A manager's calendar records the span-of-control problem before the org chart does.
The 1:1s move to every other week. Feedback waits until a project is already off course. The manager joins a meeting late because another direct report needed help, then spends the evening finishing the individual work the organization never took away.
On the org chart, this person has 12 direct reports. The number looks perfectly reasonable. The work does not.
That is the problem with asking how many direct reports is too many. It sounds like a counting question, so organizations go looking for a number: 5, 7, 8, perhaps 10 if the manager is particularly organized. The search usually ends with a benchmark presented as if every manager performs the same work under the same conditions.
There is no universal ideal span of control. A manager supervising a stable, experienced team doing similar work may support more people than a player-coach leading new hires through uncertain, interdependent work. Both boxes say "manager." Their capacity has almost nothing else in common.
The useful question is not how many names fit beneath a box. It is how much management the work requires and whether anyone has protected the time to provide it.
What the evidence says
- Gallup's 2026 analysis found a median span of 6 and an average of 12.1 among U.S. managers in 2025. A minority of very large teams pulled the average upward.
- Its global meta-analysis covered 92,252 teams in 104 organizations, 26 industries, and 46 countries. It did not find a single optimum that worked everywhere.
- A small hospital study found that department complexity and headcount were related but not interchangeable, especially in intensive care units.
- Team engagement, manager capability, individual-contributor workload, feedback, work setting, and the nature of the work all changed the picture.
- In Gallup's U.S. data, 97% of managers also did individual-contributor work, with a median of 40% of their time spent on it.
- Benchmarks are a starting signal. The real capacity test is whether the manager can perform the management work their team needs.
The magic number is attractive because judgment is expensive
An organization with 200 managers does not want 200 separate debates about span of control. It wants a rule. A rule fits in a planning model, survives a finance meeting, and makes a reorganization look consistent.
That does not make benchmarks useless. It makes them dangerous when promoted from reference point to answer.
In January 2026, Gallup published an analysis of span of control using U.S. manager data and a large global meta-analysis of engagement and performance. The median U.S. manager or leader had about 6 reports. The average had risen to 12.1 in 2025, up from 10.9 in 2024, because a smaller share of very large teams pulled the mean upward. About 66% managed fewer than 10 people, while 13% managed 25 or more.
Those numbers describe a distribution. They do not draw a line between good and bad management.
Gallup grouped spans as small at 4 to 9, medium at 10 to 19, and large at 20 or more for parts of its analysis. Across 92,252 teams, large teams could perform well when engagement and management conditions were strong. Small teams could perform poorly. The effects varied with manager talent, workload, work location, and the team's environment.
Gallup is a workplace analytics and consulting company, and some of the underlying measures are proprietary. Its analysis is broader and more useful than a recycled "rule of 7," but it should still be read for what it is: evidence against a universal number, not permission to replace one universal number with Gallup's preferred management system.
A smaller peer-reviewed pilot reached the same narrower conclusion from a very different setting. Researchers compared headcount with a department-complexity measure for 41 nurse managers across 9 hospitals. The measures were moderately related, but 54% of intensive care departments received different classifications depending on whether the researchers counted direct reports or measured complexity.
That study is too small and specialized to set a ratio for a software company, a school, or even another hospital. It does show why headcount and managerial responsibility are not interchangeable measures. Two jobs with the same number of reports can carry different coordination, risk, staffing, and operational demands.
The absence of a magic number leaves leaders with a less convenient task. They have to understand the work.
Span of control is a budget of attention
Every direct report creates recurring management work.
There is the obvious work: 1:1 conversations, feedback, performance reviews, goal setting, development, hiring, compensation, conflict, and leave. Then there is the work that rarely appears in a capacity model: reading context before a conversation, noticing a pattern across projects, following up on a promise, making an introduction, preparing a difficult message, and creating enough psychological room for someone to say what is actually wrong.
Some of that work scales. A manager can explain a shared priority once to a team. Some does not. A career conversation with 1 person cannot be efficiently batched with 11 others unless the organization has invented a very bleak form of group therapy.
Assume a manager has 10 reports and holds a 30-minute weekly conversation with each. The meetings consume 5 hours. Preparation, notes, and follow-up can easily take another 2 or 3. Add team meetings, hiring, planning, cross-functional coordination, performance work, and unexpected problems. The weekly budget disappears before the manager begins their individual contribution.
The arithmetic is not an argument that 10 is too many. It is a demand that the arithmetic be done.
Organizations often expand a manager's span because the boxes look administratively efficient. They count salary lines saved by removing a layer. They are less likely to count the coaching delayed, decisions centralized, risks missed, or work transferred upward because nobody had time to build judgment lower down.
Team dependability suffers in this environment for predictable reasons. Expectations get thinner. Invisible work stays invisible. Feedback arrives after the deadline. The manager appears to be the bottleneck, but the system assigned them more relationship and decision work than the week can hold.
The same number can describe 2 different jobs
Compare 2 managers with 8 reports.
The first leads experienced account managers who own separate portfolios, use established systems, and make similar decisions. The team members can help one another, the work is measurable, and exceptions are infrequent. The manager spends most of their time coaching, allocating resources, and handling the unusual cases.
The second leads a new cross-functional product group. Half the people joined in the last 3 months. The work is technically uncertain, roles overlap, priorities change, and nearly every decision affects another team's roadmap. The manager still writes production code for 2 days each week.
The org chart records 8 for both. It conceals the difference between them.
At least 6 features change the management load:
- Experience and stability. New people and changing teams need more context, feedback, and relationship building.
- Task similarity. Repeated, comparable work is easier to coordinate than 8 unrelated specialties.
