when similar departures repeat under the same department, the same management line, and the same way of working, the object of analysis should not be the individual.

The real object is an organizational structure that makes people stop moving and drives out those who have learned the work.

1. A company can have low overall turnover while a specific department is breaking down

1-1. Averages erase the concentration of turnover

Company-wide turnover uses the entire workforce as its denominator.

So if a company has many large, stable departments, complete turnover in a small team can be diluted in the average.

For example, if a company has several hundred employees and only a dozen or so resign each year, company-wide turnover may remain only a few percent. At the same time, if a department of about four people loses every original member over several years, its turnover in terms of retained experience is effectively 100%.

Both can be true at once.

The company has “low turnover.”
A specific department has a “0% survival rate for its former members.”

So a company-wide figure of 4% may be accurate, but it is far too coarse for organizational diagnosis.

What should actually be examined is the distribution by:

department;
direct supervisor;
permanent versus non-regular employment;
new graduate versus mid-career hire;
age group;
within 3 months, 6 months, 1 year, 2 years, and 3 years of joining;
voluntary resignation, transfer, leave of absence, and long-term absence;
job proficiency and difficulty of replacement.

In particular, when departures concentrate among mid-career hires × several months to several years after joining × a particular department or supervisor, the company should look beyond hiring mismatch and examine onboarding, role design, workload, supervisor behavior, and the safety of consultation routes.

1-2. Counting resignations alone understates the department’s loss

Internal transfers are not counted in the company-wide turnover rate.

But from the original department’s perspective, both resignation and transfer are losses of people.

Other conditions are also hard to see in the turnover rate:

requests to change the person responsible for training;
long-term absence;
leave of absence;
consultation with a harassment desk;
consultation with higher management while avoiding the direct supervisor;
“quiet withdrawal,” where communication shrinks to only what is strictly necessary for work;
starting a job search;
stopping the submission of opinions and improvement proposals;
returning decisions to the supervisor and refusing to decide independently.

So if the goal is to diagnose the health of a department, it is useful to look beyond simple turnover and consider a broader workplace exit rate.

Workplace exit rate
=
resignations
+ transfers out of the department
+ leaves of absence
+ long-term absences
+ consultations stating that continued work is difficult
÷
number of employees in the department

This is not a standardized official indicator, but it can be useful for internal diagnosis.

Even if an employee is saved by a transfer, that does not mean the original department’s problem has been solved. A transfer is relief for the individual; improvement on the causal side is a separate matter.

2. The personnel replacement occurring in a specific department

Generalizing the case so that no individual or company can be identified, the department’s trajectory looked like this:

A small group of existing members operates the department
↓
Temporary increase in staffing
↓
Multiple resignations and transfers
↓
New employees are hired
↓
Handover, training, and unresolved work concentrate on the remaining experienced people
↓
The experienced people also transfer or resign
↓
Only recently hired employees remain

The people who belonged to the department at one point were all gone several years later.

Those who remained were not the people who had built the operating system of the department or knew its past failures. They were relatively recent hires.

As a result, the same department name remained on the surface while important assets repeatedly disappeared inside it:

the reasons behind past decisions;
channels for coordinating with other departments;
unwritten deadlines;
ways of translating a supervisor’s vague instructions;
practical knowledge of which tasks are dangerous to ignore;
networks showing whom to ask when something needs to move quickly;
know-how for training newcomers;
intuition for spotting problem cases before they escalate;
organizational memory that says, “We tried this before, and it failed.”

Then unresolved work is handed to the newcomers.

Before new employees understand the work, they tend to accept the supervisor’s instructions as given, so the relationship may be relatively stable. Once they learn the work, notice contradictions and excessive burdens, and start proposing improvements, friction increases.

The department ends up in this condition:

New people are present.
But people do not remain after they become competent.

3. Why mid-career and mid-level employees leave first

3-1. Mid-career hires know work in general, but not the company’s hidden rules

Mid-career hires are experienced workers.

What they know, however, is the general way their profession or type of work is usually done.

They do not automatically know organization-specific information such as:

who is the real decision-maker;
whether verbal instructions or written documents take priority;
which department officially owns a task versus who actually processes it;
which supervisor causes a matter to stall when asked;
which related department can move it forward quickly;
which work is formal responsibility and which is goodwill-based gap filling;
what will later be called “acting without permission”;
what questions will later be called “lack of initiative”;
which deadline is the real deadline;
where responsibility will flow if something goes wrong.

Handing over this ambiguity with “You are experienced, so you can do it” is not onboarding.

It is simply forcing the mid-career employee to fill in company-specific unknowns using their own experience.

