In 1999, Donella Meadows — the systems dynamicist who had spent three decades modelling the behaviour of complex systems at MIT and Dartmouth — published an essay that distilled the most consequential strategic insight of the twentieth century into a single ranked list. The essay was called "Leverage Points: Places to Intervene in a System," and its argument was deceptively simple: within any complex system — an economy, a corporation, an ecosystem, a technology platform — there exist specific structural locations where a small shift in one thing can produce large changes in everything. These are leverage points. They are not random. They are not equally powerful. And most people, most of the time, intervene at the wrong ones.
Meadows identified twelve leverage points, ranked from least effective to most effective. The hierarchy is counterintuitive and that is precisely why it matters. The interventions that feel most concrete and actionable — adjusting parameters like tax rates, quotas, subsidies, and budgets — sit at the bottom of the list. They are the interventions that politicians, managers, and consultants reach for instinctively, and they are the interventions that almost never produce lasting systemic change. The interventions that feel abstract and difficult — changing the goals of the system, shifting the paradigm from which the system arises, cultivating the power to transcend paradigms entirely — sit at the top. They are the interventions that transform civilisations, industries, and organisations, and they are the interventions that almost no one attempts because they require operating at a level of abstraction that most decision-makers find uncomfortable.
The twelve points, from shallowest to deepest: (12) constants, parameters, and numbers — the knobs that everyone wants to turn; (11) the sizes of buffers and stabilising stocks; (10) the structure of material stocks and flows; (9) the lengths of delays relative to the rate of system change; (8) the strength of negative feedback loops relative to the impacts they are trying to correct; (7) the gain around positive feedback loops; (6) the structure of information flows — who has access to what information and when; (5) the rules of the system — incentives, punishments, and constraints; (4) the power to add, change, evolve, or self-organise system structure; (3) the goals of the system; (2) the mindset or paradigm out of which the system's goals, rules, and structures arise; and (1) the power to transcend paradigms.
The hierarchy maps an insight that experienced systems thinkers recognise immediately but that linear thinkers find baffling: the deeper you intervene in a system, the less effort is required and the greater the effect — but the deeper interventions are harder to see, harder to articulate, and harder to implement because they operate on the invisible architecture that generates visible behaviour. Adjusting a tax rate is visible, concrete, and politically legible. Changing the informational structure that determines how economic actors perceive their options is invisible, abstract, and requires understanding the system at a level that most participants never reach. Yet the informational intervention reshapes every subsequent decision in the system, while the tax adjustment reshapes only the narrow transaction it directly touches.
The hierarchy is not merely academic. It explains a pattern that recurs across every domain where humans attempt to change complex systems. Governments adjust tax rates (level 12) and wonder why economic behaviour barely shifts. Corporations restructure departments (level 10) and discover that the same dysfunctions reappear in the new structure. Schools revise curricula (level 12) without changing the incentive architecture that determines what teachers and students actually optimise for (level 5). Health systems add beds and staff (level 11) without redesigning the information flows that determine how patients move through the system (level 6). In every case, the intervention targets the tangible, visible layer of the system — the parameters — while the intangible, invisible layer — the goals, rules, information architecture, and paradigm — continues to generate the behaviour that the intervention was supposed to change. The system absorbs the shallow intervention and regenerates the original pattern, leaving the interveners frustrated and convinced that the problem is intractable when in fact it is merely mislocated.
This model is distinct from Leverage (Physics), which describes how amplification mechanisms convert small inputs into large outputs. Leverage (Systems) asks a different question: not "how do I amplify my effort?" but "where in this system should I intervene to produce the largest change?" The physics model is about the magnitude of force applied through a mechanism. The systems model is about the location of intervention within a structure. A founder who understands both knows not only how to multiply the impact of their actions but where to direct those actions for maximum systemic effect. The combination is what separates builders who optimise within existing structures from builders who reshape the structures themselves.
Section 2
How to See It
Systems leverage reveals itself whenever a small, structurally positioned change cascades into consequences that dwarf the scale of the intervention. The diagnostic signature is disproportionate systemic response — not because the intervention was forceful, but because it was placed at a point in the system's architecture where the structure amplifies its effects through feedback loops, information flows, and goal-redirecting mechanisms. When a seemingly minor policy change transforms organisational behaviour across thousands of people, a leverage point has been activated.
