Span of control is the number of people who report directly to one manager. Two things govern whether a given span is healthy: the nature of the work being supervised, and the coordination value the manager genuinely adds. For fairly routine, well-documented work (running a standardized store playbook, staffing a shift, executing a defined manufacturing process) a span of roughly six to ten direct reports is a reasonable rule of thumb: wide enough to be efficient, narrow enough that the manager still coaches and catches problems. For complex, judgment-heavy, or highly variable work, a narrower span, often four to seven, is defensible, since the coordination burden per person is simply higher. Neither number is a law; both are a starting hypothesis you test against the actual work, and Worked Example 2 walks a full recomputation.
What are layers, and what is a ghost layer?
Layers are the management levels between the top of the organization and the front line. More layers mean more hand-offs, more places for a decision to slow down or distort in translation, and more fixed management cost, but layers exist for a real reason when they carry genuine coordination and judgment a wider span could not absorb. The diagnostic test that separates a real layer from a ghost layer is that same span number: if a layer's average span is unusually thin, a manager overseeing only one or two reports, most of the time, that manager is very often relaying messages up and down a chain that would function identically without them, at the cost of one more hop every decision must clear. A layer whose average span sits meaningfully below roughly four is a strong candidate for removal, and Worked Example 2 finds and removes exactly such a layer, with the headcount cost and new span computed explicitly.
Advertisement
Should decisions be centralised or decentralised?
Centralize or decentralize is the other structural question, answered decision by decision, never once for the whole organization: should a category of decision be made centrally, or locally, close to the best information about the specific situation? The test: who has the better information for this specific call, and who bears the consequence if it is made badly. Decisions where local knowledge dominates and the downside of inconsistency is small (a store's local price response, a branch's staffing schedule) usually belong closer to the front line. Decisions where consistency, scale economics, or legal/brand risk dominate (national supplier contracts, brand standards, regulatory reporting) usually belong centrally. Getting this wrong has a real cost either way: over-centralizing produces slow, locally-wrong answers (exactly Nandavan Retail's opening problem); over-decentralizing produces brand fragmentation and inconsistent customer experience. Practice P4 drills this test on a fresh scenario.