The honest answer is that it depends on four things, none of which is the number itself.
Someone will tell you the right span of control is seven. Someone else will say five to nine. A consultant will show you a benchmark saying eight.
All of these are averages of organisations doing different work, and applying an average to your specific team produces a reorganisation with no argument behind it.
The useful question is not "how many" but "what makes a span workable", and there are four answers.
How similar the work is. A manager of twelve people doing near-identical work — a contact centre team, a processing function — has one job repeated twelve times. A manager of five people doing five unrelated specialist jobs has five jobs. The second is harder at less than half the size.
How experienced the people are. A team of experienced practitioners who need decisions rather than direction can be twice the size of a team that is half new joiners. This is why span should move during a period of hiring and usually does not: the manager who could handle eleven last year now has four people in their first six months and the same eleven-person span.
How much of the manager's time is management. A player-manager with 60% of their week on their own delivery has 40% of a manager's capacity, and a span of eight is really a span of eight on two days. Most span problems are actually this problem.
How much coordination the work needs. Teams whose members must constantly align consume manager time in proportion to the pairs of people, not the number. Doubling a highly interdependent team more than doubles the coordination load.
Forget the benchmark. Two things in your own structure will tell you where the problems are.
Spans of one or two. These are the real finding in almost every org review. A manager with one direct report is usually one of: a title given instead of a pay rise, a layer that was never removed after a departure, or a role protected during a restructure. They are expensive — a full management salary coordinating one person — and they add a layer to every decision that passes through.
Layers from the top to the front line. Count them. Each layer adds delay, dilutes the message, and gives one more person the ability to say no. Most organisations of a few hundred people need five or six and have eight.
Layers are the number that matters, and span is mostly the lever that changes it. Widening average span from six to eight in a 500-person organisation removes roughly a layer.
An organisation-wide average span of 6.4 can hide almost anything: half the managers at eleven and half at two averages to a number that describes nobody.
Look at the distribution. What you usually find is a barbell — a few very wide spans in operational teams that are straining, and a cluster of very narrow ones in head-office functions that nobody has examined because they are small enough not to show up in a headcount review.
The narrow ones are where the money is, and the wide ones are where the attrition is.
The failure mode of every span exercise is a target: "no manager will have fewer than five reports". Within a quarter you have five-person teams assembled from unrelated work, reporting to someone who understands two thirds of it, and a set of new dotted lines to compensate.
Use the number to find the questions, then answer them one at a time:
Those are answerable. "Get the average to seven" is not, and the reorganisation it produces will be undone within eighteen months.
The [Org Design & Span of Control Analyser](/products/org-design-and-span-of-control-analyser.html) calculates spans, layers and the distribution from a headcount list, flags spans of one and two, and shows depth by function — with a worked example where the widest span is not the problem. £26.