Span of control describes how many subordinates a manager supervises, and every layer removed widens it for someone.
Why it matters when the plan changes
Layers and spans are among the few organisational properties that can be measured directly, which makes them the target of most structural cost programmes. Each layer removed shortens the path information travels from the front line to the top, and information weakens at every reporting step, so delayering has a real effect on how much reality reaches the top. Sull, Homkes and Sull, writing in Harvard Business Review in 2015, found that 84% of managers say they can rely on their boss and their direct reports all or most of the time, showing how much reliance already erodes before any layer is added.
The tension is that the right span depends entirely on the work. Spans of fifteen are fine where work is standardised and self-directing, and damaging where it needs judgement and coaching. Applying one target span across an organisation optimises the number and degrades the places where management mattered, since dependencies and real ownership, not reporting lines alone, decide whether the work gets done well.
In practice
A cost programme sets a minimum span of eight across a group. In shared services it works. In a research function where each project needs substantive review, managers now have eleven reports and review nothing properly. The layer count improved and the quality of a specific kind of decision fell, which no structural metric reported.
Evidence
Span of control is the number of subordinates a supervisor has, and it is the direct determinant of layer count.
Span of control, Wikipedia (2026)Information about execution weakens as it passes through each reporting step, which is what additional layers add.
Donald Sull, Rebecca Homkes and Charles Sull, Why Strategy Execution Unravels and What to Do About It, Harvard Business Review (2015)
What it cannot tell you
Layers and spans describe structure, not the work moving through it. The same span number can be sound in standardised operations and damaging where judgement and coaching are required. The count says nothing about dependencies between roles, decision quality, or where ownership is duplicated or assumed, all of which determine whether a given structure functions.
Questions
No. It depends on how standardised the work is, how much judgement each case needs, how experienced the team is and how much coaching the role involves. A single target span applied across an organisation will be right somewhere and wrong in most places.
Because layers and spans are measurable, the savings are calculable, and information does travel better through fewer steps. Sull, Homkes and Sull, writing in Harvard Business Review in 2015, found that 84% of managers say they can rely on their boss and direct reports all or most of the time. The attention cost of removing layers goes uncounted.
Attention per person: coaching, review of individual work, and the informal noticing that catches problems early. Where the work needs those, widening the span converts a manager into a coordinator, and the function loses a quality-control step nobody had named.
Each reporting step softens information slightly, and the softening compounds across layers. Span of control, as described in the Wikipedia entry from 2026, is the direct determinant of layer count in any organisation. Fewer layers means less distortion between the work and the top, a real benefit of delayering separate from any cost argument.
By what the work requires, differing across the organisation, with the differences stated and defended. A design that varies spans deliberately is harder to explain in a cost programme and produces a structure that matches the work rather than the spreadsheet.