Meta-analytic research on newcomer adjustment identifies role clarity, self-efficacy, and social acceptance as important adjustment factors connected with job satisfaction, organizational commitment, intention to remain, and actual turnover. Research by the Japan Institute for Labour Policy and Training also reports positive relationships between organizational socialization or intention to remain and onboarding measures such as support from supervisors, senior colleagues, and peers; formal training; HR support; designated receiving staff; pre-entry meetings; and follow-up meetings after a set period.[^1][^2]

What mid-career hires need is not “pampering.”

At minimum, they need clear definitions of:

role;
priority;
completion criteria;
decision-maker;
what they may decide on their own;
where to ask questions;
periodic follow-up;
criteria for reducing existing work.

3-2. Mid-level employees collide because they can see the contradictions

During the newcomer stage, employees may not know what is abnormal.

They accept assigned work as “how this company does things.”

With experience, they begin to notice things such as:

the same data is being entered into multiple documents;
goals and metrics do not match;
only the person in charge is moving while the decision-maker remains undefined;
new work is added without reducing existing work;
tasks belonging to another department are picked up by whoever notices them;
previously approved content is later treated as unapproved;
a manager who refuses to decide becomes the evaluator only after failure;
responses to operational problems are converted into discussions of “attitude” and “fit.”

Mid-level employees raise problems because they can see the work.

But if a problematic manager equates cooperation with obedience, comments from someone who understands the work may be interpreted not as professional improvement proposals but as “defiance” or “lack of submissiveness.”

The people most valuable to the company may then be the ones most likely to leave:

those who can detect problems;
those who can compare the organization with other employers;
those whose skills are portable;
those who can structure work;
those who can move in the labor market;
those who can withdraw once they conclude improvement is impossible.

This is not necessarily a case of “weak people quitting.”

It can instead be adverse selection in which people who understand the problem and have the ability to leave are the first to go.

4. The core operating model: making employees absorb every undefined part

The problem in this workplace can be summarized as a combination of the following:

Make employees absorb everything that is undefined
+
Add new work without removing old work
+
Hand it over with “You can do this, right?”
+
Keep the volume high
+
When things stall, turn it into a question of the employee’s ability or attitude

4-1. Making employees absorb every undefined part

Things that management should decide are left undecided:

purpose;
scope;
mandatory items;
completion criteria;
reference date;
deadline;
responsible person;
approver;
boundary with related departments;
exception handling;
priority.

Yet the work still appears.

The person who notices the gap, feels responsible, or can explain things ends up filling it.

In this environment, the more capable someone is of keeping work moving, the more work they receive.

Something is undefined
↓
An employee notices and fills the gap
↓
The work somehow gets done
↓
The company misreads this as “the process works”
↓
The next case is left undefined again
↓
The same employee fills even more gaps

The ambiguity is never resolved. It is merely hidden by the cognitive load carried by competent people.

4-2. Work is added, but nothing is removed

A new request arrives.

Create training materials.
Process a large volume of documents under a short deadline.
Add another management sheet.
Plan an event and measure its effect.
Design recruiting measures.
Write standard procedures.
Continue handling inquiries from normal operations.

But nobody decides to stop existing work.

It becomes “this is necessary,” “that is necessary,” and “the previous work must continue too.”

A manager’s job is not merely to issue the instruction to add work.

They also need to decide:

what to stop;
what to postpone;
what to redistribute;
how far quality may be lowered;
whether to renegotiate the deadline.

If those decisions are not made and work is only added, that is not management. It is loading.

4-3. “You can do it, right?” is not delegation

Experienced or capable employees are told:

You can handle something like this, right?
Think for yourself.
Figure it out together.
Take initiative and move it forward.

On the surface, this sounds like giving autonomy.

In practice, however, the following conditions are attached:

the supervisor fixes the conclusion in advance;
objections to that conclusion are not accepted;
no concrete method is provided;
necessary resources are not increased;
existing work is not reduced;
the supervisor retains the right to reject the final result;
if it fails, the employee is blamed for not working out the details.

That is not delegation.

It is keeping decision rights at the top while pushing downward only the responsibility for implementing an unworkable demand.

5. Problems are raised, overridden, and returned only as implementation work

A recurring pattern in the department looked like this:

The employee points out a problem
↓
The manager pushes through their own proposal
↓
No concrete process, priority, or responsibility boundary is defined
↓
Only the implementation work remains with the employee
↓
The employee no longer knows what counts as the correct way forward
↓
More confirmation is needed and movement slows
↓
The employee is judged as “lacking initiative” or “slow to decide”

The important point is that the employee had not remained silent.

They had explained the problem.
They had objected.
They had explained that there was no workable implementation method.

The conclusion still remained unchanged, and only the implementation task came back.

