“Fewer managers” is structurally impossible — at least under the assumptions most organizations still operate on.
Consulting frameworks treat de-layering as a cost optimization problem. Benchmark ratios define what a “healthy” span looks like. Companies cut layers and report short-term savings. The logic seems airtight. Yet the arithmetic reasserts itself — through overload, coordination drift, and decision whiplash — and the manager count always creeps back.
That’s because manager count is an output determined by how information flows and decisions actually get made. The arithmetic doesn’t negotiate.
You have to redesign the structure the arithmetic operates on — cutting won’t get you there.
Like a shadow cast by an object, the number of managers is a consequence of something upstream. Organizations keep trying to reshape the shadow directly — trimming, compressing, flattening — while the object that casts it remains unchanged. The shadow always returns to its original shape because the object hasn’t moved.
Manager Count Is a Derived Property
Span-of-control ratios follow the Graicunas formula — a non-linear arithmetic that most optimization approaches overlook. Five direct reports create roughly one hundred possible relational connections; ten reports produce over one thousand. This is a deterministic constraint built into how coordination works, not a guideline open to interpretation.
The layer math is equally unforgiving. A one-thousand-person organization at an average span of six needs four layers below the chief executive. At span four, it needs five. That single extra layer accounts for fifty to one hundred additional managers [1].
Manager count is an output of structure, not an input to optimize.
The progression makes the arithmetic visible. Industry observation shows spans stretching from 1:5 to 1:15, and removing two layers from a 1:5 structure can theoretically produce 1:125 ratios [3] [6]. Each step represents a fundamental structural shift — from hierarchical relay nodes to distributed information flows to self-organizing micro-teams — not merely a headcount adjustment. Organizations that treat span as a number to benchmark miss the deeper constraint: the architecture determines how many managers the math requires.
What Actually Determines the Number
Every span-of-control benchmark carries a hidden assumption: that managers are the mandatory processors of information in the organization. Standard benchmarks place front-line operations at ten-to-twenty direct reports, professional teams at six-to-ten, and senior leadership at six-to-ten [7].
These ranges presuppose a specific architecture — one where signals travel upward through layers until they reach someone authorized to decide.
Each layer in this architecture adds latency. Information must be compressed, summarized, and translated as it moves upward. Decisions must travel back down through the same layers, losing nuance at each step.
The manager exists in this architecture as a relay node — receiving information from below, processing it, and passing it upward or laterally. The number of managers is determined by how many relay nodes the topology requires.
This assumption is structural, not inevitable. When information flows through shared dashboards, real-time protocols, or distributed decision mechanisms instead of hierarchical relays, the relay node becomes optional.
The benchmark ranges describe what spans look like under one specific design pattern — centralized information processing through positional authority. Change the architecture, and the shadow changes shape.
Why Flattening Always Regenerates
Removing layers without redesigning information flows produces wider spans that break under their own arithmetic. Increase spans without removing the low-value coordination work that filled the deleted layers, and managers become overwhelmed [2]. The overload manifests predictably: mentoring debt, coordination drift, decision whiplash, psychological cost, and burnout [3].
Flattening without redesign is like reshaping a shadow with your hands. It snaps back the moment you look away because the object hasn’t moved.
Organizations read early throughput gains as proof of success while relationship health and product quality slowly decline [3]. The shadow appears to shrink — until the distortion becomes visible elsewhere. 84 percent of companies have not redesigned jobs for artificial intelligence, installing new tools on top of architectures designed for a different era [4].
When the assumptions baked into the operating structure are invalid, no amount of commitment or resolve will make things better. The manager count creeps back because the structure still requires relay nodes. The shadow returns to its original shape.
Redesign the Architecture, the Number Resolves Itself
Haier replaced the entire coordination architecture — internal markets, real-time performance dashboards, entrepreneurial incentive mechanisms across four thousand micro-enterprises [2].
The structure changed fundamentally. Information no longer needed to travel through hierarchical relays.
This is the structural precondition that most de-layering efforts skip. Distributing authority requires first distributing clarity about what to decide. More than 25 percent of executives cite lack of strategic clarity as the top limiter of organizational effectiveness [5]. Authority patterns do not shift through directives — they shift through strategic preparation and boundary clarity that makes distributed decision-making possible.
Redesign what casts the shadow, and it shrinks on its own.
Intentional redesign of workflows must precede structural change. The sequence matters:
- First, build information flow architectures that bypass relay nodes.
- Second, distribute clarity about who decides what.
- Third, widen spans because the arithmetic now supports it.
This piece extends the thesis of Less Managers Is More Management: Why Agentic Systems Redistribute Management to Everyone — which diagnoses why authority must move to the edge. This piece provides the arithmetic precondition for why that redistribution is structurally necessary.
Sources
- Pawan Joshi: Span-of-Control Math — Graicunas formula, layer math, 1:5 to 1:125 progression
- AliveOrganisations: Haier Case Study — Internal markets, real-time dashboards, 4,000 micro-enterprises
- SAM National: Flattening Corporate Layers — September 2025 — Mentoring debt, coordination drift, decision whiplash
- Deloitte: State of AI in the Enterprise 2026 — 2026 — 84% of companies have not redesigned jobs for AI
- LHH: 2026 C-Suite Research — 2026 — 25% of executives cite lack of strategic clarity
- Andy Budd: LinkedIn Post — 2026 — 1:125 ratio observation
- Umbrex: Span of Control Layering Analysis — Front-line 10-20, professional 6-10, senior 6-10 benchmarks
Please note: 51even is an AI-first organization. We embrace AI at every step of our value creation and build our processes with a deep integration of human-AI capability. Humans always have the last decision. But this text was heavily built with AI.