The inverse signal is equally diagnostic: when large, expensive interventions produce minimal or temporary change, you are intervening at a shallow leverage point — pushing on parameters while the deeper structures regenerate the original behaviour. The most common failure mode in corporate strategy, public policy, and personal development is spending enormous resources at leverage points 10 through 12 while leaving points 1 through 6 untouched. The diagnostic question is not "did the intervention work?" but "at what level of the hierarchy did it operate?" — because the level predicts the durability and magnitude of the effect with remarkable consistency across domains.
Business Operations
You're seeing Leverage (Systems) when a company changes a single internal metric and the entire organisation's behaviour shifts within months. When Spotify moved its primary engineering metric from "features shipped" to "impact on user engagement," it did not issue new directives to its hundreds of squads. It changed the information flow — what data engineers saw, what dashboards measured, what reviews evaluated. The metric change operated at leverage point 6 (information flows) and point 5 (rules), and the behavioural cascade was vastly disproportionate to the effort of changing a dashboard. Engineers who had been building features nobody used began killing features that distracted from engagement. The system reorganised itself around the new signal without anyone specifying how.
Technology Platforms
You're seeing Leverage (Systems) when a platform changes its algorithm and millions of creators restructure their entire output strategy overnight. When YouTube shifted its recommendation algorithm from click-through rate to watch time in 2012, it intervened at leverage point 6 — the structure of information flows between the platform and its creators. The parameter change was trivial: one ranking variable substituted for another. The systemic effect was seismic: an entire ecosystem of content creators abandoned clickbait thumbnails and short provocative clips in favour of longer, more substantive content designed to retain attention. The platform did not mandate this shift. It changed the informational signal, and millions of autonomous agents adapted their behaviour to the new structure.
Public Policy
You're seeing Leverage (Systems) when a government changes a constitutional rule and decades of entrenched behaviour dissolve. New Zealand's shift from first-past-the-post to mixed-member proportional representation in 1996 operated at leverage point 5 — the rules of the political system. The change required no restructuring of government departments, no retraining of civil servants, no new spending. It altered the electoral rule, and the entire political ecosystem reorganised: new parties formed, coalition dynamics emerged, policy positions diversified, and voter behaviour shifted. One structural rule change produced more political transformation than decades of policy-level interventions had achieved.
Organisational Culture
You're seeing Leverage (Systems) when a new leader redefines what an organisation exists to do and the culture transforms without a single memo about culture. When Satya Nadella told Microsoft that its mission was to "empower every person and every organisation on the planet to achieve more," he was operating at leverage point 3 — the goal of the system. The previous implicit goal — "protect and extend Windows" — had created defensive, siloed behaviour across the organisation. The new goal redirected every feedback loop: teams that had been punished for cannibalising Windows were now rewarded for expanding the ecosystem. Nadella changed no processes, fired no executives, and issued no operational directives. He changed the goal, and the system reconfigured around it.
Section 3
How to Use It
Decision filter
"Before intervening in any system — my company, a market, a policy, a habit — I ask: at what level of the leverage hierarchy am I operating? Am I adjusting parameters (shallow, temporary), redesigning information flows or rules (mid-range, structural), or changing the goals or paradigm of the system (deep, transformational)? If I find myself at levels 10–12, I stop and ask whether a deeper intervention would produce more durable change with less ongoing effort."
As a founder
The founder's most powerful tool is not product design, hiring, or fundraising — it is the ability to set the system's goals and rules at the deepest level and let the organisation self-organise around them. Most founders spend their time at shallow leverage points: adjusting pricing (level 12), resizing teams (level 11), restructuring reporting lines (level 10). These interventions consume enormous energy and produce changes that decay the moment attention moves elsewhere.
The leverage-aware founder operates deeper. Changing the company's core metric — from revenue to customer lifetime value, from features shipped to problems solved — is a level 3 intervention that redirects every feedback loop in the organisation without specifying how any individual should behave. Redesigning information flows — making customer complaints visible to engineers in real time rather than filtered through a support team — is a level 6 intervention that changes behaviour faster than any training programme. Establishing rules that encode the company's strategic priorities — "every new feature must reduce customer effort, not just add capability" — is a level 5 intervention that shapes thousands of decisions you will never see.