The employee then becomes trapped between contradictory constraints:

following the instruction produces an incoherent result;
explaining that the instruction is flawed does not change the decision;
filling in the gaps independently expands the workload;
changing the approach independently is criticized;
asking for confirmation is labeled a lack of initiative;
if a problem occurs, the employee is required to explain it.

This is not simply poor communication.

The allocation of authority and responsibility is broken.

6. The request to measure a company sports day by “how many work hours it saves”

One case made this structure especially visible: measuring the effects of an internal event.

Management asked staff to express the effects of a sports day or social event in numerical terms such as reductions in working hours.

The employee raised the obvious issues:

How could a causal relationship between the event and reduced work time be proven?
What exactly should be measured?
Over what period should before and after results be compared?
How should workload, busy periods, and other variables be separated?
Why should an initiative intended to promote interaction be evaluated only by time savings?
Would weakly supported figures damage the credibility of the entire proposal?

The conclusion—“produce a number”—was maintained.

When the employee asked how to measure it, the design work was pushed back to the operational side as “something everyone should think about.”

The structure of the request was therefore:

Management decides in advance which number it wants
↓
The employee points out causal and measurement problems
↓
No concrete answer is given to those problems
↓
The demand to produce a number remains
↓
The employee must build the hypothesis, research, calculation, and materials
↓
Management retains final authority to accept or reject
↓
If it does not work, the employee is blamed for insufficient preparation

Ordinary effect measurement proceeds in this order:

Define the problem
→ define the purpose of the intervention
→ define the expected change
→ choose indicators
→ design measurement
→ compare

In this case, the order was reversed:

We want a number that can be used in a report
→ connect it to an existing event
→ construct the causal story afterward
→ make the employee figure out how to make it work

This is a post-hoc KPI.

At the same time, everyday sources of wear were being left untouched:

people applying pressure;
instructions changing;
responsibility being shifted;
uninterrupted work time not being protected;
calls and interruptions being unevenly distributed;
workload not decreasing;
consultation turning into an evaluation of the person;
continued departures from a particular department.

Trying to raise engagement through events in that environment reverses the order of interventions.

What an employee who is close to leaving needs first is not a sports day.

It is a workplace where ordinary work can function, consultation does not become a personality assessment, and unreasonable work can actually be reduced.

Events can be useful after the foundation is sound. They are not a substitute for everyday damage control.

7. Is this a “double bind”?

In everyday language, the following situations feel like a double bind:

Think for yourself and act proactively.
But do not act without permission.

Ask when you do not understand.
But then be told, “You do not even know that?”

Involve the people around you.
But do not make the issue bigger.

Move quickly.
But confirm everything in advance.

Deliver results.
But we will not give you the purpose, priorities, authority, or decision criteria.

Produce a plan.
But after we hold up the decision for a long time, we will suddenly demand it under a short deadline.

Whatever action the employee chooses can later be rejected under a different standard.

In organizational research, however, it is more accurate to describe this not with a single concept but as a combination of the following conditions.

7-1. Role conflict

The employee receives incompatible expectations or instructions at the same time.

decide for yourself / do not decide without permission;
move fast / confirm everything;
improve the process / do not depart from precedent;
involve others / do not talk to other departments.

7-2. Role ambiguity

The employee does not know what is expected, how far their responsibility extends, or what counts as completion.

purpose is unclear;
scope is unclear;
priority is unclear;
completion criteria are unclear;
decision-maker is unclear;
evaluation criteria appear after the fact;
exception handling is unclear.

For decades, research has examined relationships between role ambiguity and role conflict and outcomes such as job attitudes, intention to quit, actual turnover, and withdrawal behavior. Meta-analytic evidence also shows a negative relationship between role ambiguity and job performance.[^3][^4]

7-3. High demands and low control

The workload and responsibility are large, but the employee has little room to decide.

Karasek’s demand-control model proposes that the combination of high job demands and low decision latitude increases psychological strain.[^5]

This workplace was not simply busy.

Employees also faced conditions such as:

being unable to change the conclusion;
being unable to negotiate deadlines;
being unable to stop existing work;
being unable to increase staffing;
decision-makers not responding;
still being held responsible for outcomes.

The imbalance between demands and control was therefore substantial.

7-4. Intimidating or abusive supervision

Beyond operational criticism, the following behaviors were also added:

attacks on character;
fixed judgments about suitability;
criticism that someone is “too soft”;
comparisons such as “when I was young”;
evaluation of attitude;
pressure through voice or atmosphere;
converting a work problem into a problem with the employee’s way of thinking;
publicly singling out one employee for extensive corrections or rejection.

