What is the ideal span of control?

There isn't one — and any tool or consultant quoting a single number is selling you an average. Research and practice converge on ranges that depend on the work: 5–8 is a common default for mixed managerial work, wider (10–15+) suits homogeneous, process-driven teams with experienced ICs, and narrower (3–5) fits high-variability work requiring intensive coaching or judgment. The professional approach is segment-based targets: set ranges per function, location, or grade, then investigate the outliers.

What's a 1-over-1, and why does it matter?

A manager with exactly one direct report — a full management layer serving a single person. Occasionally justified (succession transitions, regulatory separation), but at any volume it's the clearest structural signature of layer accretion: decisions crossing that link gain a hop and lose nothing else. 1-over-1s and "player-coach" roles at scale are usually where a spans-and-layers review finds its first savings without touching frontline capacity.

How many layers should an organization have?

Fewer than it has, usually — but the honest answer is that layer count should follow decision speed, not benchmarks. The test isn't "do we have seven layers when peers have five"; it's "how many approvals does a routine decision cross, and what does each hop add?" Delayering that removes hops without redesigning the decisions just reassigns the same queue to fewer, busier people.

Org design

Spans & layers: the geometry of an organization

OrgTool Learn · 8 min read · Updated July 2026

Spans and layers are the two numbers that describe an organization's shape: how many people each manager manages (span of control), and how many reporting steps separate the CEO from the front line (layers). Together they determine cost, decision speed, and how diluted a message is by the time it reaches the work. They are also the most manipulated numbers in restructuring — which is why knowing how to analyze them honestly matters.

Why there's no magic number

The search for the universal ideal span is nearly a century old. V.A. Graicunas's 1933 analysis — the field's founding artifact — showed why wide spans get hard fast: the number of potential relationships a manager must attend to grows combinatorially, not linearly, with each added report. But the modern consensus runs the other way from a fixed limit: the right span depends on the nature of the work. Homogeneous, well-instrumented, process-driven work supports wide spans; high-variability work requiring coaching, judgment calls, or synthesis supports narrow ones. A support team of experienced agents can thrive at 15:1; a research group can strain at 5:1. Averages across an org are therefore close to meaningless — the signal lives in the segments.

Segment targets, not global averages

The professional method sets target ranges per segment — by function, location, grade band, or employment type — and then investigates deviations in both directions:

Below range: possible layer accretion, empire fragments, or roles kept "managerial" for compensation reasons. The extreme case is the 1-over-1.
Above range: possible manager overload — check it against attrition and the informal network before celebrating the efficiency (the widest-span managers are often the most central and most overloaded people in the company).

Two companion diagnostics complete the picture. Layer depth per person — how many hops from the top — mapped against a maximum-layers target reveals where the organization has quietly grown a basement. And grade-distance anomalies — a person graded at or above their manager — flag structures where the compensation architecture and the reporting architecture have come apart.

The structural smells

1-over-1 chains: full layers serving single individuals; each adds a hop and an approval without adding capacity.
Player-coaches at scale: managers with tiny spans who also carry full IC load — often a symptom of promoting for retention rather than for management need.
Hourglass middles: wide front lines, wide executive team, pinched middle — every decision queues through the pinch.
Span inflation at the top: executives with 12+ directs, which converts the leadership team into a status meeting and pushes real integration work into the informal network.

Why delayering-as-cost-play fails

The classic failure mode: leadership announces "we're going from seven layers to five," removes the layers, and changes nothing about the work. Six months later the spans are wider, the managers are drowning, decision queues are longer, and attrition among the newly overloaded is climbing. Structure follows work — the durable sequence is to redesign what gets decided where (which decisions move down, which approvals disappear, which coordination becomes tooling), and only then take out the geometry the redesigned work no longer needs. This is doubly true in AI-driven restructuring, where automation changes the shape of the remaining work: a heavily automated function's residual work is often more escalation-heavy, arguing for narrower spans there even as total headcount falls.

Key takeaways

There is no universal ideal span — set target ranges per segment and analyze the outliers.
Track layer depth and grade-distance anomalies alongside spans; they expose accretion spans alone hide.
Hunt the smells: 1-over-1s, player-coaches, hourglass middles, span-inflated executive teams.
Redesign the work before removing the layers; delayering without decision redesign reliably rebounds.
Wide is not automatically efficient — check wide spans against overload and network centrality before locking them in.