The hierarchy explains why some founders transform industries while others merely compete in them. The founders who intervene at levels 1 through 4 — changing paradigms, redefining what a market is for, establishing the power for their organisation to continuously self-organise — build systems that adapt and improve without continuous executive intervention. The founders who intervene at levels 8 through 12 build systems that require their constant presence to function, because the shallow interventions decay as soon as the force behind them is removed.
As an investor
The leverage-point hierarchy provides a diagnostic for evaluating management quality that no financial metric captures. When analysing a company, ask: at what level of the hierarchy does the leadership team habitually intervene? Companies led by executives who spend their time adjusting parameters — renegotiating supplier contracts, tweaking pricing, reallocating headcount — are operating at shallow leverage points. They may produce steady incremental improvement, but they will never generate the nonlinear value creation that transforms industries.
Companies led by executives who redesign information flows, restructure incentive architectures, and redefine organisational goals are operating at deep leverage points. These companies produce outcomes that appear discontinuous — sudden improvements in execution, rapid culture shifts, unexpected strategic pivots that succeed — because deep interventions cascade through the entire system rather than affecting only the immediate point of contact. The investor's edge is recognising which level of the hierarchy the management team operates at, because that level determines the shape of the company's future value curve: linear improvement from shallow interventions, or nonlinear transformation from deep ones.
As a decision-maker
Every decision-maker faces a constant temptation to intervene at the shallowest available leverage point because shallow interventions are visible, immediate, and politically legible. Announcing a budget increase, hiring a new team, or adjusting a quarterly target produces the appearance of action. Redesigning an information flow, changing a rule, or challenging the paradigm that generated the problem requires patience, abstraction, and the willingness to be misunderstood in the short term.
Meadows's hierarchy provides the discipline: before any intervention, locate it on the twelve-point scale. If you are operating at levels 10–12, ask whether a deeper intervention exists that would address the root structure rather than the surface symptom. If a deeper intervention exists but feels too abstract or too difficult to communicate, that discomfort is a signal — you have found the actual leverage point, and the resistance you feel is the system protecting its current structure from the change it most needs.
Common misapplication: Assuming that deeper always means better in every situation. Meadows herself warned that the hierarchy describes the potential impact of interventions at each level — not the feasibility. Changing a paradigm is the highest-leverage intervention, but it is also the hardest to execute and the slowest to produce visible results. Sometimes the correct intervention is a parameter adjustment — a price change, a deadline, a hiring decision — because the situation demands speed, not depth. The discipline is matching the depth of the intervention to the time horizon and scope of the problem, not reflexively reaching for the deepest possible lever.
A second misapplication is intervening at a deep leverage point in the wrong direction. Meadows noted that people who discover leverage points frequently push them the wrong way. A founder who changes the company's goal to "maximise growth at all costs" has intervened at level 3 — a deep leverage point — but in a direction that produces systemic destruction rather than systemic health. Depth amplifies the intervention; it does not guarantee the intervention is wise. The amplification works in both directions: a well-aimed deep intervention produces transformational improvement that propagates through every feedback loop in the system, while a poorly aimed deep intervention produces transformational damage that propagates through those same loops with equal efficiency.
A third misapplication is treating the hierarchy as permission to ignore shallow leverage points entirely. Parameter adjustments — pricing, headcount, budgets — are not useless. They are necessary operational tools. The error is treating them as strategic tools, deploying them as solutions to structural problems. A price adjustment handles a competitive response. A goal change handles a strategic misdirection. The discipline is matching the intervention depth to the problem depth, not reaching for the deepest available lever regardless of the situation.
Section 4
The Mechanism
Section 5
Founders & Leaders in Action
The leaders who produce the most durable systemic transformations share a structural habit: they intervene at the deepest leverage points available to them rather than spending their energy adjusting parameters at the surface. They change goals, redesign information flows, rewrite rules, and challenge paradigms — not because these interventions are easy, but because they are the interventions that cascade through every feedback loop in the system, reshaping behaviour without requiring continuous executive force.