Research on abusive supervision has examined relationships with job dissatisfaction, emotional exhaustion, intention to quit, withdrawal behavior, and other outcomes.[^6]

Once these factors overlap, difficulty taking action cannot be explained only as a personality problem.

8. It is not “lack of initiative”; action itself has become risky

One employee was not someone who had been passive from the beginning.

They had been doing things such as:

spotting problems;
proposing alternatives;
preparing materials ahead of schedule;
checking with relevant people;
standardizing work;
improving efficiency with VBA and other systems;
converting training into video, scripts, and procedures;
organizing undefined requirements;
going directly to coordination partners for answers.

But experiences accumulated when going through the direct supervisor:

Decide independently
→ get told you acted without permission

Confirm in advance
→ get told to think for yourself

Point out a problem
→ get the issue reframed as your suitability, attitude, or mindset

Coordinate with others
→ get told you are being one-sided or making the issue too large

Submit materials
→ have new criteria added afterward

Return a decision to the supervisor
→ receive no response and the case stalls

Approach the deadline
→ get told you are slow and lack initiative

In such an environment, it is rational for an employee to:

leave a written record instead of relying on verbal exchanges;
return decisions to the supervisor;
include relevant people in email;
increase advance confirmation;
avoid deciding alone;
lock in already approved content;
return out-of-scope work to its source;
reduce unnecessary proposals.

From outside, it may look as if the employee has become slower.

More accurately, the employee has accumulated defensive overhead because moving quickly increases the risk of being rejected later.

Initiative did not disappear.

Taking initiative became a risk to evaluation, responsibility, and personal dignity.

9. Work moves forward when the supervisor is bypassed

There is a very clear phenomenon that helps identify this problem:

Cases stall when they are sent to the direct supervisor, but move when the employee asks someone who actually knows the work or a related department.

The employee and the case are the same.

Only the consultation route changes.

When routed through the direct supervisor

understanding the content takes time;
the discussion expands into abstract coaching;
new issues are added;
no decision is made;
the work is sent back;
progress stops while waiting for a response;
in the end, it becomes the subordinate’s failure to explain.

When the practitioner or related department is asked directly

the facts become clear;
the rule-based answer is provided;
the responsibility boundary is confirmed;
the required process becomes clear;
the case can move forward quickly.

This difference matters.

If the employee’s aptitude were the problem, changing only the person asked would not suddenly make the work proceed.

If the case itself were extremely difficult, it would stall no matter who was asked.

But if it stalls only when routed through a particular manager, that manager may be the bottleneck for approval, information, and decision-making.

The problematic manager demands “initiative” from subordinates while stopping decisions at their own desk.

Then delays caused by the manager’s waiting time are returned as a problem with the subordinate’s way of working.

The supervisor does not decide
↓
The case stalls
↓
The deadline approaches
↓
The subordinate is told “you are slow”
↓
Only the record of “coaching” the subordinate remains

This is a structure that converts managerial delay into subordinate delay.

10. The same structure appears with another team member

Another member was not someone who silently swallowed every problem.

When necessary, they voiced strong opinions.
They pointed out impossible conditions.
They questioned the proposed method.

But management pushed through its own conclusion without changing it.

No concrete method was then provided.

The team member was left with this situation:

they had pointed out the problem;
the conclusion still did not change;
they had to figure out how to implement that conclusion;
they might have to explain any failure;
they had no authority to change the decision;
they no longer knew what counted as the correct action.

The problem was not that they “could not speak strongly enough.”

They had spoken strongly, but the decision remained unchanged and only uncertain implementation work came back to them.

In this condition, it is plausible that smiling decreases, movement slows, confirmation increases, and fatigue begins to show.

Even if the recorded direct reason for resignation is something else, workplace stress can erode the reasons to stay. When another factor appears, leaving then becomes a rational choice.

The company’s records may say “family circumstances,” “personal reasons,” or “career reasons.”

But those may only be the final trigger. The workplace structure that depleted the person’s remaining willingness to stay is not written in the resignation category.

11. Selecting for people who “smile and nod” and losing those who mature

A hypothesis can be formed that the manager in this department preferred people who tended to:

listen with a smile;
agree first;
avoid strong objections;
accept ambiguous requests;
manage the other person’s mood;
return emotional reassurance;
say “Understood” and get moving.

For convenience, call this being expected to play a “mothering role.”

Cooperation, listening, and receptiveness are of course important at work.

The problem begins when these qualities become not measures of job competence, but measures of how easy the employee is for the supervisor to handle.

During the newcomer stage, employees may accept things with a smile because they cannot yet see the contradictions in the work.

As they learn the job, questions appear:

Why are we doing this task?
What counts as finished?
Is this volume realistic?
Can this metric actually measure the effect?
Who decides?
Why is existing work not being reduced?
Why does consultation turn into a discussion of personality?

