There is no correct number of direct reports. There is a number for a given kind of work, and organisations routinely apply one figure across situations that require very different ones.
What is reliably true is that something specific breaks first when span is too wide, and knowing what it is makes the problem diagnosable before it becomes attrition.
For a practical reference on employee attendance tracking software, Monitask provides this page.
What span actually has to cover
A manager's time per report is not one thing. It is several, and they scale differently.
One-to-ones. Roughly fixed per person. Thirty minutes weekly across twelve people is six hours before anything else.
Reviewing work. Scales with output, not headcount.
Career development. The most compressible and the first thing cut, which is why it disappears silently.
Hiring and onboarding. Lumpy and expensive.
Performance problems. Not proportional to headcount at all — one difficult situation can consume a manager for months.
Coordination outward. Represents the team to the rest of the organisation, and grows with team size and with organisational complexity.
The last two are why apparently reasonable spans fail suddenly. A manager comfortable with ten reports is not comfortable with ten reports and one performance case.
What determines the workable number
How much direction the work needs. Experienced people doing well-defined work need far less. A team of new joiners on ambiguous work needs far more.
How interdependent the work is. Independent workstreams coordinate themselves. Highly interdependent work pushes coordination onto the manager.
How much the manager also does hands-on. A player-manager has less time, and the hands-on work is invariably what gets prioritised because it is more visible.
Rate of change. Stable work needs less management than work being redefined quarterly.
Distribution. Distributed teams need more deliberate management effort, because none of it happens incidentally. See hybrid without a two-tier team.
Broadly: high-autonomy, stable, independent work supports wide spans. Ambiguous, interdependent, changing work with junior people supports narrow ones. Applying the same target to both is the error.
What breaks first, in order
Useful because these are diagnosable before anyone complains.
1. Development conversations stop. Not cancelled — quietly deprioritised. One-to-ones become status updates because status is urgent and development is not.
2. Feedback becomes annual. Correction stops happening in the moment, which is where it works. See how learning at work actually happens.
3. The manager stops knowing what people are working on. They know what is reported, which is a different thing.
4. Hiring quality falls. Insufficient time to do it properly, so it gets rushed.
5. Problems are noticed late. Both performance problems and people about to leave.
6. The strongest people go first. They need the least management, receive the least, and have the most options.
If you recognise the first two in a team, the span is too wide, whatever the number is.
The failure at the other end
Narrow spans have their own costs and are less discussed.
Over-management. A manager with three reports has time to involve themselves in everything, and frequently does.
Too many layers. Narrow spans multiply management levels, which lengthens decision paths and adds distortion at each step. See decision rights.
Expensive. More managers, doing less managing each.
Career distortion. When management is the only progression route and spans are narrow, you create manager roles to promote people rather than because the work needs them — and then you have managers with three reports and no other job. See the IC and management fork.
Fixing a span that is too wide
Not automatically by splitting the team, which is the reflex.
Reduce what the manager has to do per person. Better-defined work, clearer decision rights, and more autonomy reduce the management load without changing the structure. Often this is the whole fix.
Introduce a lead role without a management layer. Someone who coordinates the work without owning careers. Effective, and it needs to be explicit about what it does and does not include, or it becomes an unpaid manager job.
Split, if the work genuinely divides. Splitting a team whose work is interdependent creates a coordination problem worse than the span problem.
Remove the hands-on load if the manager is also delivering. This is the most common actual cause and the least often addressed, because the delivery work is what everyone can see.
Deal with the performance case. If one situation is consuming a third of the manager's capacity, the span is not the problem.
The number worth tracking
Not span itself. The proportion of scheduled one-to-ones that actually happened in the last quarter.
It is available from any calendar system, it is not gameable in any useful way, and it drops before anything else does. A team where a third of one-to-ones were cancelled has a capacity problem that will appear as attrition in about two quarters.
For broader public guidance and background, consult the OECD.