The common pattern is patience combined with structural precision. These leaders endure short-term ambiguity — the period between a deep intervention and its visible systemic effect — because they understand that the delay is not a sign of failure but a structural property of deep leverage points propagating through complex feedback architectures.
What distinguishes them from leaders who merely announce new visions or reorganise boxes on an org chart is follow-through at the structural level. A goal change that is not encoded into metrics, incentive structures, and information flows decays into rhetoric. A paradigm shift that is not reinforced by consistent resource allocation and leadership behaviour reverts to the old paradigm within months. The leaders below succeeded not only because they intervened at the right depth but because they ensured that every lower-level structure in the system was realigned to sustain the deep intervention. The hierarchy works in both directions: a deep intervention cascades downward through every level, but only if the lower levels are allowed to reorganise around the new structure rather than being held in their previous configuration by institutional inertia.
Bezos consistently intervened at Meadows's deepest leverage points. His most consequential act as CEO was not a product decision or an acquisition — it was establishing "customer obsession" as the goal of the entire system (leverage point 3). Every feedback loop at Amazon was calibrated to this goal: metrics tracked customer satisfaction, promotions rewarded customer-centric thinking, resource allocation favoured projects that reduced customer friction. The goal change was a single structural intervention that propagated through hundreds of teams and thousands of decisions without Bezos specifying any of them.
At leverage point 6, Bezos redesigned Amazon's information flows with the "empty chair" practice — placing an empty chair in meetings to represent the customer, forcing every discussion to confront the question of customer impact. More structurally, he mandated that teams write six-page narratives instead of PowerPoint presentations, changing the informational medium through which decisions were made. The shift from slides to prose altered the depth and rigour of strategic thinking across the entire organisation — a single rule change at leverage point 5 that reshaped the cognitive architecture of Amazon's decision-making system. Bezos understood intuitively what Meadows articulated theoretically: the place you intervene matters more than the force you apply.
Nadella's transformation of Microsoft is the clearest modern case of a leader operating at leverage points 2 and 3 simultaneously. The paradigm from which Microsoft operated under Ballmer — "we are a Windows company" — sat at leverage point 2, the mindset from which all goals, rules, and structures arose. Every strategic decision was filtered through the question "does this protect Windows?", and the feedback loops of the organisation enforced this paradigm ruthlessly.
Nadella's intervention was to change the paradigm itself: Microsoft was no longer a Windows company but a cloud and platform company. This single shift at the deepest structural level cascaded through every layer of the hierarchy. The goal changed from protecting Windows revenue to growing cloud adoption. The rules changed: teams were no longer penalised for cannibalising Windows. The information flows changed: success metrics shifted from Windows licence sales to Azure consumption and Microsoft 365 subscriptions. Nadella did not need to reorganise departments, issue operational directives, or micromanage the transition. He changed the paradigm, and the system reconfigured around it — from a $300 billion market capitalisation in 2014 to over $3 trillion by 2024. The magnitude of the outcome was disproportionate to the apparent simplicity of the intervention, which is the defining signature of a deep leverage point correctly activated.
Grove's decision to exit the memory chip business and focus Intel on microprocessors was an intervention at leverage point 3 — changing the goal of the system. Intel's identity, culture, and feedback loops were all organised around memory manufacturing. The goal — "be the best memory company" — determined what engineers worked on, what the sales force sold, and what metrics the board evaluated. The system was coherent and optimised, and it was leading Intel toward extinction as Japanese manufacturers achieved cost advantages that no amount of parameter-level intervention could overcome.
Grove reframed the question at the paradigm level: "If we got kicked out and the board brought in a new CEO, what would he do?" The answer was obvious from outside the existing paradigm but invisible from within it. The new CEO would exit memory and focus on processors. Grove's genius was recognising that the intervention had to occur at the level of the system's goal and identity — not at the level of operational efficiency or cost reduction. He changed what Intel existed to do, and the entire organisational system — R&D allocation, talent development, customer relationships, manufacturing strategy — reorganised around the new goal. The parameter-level interventions that would have been required to compete in memory (cost reductions, factory upgrades, pricing adjustments) were rendered irrelevant by a single structural intervention at the goal level.