- Interdependence. When each person's work changes everyone else's, the manager carries more integration and conflict work.
- Uncertainty. Novel work creates more decisions, exceptions, and learning.
- Geography and time. Distributed work reduces shared context and can stretch conversations across the day.
- Support and systems. Clear goals, good information, administrative help, and healthy peer networks reduce avoidable management load.
These features can also move. A manageable team becomes demanding during rapid hiring, a product launch, a merger, or a period of poor performance. A temporary span may need temporary support even when the long-term structure is sound.
Capacity is not an enduring trait attached to a manager. It is a relationship between the person, the work, and the system around them.
Player-coaches pay twice
The phrase "player-coach" sounds energetic. In practice, it often means the company wants a full manager and most of an individual contributor for the price of 1 person.
Gallup found that 97% of U.S. managers in its study reported some individual-contributor responsibilities. The median manager spent 40% of their time on that work. Managers above that threshold tended to have smaller teams, and higher individual workloads became more difficult as spans widened.
This should not be surprising. Individual work and management work compete not only for hours but for attention. A technical task may require uninterrupted concentration. A direct report's urgent problem does not care.
The conflict creates 3 common patterns:
- The manager protects production work, and management becomes reactive.
- The manager protects the team, and individual work moves into evenings.
- The manager tries to do both in normal hours, and neither receives enough thought.
Organizations sometimes interpret the resulting strain as a personal efficiency problem. The manager gets a time-management course. Their job remains mathematically ambitious.
Player-coach roles can work. They are often useful in small teams where technical credibility, hands-on contribution, and close coaching reinforce one another. The design becomes unstable when the organization widens the span without removing individual work, or keeps the individual work while quietly increasing the complexity of the team.
Before adding reports, ask which existing responsibilities will stop. If the answer is none, the organization is not changing the span of control. It is increasing the wager.
Great managers do not have infinite range
Manager capability matters. Gallup's research found that people it classified as having stronger management talent sustained engagement better across larger teams than others. That finding is useful and easy to abuse.
One response is to invest in selection and development. Another is to conclude that good managers can absorb almost anything. The first improves the system. The second rewards capability with overload.
Skill changes what a manager can do with time. It does not create more time.
A skilled manager can run a focused conversation, delegate clearly, build peer support, and distinguish a real escalation from one somebody could solve. They may manage a larger span because they reduce unnecessary work and help the team become less dependent on them.
But a manager cannot listen to 2 people at once. They cannot provide thoughtful feedback on work they have had no time to understand. They cannot build trust by being permanently unavailable and then compensate with a particularly efficient quarterly review.
There is also a system effect. When capable managers carry very large teams, the organization can become dangerously dependent on their endurance. The design appears successful until that person takes leave, moves roles, or simply reaches a limit. What looked like an efficient structure turns out to have been a private subsidy from one manager's attention.
Do not use exceptional performance to declare a job normally designed.
Audit management work before changing the chart
A useful span review begins with evidence, not a target ratio.
For each manager, examine:
What management work is required?
List the recurring responsibilities: coaching, feedback, hiring, performance, planning, approvals, stakeholder work, conflict, development, and administration. Add the work created by current conditions, such as rapid growth, low role clarity, or a major change.
How much individual work remains?
Count project ownership, customer delivery, technical production, and operational duties. Do not describe these as "staying close to the work" if they consume half the week. Name the hours.
How dependent is the team on the manager?
Look at which decisions, relationships, and information flow through 1 person. High dependency may reflect weak delegation, but it may also reflect formal authority, inexperience, or a process that requires the manager's approval.
How healthy is the layer below?
Peer support, clear roles, and distributed expertise can make a broader span workable. Role ambiguity makes even a small team expensive to manage because every boundary becomes a conversation.
What evidence shows strain?
Use skipped conversations, delayed reviews, manager after-hours work, regretted turnover, unresolved conflicts, escalation volume, decision time, and team feedback. A neat chart is not contrary evidence.
This audit will not produce a universal number. It will produce something more useful: a reasoned view of where management capacity and management demand no longer match.
The fix is not always another manager
When a span is too wide for the work, adding a management layer is 1 option. It is not the only one.
An organization can:
- remove individual-contributor work from the manager;
- clarify decision rights so fewer choices require escalation;
- reduce administrative work or provide operational support;
- build stronger peer coaching and specialist support;
- separate work with genuinely different needs;
- stabilize priorities so the team spends less time renegotiating them;
- develop an experienced lead without pretending that an unpaid coordination role is a promotion;
- add a manager where the relationship and judgment work truly warrants one.
The right response depends on the source of load. Adding a manager to compensate for unclear ownership may create another box around the confusion. Removing a manager while keeping every approval may leave the same hierarchy operating invisibly.
This is why reorganizations should follow a diagnosis. Span is one design variable inside a larger operating system.
It also interacts with matrix work. A person may have 1 formal manager and several project leaders competing for attention. The manager's span looks modest while the employee experiences multiple lines of authority. Dotted lines carry their own clarity tax.
Count what the number leaves out
The span of control belongs on an org chart. It is a useful structural fact. It just is not a capacity model.
Before deciding that 6 is efficient, 10 is modern, or 20 is admirably flat, count the work hidden beneath the number. Count the conversations, decisions, new hires, exceptions, interfaces, and hours of production expected from the same person. Ask what will stop when another report arrives.
A manager with enough time to manage can help a team become clearer, more capable, and less dependent. A manager with a permanently impossible job becomes an escalation address with a calendar.
The boxes are easy to count. Attention is the scarce part.