Dunbar's Number

Overview

Dunbar's number (~150) is the cognitive limit on stable social groups the human neocortex can sustain, with nested layers at ~5, ~15, ~50, then outer bands at ~500 and ~1,500. Derived by Robin Dunbar (1992) from primate neocortex ratios. Key implication: below ~150, informal trust coordinates behavior; above ~150, formal hierarchy becomes architecturally mandatory — not a cultural choice.

Composes with network-effects (different coordination mechanisms below vs. above 150), principal-agent (exceeding 150 multiplies agency problems informal trust can no longer cover), founder-mindset (founder relationship bandwidth peaks at ~150 — first breakpoint demanding shift from doing to delegating structure).

When to Use

• Headcount approaching or crossing 150 (or 50, or 500)
• "Coordination getting harder" complaints with no obvious structural cause
• Deciding to split a team, open a new office, or create a sub-unit
• Designing org structure for a scaling startup; evaluating acquisition integration plans
• Diagnosing why informal culture "felt different" after a headcount milestone
• AI-native startup hiring fast on abundant AI-era capital / AI-adoption demand — "we scaled from 20 to 150 in a year and coordination is breaking"

Not when: group well below 50 and problems are interpersonal; constraint is incentive misalignment (use principal-agent); challenge is growing a network beyond 150 (use network-effects).

Coaching Novices (Adaptive Front Door)

• Engine mode: user has a concrete org design or headcount question → run The Process directly.
• Coach mode: user is unfamiliar → guide step by step.

In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.

1. One-liner: once a group exceeds ~150 people, informal trust cannot coordinate it — formal structure is architecturally required by the human neocortex.
2. Check fit: what is the current headcount? Which Dunbar layer, which breakpoint is approaching?
3. Elicit the structure: what coordination mechanisms exist? Where are they breaking down?

[WAIT — do not advance until user responds]

4. Run The Process one step at a time with their input.

[WAIT — do not advance until user responds]

5. Close: group layer + breakpoint diagnosis + structural prescription (split / formalize / layer) named.

[WAIT — do not advance until user responds]

The Process

Step 1 — Identify layer: headcount → Dunbar layer (5/15/50/150/500/1,500) → distance to next breakpoint.

Step 2 — Audit coordination: how does the group coordinate? (informal trust / formal process / hierarchy) — where are breakdowns?

Step 3 — Diagnose mismatch: does coordination mechanism match group size? (below 50: direct knowledge ok; 50–150: structured trust ok; above 150: formal hierarchy required)

Step 4 — Prescribe structure: A) Split — sub-units each below 150 with explicit interfaces; B) Formalize — replace relationship-based coordination with documented processes; C) Layer — add management tier.

Step 5 — Set thresholds: current ceiling / trigger for next structural review / Gore parking-lot rule (headcount cap per unit).

Step 6 — Validate culture claims: "Lost startup feel" — Dunbar exceeded or culture drift? "Need better communication" — is structure the actual solution?

Output template

Dunbar Org Analysis: <name>
Current layer: [X] | Distance to next breakpoint: [N]
Coordination mechanism: [informal/formal] | Breakdown points: [list]
Root cause: [architectural / interpersonal / incentive]
Prescription: [split / formalize / layer] | Sub-unit targets: [≤150] | Trigger: [headcount]

→ Method in Action: Dunbar 1992 + W.L. Gore & Associates → 2026 lens: Scaling an AI-native startup team (2023–2026) — how AI capital compresses the timeline to the ~50 and ~150 breakpoints

Pack: Dunbar Number Across Domains

Domain	Layer	Key implication
Founding team (1–15)	Sympathy group	No process needed; direct knowledge
Seed-stage (15–50)	Affinity group	Informal still works; minimize overhead
Series A (50–80)	Approaching ceiling	Begin formalizing roles + reporting
Series B / Pre-150 (80–149)	Just below limit	Build management layer before hitting 150
150-person threshold	Dunbar crossed	Mandatory: documented processes + explicit hierarchy
150–500 / 500+	Acquaintance / multiple groups	VP layer + OKRs; federated units each below 150
Remote teams	~50–70 (compressed)	Video/async reduces relationship maintenance efficiency

→ Primary sources: references/sources.md

Common Rationalizations

[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.

Fake move	Reality
[D] "We just need better communication tools"	Above 150, the problem is neocortical capacity. Slack does not fix Dunbar.
[D] "We can maintain flat culture as we grow"	Flat culture works below 50. Above 150, informal coordination structurally fails.
[D] "We'll hire great people who don't need management"	Quality of people doesn't change cognitive limits.
[D] "Communication problems are a people issue"	At ~150, communication problems are architectural. HR interventions treat the symptom.
[D] "The founders just need to be more visible"	Founder visibility works below 150. Above it, all-hands don't substitute for structure.
[D] "We lost our startup feel — must have been wrong hires"	The startup feel was a Dunbar-layer property of a sub-50 group. Hires are not the issue.
[D] "More team-building will restore cohesion"	Works within the 150-layer. Across groups above 150, it produces temporary warmth, no structural repair.
→ Add [O] entries here after each real use — paste the actual failure pattern	What went wrong and why

Red Flags

• Headcount crossed 150 without any structural change or management layer addition
• Founder still coordinating a 200-person company via personal relationships
• "Communication problems" diagnosed as interpersonal without checking if Dunbar limit was crossed
• Org chart flat beyond 50 with no explicit coordination mechanism
• Acquisition integration plan assumes two informal cultures can merge without formal structure

Verification

• Headcount mapped to Dunbar layer explicitly; distance to next breakpoint calculated
• Coordination mechanisms audited against group size requirements
• Root cause identified: architectural vs. interpersonal vs. incentive
• Structural prescription chosen (split / formalize / layer); sub-unit targets set below 150
• Growth trigger for next structural review defined
• "Culture problem" vs. "architecture problem" distinction made explicitly

---

Part of deciqAI Knowledge Skills — 227 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/c/dunbars-number · ⭐ Star the repo → https://github.com/deciqAI/knowledge-skills · Contributions welcome.

Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/dunbars-number.json