Lee governed Singapore by intervening at the deepest available leverage points in the national system. His most powerful intervention was at leverage point 5 — the rules of the system — where he established incorruptible rule of law and eliminated corruption from the civil service. In most developing nations, corruption operates as a reinforcing feedback loop: officials extract rents, which attracts rent-seekers to government, which increases corruption, which degrades public services, which reduces economic activity, which increases the incentive for officials to extract rents. Breaking this loop required not a parameter adjustment but a rule change so severe and so consistently enforced that the feedback structure itself was redesigned.
Simultaneously, Lee intervened at leverage point 6 — information flows — by making English the medium of instruction, connecting Singapore's workforce directly to global information, capital, and trade networks. And at leverage point 3, he set the national goal as economic development through meritocratic competition rather than ethnic patronage or ideological orthodoxy. Each intervention was structurally deep, and the combination produced cascading effects that no amount of surface-level policy could have generated. Singapore's transformation from a resource-poor city-state to one of the world's wealthiest nations within a single generation is the result of three deep leverage-point interventions compounding through the feedback architecture of an entire society.
Dalio's "radical transparency" at Bridgewater is a deliberate intervention at leverage point 6 — the structure of information flows. In most organisations, information about performance, disagreements, and mistakes flows through informal channels with massive delay and distortion. Junior employees cannot see senior leaders' reasoning. Feedback about poor decisions is filtered, softened, or suppressed entirely. The resulting information structure produces a system that is blind to its own errors and slow to self-correct.
Dalio redesigned the information architecture: all meetings were recorded, all feedback was documented, all performance assessments were visible to all participants. The intervention was uncomfortable — it violated every social norm about information privacy in organisations — but it operated at a deep leverage point. By changing who had access to what information and when, Dalio changed the behaviour of every agent in the system without issuing behavioural directives. Employees who knew their reasoning would be visible reasoned more carefully. Leaders who knew their mistakes would be documented corrected faster. The system's error-correction rate increased by orders of magnitude, producing Bridgewater's sustained outperformance across multiple market cycles — all traceable to a single structural intervention in the information architecture.
Section 6
Visual Explanation
Meadows's hierarchy reveals that the places where we habitually intervene — parameters, buffers, and physical structures — are the shallowest leverage points, producing small and temporary effects. The deepest leverage points — goals, paradigms, and the power to transcend paradigms — are abstract, invisible, and transformational.
The visual arrangement below maps the twelve points from shallow (bottom) to deep (top), showing why most interventions fail to produce lasting change and where the structurally aware leader directs their effort. The increasing visual weight toward the top of the hierarchy reflects the increasing systemic impact: each level upward produces effects that are not merely stronger but qualitatively different, because deeper interventions reshape the feedback structures that generate all surface-level behaviour rather than merely adjusting the behaviour itself.
Section 7
Connected Models
Leverage (Systems) occupies a unique position in the mental model lattice: it is a meta-strategic framework that tells you where to apply every other model. Systems Thinking provides the diagnostic vocabulary for mapping feedback structures; Leverage (Systems) ranks those structures by their capacity for intervention. Incentives explains how rules shape behaviour; Leverage (Systems) explains why rules are a mid-range intervention point and what sits above them.
The six connections below map how this model reinforces frameworks that share its structural logic (by identifying the system dynamics that make those frameworks powerful), creates tension with frameworks that assume simpler intervention strategies (by revealing that parsimony and incrementalism can be traps in complex systems), and leads naturally to the systemic phenomena that emerge when deep leverage points are activated (by explaining the structural mechanisms through which paradigm shifts and metric dysfunction propagate through feedback architectures).
Reinforces
Systems Thinking
Leverage (Systems) is the operational extension of Systems Thinking. Systems Thinking provides the diagnostic framework — stocks, flows, feedback loops, delays — for understanding why a system behaves as it does. Leverage (Systems) answers the question that Systems Thinking raises but does not resolve: given that I understand the system's structure, where exactly should I intervene? Meadows's hierarchy ranks the structural elements that Systems Thinking identifies, converting a descriptive framework into a prescriptive one. The reinforcement is reciprocal: Systems Thinking without leverage-point awareness produces diagnosis without strategy; leverage-point analysis without Systems Thinking produces intervention without understanding. The combination — mapping the feedback structure and then identifying the deepest accessible intervention point — is the most complete strategic methodology available for operating in complex systems.