When the employee raises those questions, they may be judged as “different from when they joined” or “less cooperative than before.”

In reality, the person did not necessarily change.

They learned enough to see the work.

The Attraction-Selection-Attrition model proposes that the composition of organizations changes as people are attracted, selected, and leave in ways that create similarity.[^7]

In this case, that mechanism may be functioning in reverse:

Prefer receptive people who do not object
↓
Relations are good during the newcomer stage
↓
As employees learn the work, they notice contradictions
↓
When they develop opinions, they are labeled difficult
↓
They transfer or resign
↓
New people who do not yet know the problems are hired again

The department then tends to retain one of two groups:

newcomers who do not yet know the problems;
people who have stopped talking about the problems.

That is not healthy retention.

It is maintaining surface-level obedience by losing the people who have developed.

12. Once personal attacks and pressure are added, the issue is no longer just bad work design

If the only problem were vague instructions, clearer work definitions could solve it.

But when a person who raises a problem receives comments such as:

your suitability is the problem;
your attitude is bad;
you are soft;
you are young;
things were tougher when I was your age;
you are one-sided;
you are not involving others enough;
you are not thinking;
this is a mental issue;

the situation moves to another level.

The real questions should be:

Are the instructions contradictory?
Is the workload realistic?
Do role and authority match?
Is the manager returning decisions?

Instead, they are transformed into:

Is this employee suited to the job?
Is this employee’s thinking correct?
Is this employee mature enough as a professional?

That transformation makes it difficult to discuss the operational contradiction itself.

Employees learn, “If I point out a problem with the instructions, an assessment of my personality begins.”

Reporting and consultation then decrease.

From management’s perspective, it may look as though “the person never said anything.”

But that is not what happened.

Speaking up had become dangerous.

13. Why employees may want to talk to HR or a consultation desk instead of the direct supervisor

When a subordinate goes around them and speaks to HR or higher management, a supervisor may think:

I wish they had told me directly if they had a request.
I cannot know if they do not tell me.

But the fact that someone is the direct supervisor can be exactly what makes them difficult to approach.

After the consultation, the employee still has to see that person every day.
That supervisor assigns work.
That supervisor evaluates them.
If no transfer occurs, the relationship continues.

It is therefore rational for the employee to want to speak to another department, HR, a harassment consultation desk, or a higher-level manager.

If the direct supervisor is part of the causal problem, the direct route cannot function as the consultation route.

One employee kept a chronological Excel log of harassment-related incidents.

The log did not contain emotions alone.

It recorded:

date and time;
place;
the actual words spoken;
instructions given;
what happened before and after;
how the employee responded;
effects on work;
effects on physical and mental condition;
recurrence of the same pattern;
the response the employee wanted.

Then, instead of going through the direct supervisor, the employee used:

internal consultation desk
→ higher-level manager
→ transfer

This was not rebellion against the supervisor.

It was simply refusing to treat the source of the problem as the consultation channel.

Japan’s Ministry of Health, Labour and Welfare requires employers, as part of measures to prevent workplace power harassment, to communicate policy, establish consultation systems, confirm facts, respond appropriately to victims and alleged perpetrators, prevent recurrence, and prohibit disadvantageous treatment because an employee sought consultation.[^8]

The system does not assume that employees must speak only to their direct supervisor.

14. A transfer can be the right answer for the person without preventing recurrence

Separating an employee from a problematic manager or department can be an effective form of protection.

A transfer can:

reduce contact;
change the evaluator;
move the employee to a department where their strengths fit better;
stop physical and psychological depletion;
retain the person inside the company instead of losing them to resignation.

So a transfer is not “running away.”

But if the company stops at the transfer, the cycle continues:

Employee A is worn down
↓
A is transferred
↓
A is saved
↓
The problematic manager and undefined work remain
↓
Employee B is worn down
↓
B resigns
↓
A newcomer is hired

A transfer is emergency transport.

Emergency transport may be necessary, but it does not put out the fire.

After the transfer, the company should examine the original department for:

repeated resignations, transfers, or absences under the same manager;
tenure and hiring route of people who left;
role ambiguity and workload;
frequency of instruction changes;
waiting time for decisions;
references to personality and suitability;
employee-side evaluation of managers;
retaliation or relationship deterioration after consultation;
recurrence after new staff are hired.

15. Turnover spreads to coworkers and increases the burden on those who remain

When one person resigns, their work does not disappear.

It shifts to those who remain:

normal operations;
unresolved cases;
handover;
training new employees;
inquiry handling;
explanation of past background;
repair of relationships with other departments.

The remaining employees’ workload therefore rises.

