Outgrace, home
LeadershipData brief

Span of control: what our research search could and could not find

Four sources we checked hold nothing on how many direct reports a manager can lead. Here is what that means for sizing your layers.

Magnifying glass, seek, detective, looking for, enlarge, zoom
Photo: viarami / Pixabay

We set out to size management layers with research. Span of control is the number of people one manager leads directly. Leaders ask what that number should be because it sets how many layers sit between the chief executive and the front line, and layers cost money and slow decisions. Our researcher searched the World Bank, the National Bureau of Economic Research, Brookings and the Pew Research Center. None of the four sites holds research on span of control, direct-report counts or manager effectiveness. This article therefore cannot give the range a reader would want. It explains what the search found, what follows from it, and how a leader can size layers from their own evidence.

What the data shows

The short answer is that the sources we checked show nothing on this question. Across the four sites the subjects were debt, development finance, AI and jobs, health spending, entrepreneurship, polling and religion. None of these is organisational design, and none measures how many people a manager can lead well. The only item that comes close to work is a Pew Research Center finding that 75% of Democrats and 68% of Republicans predict fewer jobs over the next 20 years, with AI seen as a threat to jobs. That is a view about the labour market in general. It says nothing about the number of people a manager should supervise.

The gap matters because the question is usually asked with confidence. Board papers often carry a target span, sometimes taken from a competitor's chart or from a rule of thumb passed between executives. A target of that kind looks like evidence. When the source is traced back, it is often a habit, a single case or a consultant's preference. Our search does not prove that such targets are wrong. It shows only that the four sources we checked cannot be used to defend any particular number.

What it means for a leader

Without research, the honest starting point is your own data. A span is not a fixed feature of good management. It is a choice about how much one manager can supervise, decide and coach before the work slips. In our experience the right number depends on how routine the work is, how far apart the people sit, and how much of the manager's own week goes to tasks someone else could do. Those facts can be pulled by a finance lead or an HR team within a few weeks. They are more useful than a range taken from a study that never measured your business.

Keep two ideas apart. A span is a count of direct reports. A layer is a place in the reporting chart. They move independently. A company can widen spans and add layers at the same time, and many do, because the chart was drawn in one period and the work changed in another. Before any redesign, check which of the two you are actually trying to change. Adding a layer to fix a slow decision rarely works unless the decision rights change with it.

Define effectiveness before you count anything. If effectiveness means output, measure output for each team. If it means keeping good people, measure leavers and internal moves. If it means quality of decisions, pick a sample of decisions and review them. A span that suits one definition may be wrong for another. Any study you later find will have made the same choice, so check the definition it used before you compare its number with yours.

Start by counting, not guessing. List every manager, the number of direct reports, and the number of layers between each person and the chief executive. Then mark the managers whose direct reports ask them for approval more often than they ask for coaching. A sensible test is whether a manager spends most of the week on decisions that a written rule could settle. If so, the span may be too narrow for the work, or the decision rights are unclear. In either case the problem is not solved by adding a layer.

Two more checks help. Look at decision time: how many days pass between a request and an answer, and at which layer the delay sits. Look at managers' own calendars for a typical fortnight. A manager who leads four teams and spends a day a week in one approval meeting has a design problem, not a people problem. Both checks produce numbers a board can read, and neither depends on an outside figure that may not apply to a company of your size.

What leaders get wrong

  • Copying a span from another company's organisation chart. A chart shows boxes, not the work done inside them.
  • Setting one span for the whole company. A customer-service floor and a research team seldom need the same number.
  • Treating a new layer as the answer to slow decisions. A layer adds a handoff and removes one only if the decision rights change.
  • Widening spans to cut cost without moving approvals. The manager simply acquires more names and the same queue.

Most of these errors share one cause. The chart was drawn before anyone wrote down what each layer is for. When the decisions are unclear, every number on the chart is arguable, and the debate becomes a contest between the people with the most influence. Writing the decisions down first removes much of that argument, and it does so without any outside figure at all.

What to do about it

Work in three steps before the next board cycle. First, write down the decisions each layer may take alone, and the amount or level of risk above which it must escalate. Second, count the direct reports and the approvals per manager for the last full quarter. Third, compare the two lists. Where a manager escalates routine items, the span is being set by the decision rights rather than by the people. Fix the rights before you change the chart. If the changes touch employment contracts, tax, or a restructuring that needs consultation, put them to your lawyer or accountant before anything is announced. This article cannot judge those questions.

Ask the managers themselves which approvals take longest. They usually know, and they tend to name the step that nobody ever wrote down. Their answers are evidence too, and they are cheaper to collect than a benchmarking exercise. Treat them as a starting list, then check each item against the approval records before acting on it.

Then decide what evidence would change your mind. Choose one or two measures before any change, such as decision time, staff answers on whether they can get help when they need it, or the share of a manager's week spent on approvals. Agree the threshold in advance and write it into the paper that proposes the change. A design change that cannot be judged against a measure will be judged by whoever argues longest, and that is rarely the best-informed person in the room.

Limits of the data

The limit is total. The four sources the researcher checked contain no research on span of control, direct-report counts or manager effectiveness, so this piece gives no range. We have not filled that gap from memory. A range offered without a named source would be a guess, and our readers are right to discount guesses. A future edition would need a fresh search across organisational-design and labour-economics research, with each finding attributed to its publisher and year. Until that exists, treat any published span figure, including figures in consultancy material, as a claim to check against the study behind it before relying on it.

Newsletter

The Briefing

One email a week: the leaders we profiled, the ideas worth stealing, and nothing else.

Weekly, free, one click to leave. Read our privacy policy.