Reinforces
Incentives
Incentives sit at leverage point 5 in Meadows's hierarchy — the rules of the system. The Incentives model explains that agents within a system respond rationally to the reward and punishment structures they face, often producing behaviour that the system's designers did not intend. Leverage (Systems) reinforces this insight by placing incentive design in its proper structural context: incentives are powerful because they are a mid-range leverage point that shapes every downstream behaviour, but they are not the deepest intervention available. A system whose goals are misaligned will produce dysfunctional behaviour regardless of how cleverly its incentives are designed, because the incentives are calibrated to serve the wrong goal. The reinforcement is that understanding incentives leads you to ask "what rules are shaping behaviour?" and understanding leverage points leads you to ask "what goals are the rules serving?" — each framework deepening the other.
Tension
Section 8
One Key Quote
"Folks who do systems analysis have a great belief in 'leverage points.' These are places within a complex system where a small shift in one thing can produce big changes in everything. The leverage point frequently is not intuitive. Or if it is, we too often push it in the wrong direction."
— Donella Meadows, 'Leverage Points: Places to Intervene in a System' (1999)
The statement captures the double edge of systems leverage: the most powerful intervention points exist, but they are counterintuitive and dangerous. Most people either cannot find them — intervening at parameter levels that produce no lasting change — or find them and push in the wrong direction, amplifying the very dysfunction they intended to correct. The strategic discipline is finding the deep leverage point and having the structural understanding to push it in the right direction — a combination that requires mapping the system before moving the lever.
Meadows spent thirty years watching well-intentioned policymakers, executives, and reformers discover leverage points and then push them in precisely the wrong direction. Growth-oriented governments that identified the financial system as a leverage point deregulated it into crisis. Efficiency-focused executives who identified organisational structure as a leverage point centralised it into rigidity. The leverage point amplifies whatever passes through it — wisdom and folly alike. The depth of the intervention determines the magnitude of the effect. It does not determine the direction. That requires understanding the system well enough to know which way "better" actually points — and that understanding is rarer than the willingness to act.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
Leverage (Systems) is the model that explains why most organisational change efforts fail and why the rare ones that succeed seem almost effortless in retrospect. The pattern is consistent: when leaders intervene at shallow leverage points — adjusting budgets, reshuffling teams, launching initiatives — the system absorbs the intervention and returns to its previous behaviour within months. When leaders intervene at deep leverage points — changing the goal, redesigning information flows, challenging the paradigm — the system reorganises itself around the new structure without ongoing executive force. The effort of the deep intervention is modest. The effect is transformational. The gap between effort and effect is Meadows's hierarchy made operational.
The most important insight for founders is that most of what passes for strategic leadership is parameter adjustment. Raising prices, cutting costs, hiring, firing, reorganising departments, setting quarterly targets — these are all interventions at levels 10 through 12, the shallowest leverage points. They consume enormous executive energy and produce effects that decay the moment attention shifts. The founder who changes the company's core metric from revenue to customer lifetime value (level 3), who makes customer complaints visible to every engineer in real time (level 6), who establishes a rule that every product decision must reduce customer effort (level 5) — this founder has made three interventions that will reshape organisational behaviour for years without any additional effort. The interventions are structurally self-sustaining because they operate at levels where the system's feedback loops propagate and reinforce the change.
The hierarchy also explains why some leaders seem to have a "reality distortion field."Steve Jobs, Elon Musk, Jeff Bezos — the leaders described as having an almost gravitational effect on the people and organisations around them — were operating at leverage points 2 and 3. They changed what people believed was possible (paradigm) and what the organisation existed to do (goal). These are the deepest structural interventions available to a leader, and their effects propagate through every feedback loop in the system without the leader needing to specify or direct the propagation. The "reality distortion field" is not charisma. It is deep structural leverage producing effects that appear disproportionate to the observable effort because the real work — the paradigm shift — is invisible.