Their psychology changes at the same time:

Is this a workplace bad enough that this person chose to leave?
Do I really need to stay here?
Maybe leaving is acceptable.
Maybe speaking up will not change anything.

Research across multiple organizations has shown “turnover contagion,” in which coworkers’ job embeddedness and job-search behavior influence voluntary quitting among surrounding employees.[^9]

Meta-analytic evidence also shows a negative relationship between organization- or workplace-level turnover rates and organizational performance.[^10]

In a small department, the effect of one departure is large.

One person leaves
↓
The remaining employees receive more work
↓
Training overlaps with normal operations
↓
Slack disappears
↓
Improvement and standardization become impossible
↓
Work becomes even more person-dependent
↓
The next person leaves

Turnover itself strengthens the conditions for the next round of turnover.

16. The quality of a supervisor really can affect productivity and retention

If managers are treated as mere communication relays, this problem is easy to underestimate.

But supervisor quality affects team productivity and retention.

Lazear, Shaw, and Stanton estimated that replacing a bottom-10% boss with a top-10% boss produced an increase in total output greater than adding one worker to a nine-person team, and they also found that employees with better bosses were less likely to leave the company.[^11]

That result cannot be mechanically applied to every workplace.

But at minimum, it supports asking:

Is the subordinate slow?
Or does the entire team become slow when work passes through the supervisor?

Manager evaluation should not focus on whether the manager personally looks busy.

It should ask whether the manager:

returns decisions;
sets priorities;
removes work;
defines roles;
creates clear boundaries with other departments;
allows subordinates to raise problems safely;
retains and develops the same employees over time;
has created a situation where work only moves when subordinates bypass them.

17. What companies should measure

Company-wide turnover alone cannot reveal problems like this.

17-1. Talent outflow

turnover rate by department;
turnover rate by direct supervisor;
retention of mid-career hires at 3 months, 6 months, 1 year, 2 years, and 3 years;
voluntary turnover among permanent employees in their thirties;
number of transfer requests;
number of actual transfers;
leaves of absence and long-term absences;
use of consultation desks;
requests to change the person responsible for training.

17-2. Quality of turnover

departure of highly rated or core operational employees;
departure of people who take a long time to replace;
departure of people who had been proposing improvements;
departure of people who hold substantial operational knowledge;
departure of trainers for newcomers;
departure of people with specific qualifications or experience.

Do not look only at how many people left. Look at who left.

17-3. Work design

number of undefined tasks;
tasks with no clear owning department;
tasks with unclear completion criteria;
number of existing tasks removed when new work is added;
frequency of instruction changes;
number of reversals after approval;
proportion of instructions given only verbally;
number of duplicate entries of the same information;
number of tasks sent back from other departments.

17-4. Management bottlenecks

time from a request for a decision to the answer;
number of cases stalled waiting for supervisor approval;
number of cases solved by bypassing the supervisor;
decisions made per hour of meetings;
number of cases stopped at “we will consider it”;
clarity of reasons when work is returned to subordinates;
percentage of cases with explicit priorities;
percentage of added tasks accompanied by a decision to stop or postpone other work.

17-5. Psychological safety and harassment risk

whether employees feel they can raise problems without disadvantage;
whether they can disagree with a supervisor’s instructions;
whether feedback addresses work rather than personality or suitability;
whether relationships deteriorate after consultation;
whether consultation routes other than the direct supervisor actually work;
whether behavior changes after harassment training;
whether the same behavior has recurred after 3 months or 6 months.

18. What companies should fix first

18-1. When adding new work, decide what will be reduced

Adding instructions is not management by itself.

new work
+ existing work
+ interruptions
+ training
+ meetings

If all of these are maintained, the system becomes unprocessable.

When assigning new work, a manager should decide at the same time:

which work will stop;
which work will be postponed;
which work may be done at lower quality;
which work will be given to someone else;
which deadlines will be renegotiated.

18-2. Define purpose, completion criteria, authority, and the decision-maker

At minimum, put the following in writing:

Purpose:
Deliverable:
Mandatory conditions:
Optional conditions:
Deadline:
Priority:
Scope of responsibility:
What the person in charge may decide:
Midpoint for review:
Final decision-maker:
Person responsible for failures or exceptions:

18-3. Set response deadlines for managers

If subordinates have deadlines, managers should have deadlines for decisions.

This prevents the pattern where a supervisor does not decide for days and then returns the matter under a sudden short deadline.

Define:

how many business days for ordinary cases;
how many hours for urgent cases;
where a case escalates if the manager cannot decide;
what automatic escalation occurs if there is no response.

