Leverage Points: Where to Push to Change Everything
In 1999, systems thinker Donella Meadows published an essay that became one of the most influential pieces of writing in the field of complex systems. It was called Leverage Points: Places to Intervene in a System.
Her central argument: not all interventions are equal. Some places in a system are so sensitive that a small push there produces enormous change. Others are nearly inert — you can push with all your strength and nothing moves.
The tragedy she identified: organizations almost universally focus their energy on the least powerful leverage points.
The Leverage Hierarchy
The Hierarchy of Leverage
| Rank | Leverage Point | Example |
|---|---|---|
| 12 (Least Powerful) | Numbers (constants, parameters) | "Change the tax rate by 2%"; "Hire 10 more engineers"; "Cut the budget by 15%" → Tweaks the system. Rarely changes behavior. |
| 11 | Buffer sizes relative to flows | "Build 6 months of cash reserves"; "Add more server capacity" → Stabilizing, but expensive and slow to change. |
| 10 | Stock-and-flow structure | "Redesign the supply chain"; "Rebuild the architecture" → Powerful but often physically constrained. |
| 9 | Delays in feedback loops | "Make deployment faster"; "Shorten the feedback cycle" → Significant. Often underestimated. |
| 8 | Strength of negative feedback loops | "Tighten the quality control process"; "Make prices more responsive to supply" → Directly affects system stability. |
| 7 | Gain of driving positive loops | "Invest in the network effect"; "Double down on what's compounding" → Can produce exponential change. |
| 6 | Structure of information flows | "Who has access to what data?"; "What gets measured and reported?" → Massive. Information gaps are system gaps. |
| 5 | Rules of the system | "Change the incentive structure"; "Change the legal constraints" → The constitution of the system. |
| 4 | Power to change the rules | "Who gets to set the rules?"; "Who controls governance?" → Meta-leverage: control the rule-makers. |
| 3 | Goals of the system | "What is the system actually optimizing for?"; "Change the target, not just the behavior." → Transforms behavior without changing structure. |
| 2 | Mindset / paradigm | "The shared beliefs that produce the rules"; "What the system believes about itself" → The source of goals, rules, structures. |
| 1 (Most Powerful) | Power to transcend paradigms | "Hold no paradigm as absolute truth"; "Stay flexible at the level of world-view" → Unlimited. Rarely achieved. |
Most reorganizations, budget cycles, and quarterly plans operate exclusively at levels 12 and 11. They adjust numbers. They are shocked when nothing changes.
Where Organizations Actually Spend Their Energy
Let's be honest about what most change initiatives look like:
| Where Change Energy Goes | Where High-Leverage Change Lives |
|---|---|
| 83% Numbers & parameters: "Cut headcount by 10%", "Increase sales targets by 20%", "Reduce meeting time by 30%" | 5% Numbers (necessary maintenance) |
| 12% Structural tweaks: "Reorganize teams", "Move to a new office", "Switch project management tools" | 10% Structure (occasionally transformative) |
| 4% Information flows: "Add a new dashboard", "More reporting requirements" | 20% Information & feedback loops (high ROI) |
| 1% Rules & incentives: "Change how bonuses are calculated" | 30% Rules & incentives (very high ROI) |
| 0% Goals, paradigms, purpose (almost never touched) | 35% Goals, paradigms, mindset (transformational) |
High-Leverage Points
The Information Flow Leverage Point
Leverage point #6 — the structure of information flows — is consistently underused and undervalued.
Systems behave badly not because people are bad, but because they lack information. The person making the decision doesn't experience the consequence of the decision.
Classic examples of information flow failures:
- The engineer who writes code never sees the support tickets it generates
- The executive who sets growth targets never talks to the customers being over-sold
- The architect who designs the system never does on-call for it
- The product manager who prioritizes features never watches a user struggle with them
In each case, the fix isn't training or process. It's closing the information loop — ensuring the people making decisions receive the consequences of their decisions.
| Broken Loop | Fixed Loop |
|---|---|
| Engineer writes code | Engineer writes code |
| ↓ (6 months later, different team) | ↓ (immediately, automatically) |
| Support ticket arrives | Error rates, support ticket rate attributed to code |
| ↓ (buried in a report) | ↓ (engineer sees in dashboard) |
| "Tech debt is increasing" | Direct feedback on quality of decision |
| ↓ (nobody connects to original decision) | ↓ (informs next decision) |
| Engineer writes more code with same pattern | Engineer writes better code |
Same engineer. Same intentions. Radically different outcome because the information loop is closed.
The Paradigm Leverage Point
The most powerful accessible leverage point — and the most uncomfortable — is changing the shared beliefs that produce the rules, goals, and structures of a system.
Paradigms are the source code of civilization. They are the unquestioned assumptions that everyone in the system treats as obviously true:
- "Growth is always good"
- "Competition produces better outcomes than cooperation"
- "Technical debt is a business decision, not an engineering one"
- "Users don't read documentation"
- "This is just how things work here"
Changing a paradigm doesn't require power or resources. It requires pointing, clearly and persistently, at evidence that contradicts the assumption. It requires modeling the alternative. It requires patience.
| Phase | Description |
|---|---|
| Phase 1: Anomalies accumulate | Evidence that the current paradigm doesn't explain reality starts piling up. Ignored, dismissed, rationalized. "Edge cases." |
| Phase 2: Crisis | The anomalies become too large to ignore. The old explanations stop working. Anxiety in the system increases. |
| Phase 3: Competing alternatives | New frameworks emerge to explain the anomalies. Advocates compete for adoption. The system is in flux. |
| Phase 4: Tipping point | One framework gains enough adoption to become the new default assumption. The paradigm has shifted. |
| Phase 5: New normal | The new paradigm is now "obvious." People forget there was ever a debate. New anomalies begin to accumulate. |
Most people wait for Phase 2 or 3 to act. The leverage lives in Phase 1: naming the anomalies before they become crises.
Applying Leverage
The Meta-Skill: Seeing the Level You're Operating At
The highest-leverage skill isn't knowing where the leverage points are. It's knowing which level you're actually operating at when you think you're doing something powerful.
It's easy to believe you're changing the culture (paradigm-level) when you're actually just changing the ping-pong table (numbers-level). It's easy to believe you're changing incentives (rules-level) when you're actually just renaming the OKRs (information-level).
Ask ruthlessly: what level of the leverage hierarchy does this intervention actually touch?
If the answer is "numbers," don't be surprised when the system shrugs and continues as before.
The systems that survive and thrive are the ones governed by people who understand what they're actually changing — and have the discipline to push at the levels that matter.
This concludes the five-part Systems Thinking series. The concepts here — stocks and flows, feedback loops, emergence, and leverage points — are a lens, not a formula. Apply them to what's in front of you, and the world becomes legible in ways it wasn't before.