For investors, the leverage-point hierarchy is a diagnostic tool for management quality. When evaluating a leadership team, ask: at what level of the hierarchy do they habitually intervene? A CEO who spends their time adjusting budgets and reorganising teams is operating at levels 10 through 12. A CEO who is redesigning information flows, changing incentive architectures, and redefining the company's purpose is operating at levels 3 through 6. The second CEO will produce nonlinear outcomes because deep leverage points cascade through the entire system. The first CEO will produce linear outcomes at best because shallow leverage points affect only the immediate point of contact.
Section 10
Test Yourself
Systems leverage is invoked loosely whenever someone suggests a "strategic" intervention, but Meadows's framework demands precision: where on the twelve-point hierarchy does the intervention actually operate, and is there a deeper point that would produce a more durable effect? These scenarios test your ability to locate interventions on the hierarchy, distinguish shallow from deep leverage, and recognise when an intervention is aimed at the right level versus when it is consuming effort at a parameter level while leaving the generating structure intact.
The key diagnostic: does the intervention change the system's behaviour temporarily (shallow leverage) or does it change the structure that generates the behaviour (deep leverage)? Temporary change requires continuous force to maintain. Structural change is self-sustaining because the feedback loops propagate and reinforce it.
A secondary diagnostic is the reversion test: if you remove the intervention, does the system's behaviour snap back to its previous pattern? If it does, you operated at a parameter level — you were pushing against the system's structure rather than changing it. If the new behaviour persists without ongoing force, you changed the structure itself, and the system's own feedback dynamics now maintain the new pattern. Every scenario below can be evaluated with this test.
Where is the leverage point?
Scenario 1
A hospital system struggling with patient readmission rates creates a new dashboard that shows each physician their individual readmission statistics compared to department averages, updated weekly. Within six months, readmission rates drop 22% — without any changes to clinical protocols, staffing, or incentives.
Scenario 2
A SaaS company experiencing declining growth increases its marketing budget by 40% and hires twenty additional sales representatives. Growth accelerates for two quarters, then returns to the pre-intervention trend line. The CEO proposes another budget increase.
Scenario 3
A country struggling with deforestation passes a law granting indigenous communities legal title to their ancestral forestlands. Within a decade, deforestation rates in titled areas drop 75% compared to untitled areas with equivalent ecological characteristics.
Section 11
Top Resources
The leverage-points framework emerged from decades of system dynamics research, and the strongest resources are those that connect Meadows's hierarchy to the modelling tradition that produced it. Start with Meadows's original essay for the framework itself — it is short, free, and the single highest-leverage reading in the systems dynamics canon. Then read her primer for the systems vocabulary that makes the framework operational. Senge provides the organisational translation, converting abstract leverage-point theory into management practice. Forrester provides the mathematical foundation that demonstrates why structural interventions outperform parameter adjustments. Sterman provides the full quantitative toolkit for modelling the systems in which leverage points operate.
The intellectual progression matters: the essay provides the hierarchy, the primer provides the vocabulary, Senge provides the organisational application, Forrester provides the mathematical proof, and Sterman provides the simulation methodology. Each resource builds on the previous one.
The reader who engages with all five will possess the most complete available toolkit for identifying where to intervene in any complex system — and the discipline to resist the perennial temptation of the shallow, visible, politically comfortable intervention that changes nothing.
The foundational essay. Meadows presents the twelve leverage points in ascending order of effectiveness, with examples from ecology, economics, and policy. The essay is concise — roughly 8,000 words — and immediately actionable. It is the single most important strategic document produced by the systems dynamics tradition, and every founder, investor, and decision-maker should read it. The key insight is on the first page: the leverage points are counterintuitive, and when we find them, we usually push them in the wrong direction.
The companion text that provides the systems vocabulary — stocks, flows, feedback loops, delays — needed to identify and operate at leverage points. Without understanding the feedback structures that Meadows's hierarchy ranks, the hierarchy itself is a list rather than a tool. This primer converts it into a tool by teaching you to map the structures in which leverage points are embedded. Chapter 6, on leverage points, expands the original essay with additional examples and practical guidance.