18-4. Include “ability to retain people” in manager evaluation

Evaluate not only short-term output but also:

retention of subordinates;
ramp-up of mid-career hires;
consistency of instructions;
role definition;
decision speed;
workload reduction;
feedback from subordinates;
concentration of transfers, leave, and consultations;
presence or absence of personal attacks or intimidation;
standardization and handover;
whether successors are developing.

18-5. Choose consultation interviewers carefully

Even if an employee comes to HR expecting a neutral and safe discussion, the consultation desk becomes a source of secondary harm if the interviewer shifts the conversation to:

attitude;
softness;
age;
suitability;
how things were “when I was young”;
“just come to work first”;
“harsh, but correct.”

The interviewer’s job is not to assess the employee’s character.

It is to examine:

facts;
chronology;
operational impact;
whether continued work is possible;
the employee’s requested response;
temporary protection;
separate interviews with involved parties;
prevention of retaliation;
transfer or contact restrictions;
investigation of the causal side.

18-6. Reduce everyday wear before adding events

Sports days, drinking parties, festivals, social gatherings, and internal recreation are not always meaningless.

But the company should first reduce:

pressure;
personal attacks;
blame shifting;
unclear instructions;
excessive workload;
lack of uninterrupted work time;
concentration of calls and interruptions;
situations where consultation changes nothing.

Adding events while daily wear remains can simply become additional work for already exhausted employees.

19. Defensive practices that actually helped on the individual side

An individual cannot solve a structural problem alone.

But some operating practices can reduce damage.

19-1. Change the decision route

When a case stalls if it is sent to the direct supervisor, and within the limits of authority and internal rules, ask:

the department that owns the policy;
the practitioner;
the previous person responsible;
the related department;
the final approver.

This is not acting without permission. It is connecting to the correct source of information.

19-2. Lock in what has already been approved

Leave agreements in writing.

I will proceed using the content previously shared and approved as the baseline.
If there is a change, I will confirm the changed point, the reason, and the effect on the deadline.

This helps prevent the entire matter from later being treated as if it had never been approved.

19-3. Turn ambiguity into questions

I can handle this.
Please confirm the scope, reference date, submission date, mandatory items, and final decision-maker.

Instead of saying “I do not understand,” convert missing definitions into concrete items.

19-4. Return work to its source

Do not turn sales calls, packages for other departments, visitors with no clear owner, facility inquiries, or unresolved meeting issues into your own permanent responsibilities when the source or owning department lies elsewhere.

source
→ first contact point
→ responsible department

Return the work there.

Do not permanently fill undefined gaps with personal goodwill.

19-5. Log facts, inferences, and emotions separately

Facts:
what was actually said, instructions, date and time, participants

My response:
what I did in advance, confirmations, alternatives, submissions

Impact:
delay, rework, workload, health, intimidation

Inference:
possible interpretations of the other person’s intent

Emotion:
fear, anger, exhaustion, helplessness

This separation makes it harder for the situation to be dismissed as “you are only complaining because you dislike the person.”

19-6. Avoid close-range combat

If someone turns contradictions into personality arguments, defeating them verbally on the spot is unlikely to help.

put the issue in email;
include relevant parties;
define boundaries in writing;
do not enter an emotional verbal fight;
escalate to higher management when necessary;
pursue transfer or job change in parallel.

This is not avoidance. It is resource allocation.

20. The core structure revealed by this case

The overall cycle can be summarized as follows:

There is a great deal of undefined work
↓
Capable employees fill the gaps
↓
The company mistakes that gap filling for a standard operating model
↓
New work is added without reducing old work
↓
When problems are raised, management pushes through its conclusion
↓
Only the implementation details and operational work return to the employee
↓
The employee carries outcome responsibility without decision rights
↓
Confirmation increases and movement slows
↓
Management reframes the issue as “initiative, fit, or attitude”
↓
Personal attacks and pressure close off consultation routes
↓
Employees bypass the supervisor, transfer, or resign
↓
New employees are hired
↓
Once those newcomers learn the work, they notice the same contradictions
↓
They leave again

In this cycle, a manager may feel that “subordinates never develop.”

But the opposite may be true.

They notice the contradictions because they developed, and they can leave because they developed.

The company can say “mid-career hires these days are weak,” “the person was not a fit,” or “they left for family reasons.”

But if people at the same stage of tenure repeatedly leave the same department under the same supervisor, individual explanations are not enough.

Conclusion

A company-wide turnover rate of 4% can be useful as a high-level indicator.

But it cannot reveal complete replacement of personnel inside a specific department.

What was happening in this department was not merely a personality mismatch.

It included:

insufficient onboarding for mid-career and mid-level employees;
absorption of undefined work;
adding work without removing work;
management that forces through a conclusion and pushes implementation back downward;
contradictory instructions;
high demands with low control;
slow managerial decision-making;
returning delays caused by management waiting time to subordinates;
converting operational concerns into judgments about suitability, attitude, and personality;
inability to use the direct supervisor as a safe consultation route;
leaving the causal side untouched after transfers or resignations;
hiding the problem temporarily by hiring another newcomer.

When employees begin to increase confirmation, avoid making decisions, and bypass their supervisor, do not immediately conclude that they “lack initiative.”

First ask:

Has the environment become one where acting independently gets rejected, asking for confirmation gets treated as incompetence, sending a decision to the supervisor causes it to stall, and only the responsibility for the result comes back?

If the people who understand the work are the ones who keep disappearing from a department, the key question is not whether those who left were suitable.

It is whether the department and its management system are suitable for developing and retaining people.

The people left at the end are not necessarily proof of a healthy organization.

They may simply be newcomers who have not yet discovered the problem, and employees who have stopped speaking about it.

References and sources

[^1]: Bauer, T. N., Bodner, T., Erdogan, B., Truxillo, D. M., & Tucker, J. S. (2007). Newcomer adjustment during organizational socialization: A meta-analytic review of antecedents, outcomes, and methods. Journal of Applied Psychology, 92(3), 707–721.
https://pubmed.ncbi.nlm.nih.gov/17484552/

[^2]: Japan Institute for Labour Policy and Training (2022), research on mid-career job changes, skills development, and career formation, including materials from the Labour Policy Forum. The research examines relationships between organizational socialization or retention intentions and interpersonal support for mid-career hires, formal support, and onboarding measures.
https://www.jil.go.jp/institute/reports/2022/documents/0215.pdf
https://www.jil.go.jp/event/ro_forum/20220726/resume/01-kenkyu-fujimoto.pdf

[^3]: Jackson, S. E., & Schuler, R. S. (1985). A meta-analysis and conceptual critique of research on role ambiguity and role conflict in work settings. Organizational Behavior and Human Decision Processes, 36(1), 16–78.
https://doi.org/10.1016/0749-5978(85)90020-2

[^4]: Tubre, T. C., & Collins, J. M. (2000). Jackson and Schuler (1985) revisited: A meta-analysis of the relationships between role ambiguity, role conflict, and job performance. Journal of Management, 26(1), 155–169.
https://doi.org/10.1177/014920630002600104

[^5]: Karasek, R. A. (1979). Job Demands, Job Decision Latitude, and Mental Strain: Implications for Job Redesign. Administrative Science Quarterly, 24(2), 285–308.
https://www.scienceopen.com/document?vid=60c75f8b-4a55-43f8-9c84-1e4ba12e7325

[^6]: Bhattacharjee, A., & Sarkar, A. (2022). Abusive supervision: a systematic literature review. Management Review Quarterly.
https://pmc.ncbi.nlm.nih.gov/articles/PMC9376912/

[^7]: Schneider, B. (1987). The People Make the Place. Personnel Psychology, 40(3), 437–453.
https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1744-6570.1987.tb00609.x

[^8]: Japan’s Ministry of Health, Labour and Welfare, materials on workplace harassment and the “No Harassment” workplace support portal. They explain the three elements and six representative categories of workplace power harassment, consultation systems, fact-finding, response measures, recurrence prevention, and the prohibition of disadvantageous treatment for seeking consultation.
https://www.mhlw.go.jp/content/11900000/001338359.pdf
https://www.no-harassment.mhlw.go.jp/foundation/harassment_list/

[^9]: Felps, W., Mitchell, T. R., Hekman, D. R., Lee, T. W., Holtom, B. C., & Harman, W. S. (2009). Turnover Contagion: How Coworkers' Job Embeddedness and Job Search Behaviors Influence Quitting. Academy of Management Journal, 52(3), 545–561.
https://doi.org/10.5465/AMJ.2009.41331075

[^10]: Park, T.-Y., & Shaw, J. D. (2013). Turnover rates and organizational performance: A meta-analysis. Journal of Applied Psychology, 98(2), 268–309.
https://pubmed.ncbi.nlm.nih.gov/23244224/

[^11]: Lazear, E. P., Shaw, K. L., & Stanton, C. T. (2015). The Value of Bosses. Journal of Labor Economics, 33(4), 823–861.
https://www.nber.org/papers/w18317

AdBooks on this topic

Books on management
Book search results for management, the field this article belongs to.
Amazon (search)

Links to store search results. It is not a recommendation of a specific product or seller; no prices, stock or ratings are shown. We may earn a commission if you buy through these links (the price you pay does not change). Commissions do not affect which books are listed. As an Amazon Associate I earn from qualifying purchases.