Senge translates systems leverage into organisational management practice. His system archetypes — "shifting the burden," "fixes that fail," "limits to growth" — are recurring patterns that reveal where leverage points exist in organisational systems and why the default interventions (shallow) fail while the structural interventions (deep) succeed. The book provides the pattern-recognition templates that enable leaders to identify leverage points in their own organisations without building formal system dynamics models.
The foundational text of system dynamics, which demonstrated through computer simulation that organisational behaviour is generated by feedback structure, not by individual decisions. Forrester's models of supply chains, production systems, and corporate dynamics revealed that the same structures produce the same behaviours regardless of who occupies the roles — the empirical foundation for Meadows's insight that structural interventions outperform parameter adjustments. Technically demanding but transformative for anyone who wants to understand why leverage points work.
The most comprehensive modern treatment of system dynamics applied to business strategy. Sterman, Forrester's intellectual successor at MIT, covers the full toolkit for modelling the systems in which leverage points operate: causal loop diagrams, stock-and-flow models, simulation, and policy design. The book includes detailed case studies that demonstrate how identifying the correct leverage point — rather than the obvious one — produces dramatically better strategic outcomes. Essential for anyone who wants to move from conceptual leverage-point thinking to quantitative systems intervention.
Companies that illustrate this model
Strategy playbooks where this pattern shows up in practice.
Leverage Points — Twelve places to intervene in a system, ranked from shallowest (parameters) to deepest (transcending paradigms). Depth of intervention determines magnitude and durability of systemic change.
[Gall's Law](/mental-models/galls-law)
Gall's Law states that a complex system that works has invariably evolved from a simple system that worked, and that a complex system designed from scratch never works. The implication is that intervention should be incremental — start simple, iterate, evolve. Leverage (Systems) creates tension by arguing that incremental, parameter-level interventions (the simplest possible changes) are often the least effective, and that the most powerful interventions target the deep structural elements — goals, paradigms, self-organisation — that Gall's Law would counsel approaching with extreme caution. The tension is productive: Gall's Law warns against overambitious structural redesign that ignores emergent complexity, while Leverage (Systems) warns against timid parameter-tweaking that ignores structural causes. The resolution lies in intervening deeply but narrowly — changing a single goal or information flow rather than redesigning the entire system simultaneously.
Tension
[Occam's Razor](/mental-models/occams-razor)
Occam's Razor counsels preferring the simplest explanation and, by extension, the simplest intervention. Leverage (Systems) reveals that the simplest interventions — parameter adjustments at levels 10 through 12 — are often the least effective because they fail to address the feedback structures that generate the behaviour. The model argues that the most effective interventions require understanding the system at a level of structural complexity that Occam's Razor would discourage. The tension is between analytical parsimony and structural accuracy: a simple explanation of organisational dysfunction (the team lacks talent) leads to a simple intervention (hire better people), while the systems explanation (the information architecture prevents good people from making good decisions) leads to a structurally deeper and more effective intervention. The resolution is to apply Occam's Razor to the intervention, not the analysis — find the single deepest leverage point that addresses the structural cause, rather than deploying multiple shallow interventions at the symptom level.
Leads-to
[Paradigm Shift](/mental-models/paradigm-shift)
Leverage point 2 — the mindset or paradigm out of which the system's goals, rules, and structures arise — connects directly to Thomas Kuhn's concept of the paradigm shift. Meadows placed paradigms near the top of her hierarchy because changing a paradigm redirects every goal, every rule, and every feedback loop in the system simultaneously. A paradigm shift in an industry — from "software is a product you buy" to "software is a service you subscribe to" — restructures the entire competitive landscape without any individual actor designing the restructuring. Leverage (Systems) leads to Paradigm Shift by identifying the paradigm as a leverage point and explaining why paradigm changes produce effects that are disproportionate to the apparent effort: the paradigm is the generative structure from which all lower-level structures derive, and changing it propagates through every level of the hierarchy simultaneously.
Leads-to
[Goodhart's Law](/mental-models/goodharts-law)
Goodhart's Law — "when a measure becomes a target, it ceases to be a good measure" — is a predictable consequence of intervening at the wrong leverage point. When an organisation makes a metric the target, it is intervening at leverage point 12 (parameters) rather than at leverage point 3 (goals). The metric is a proxy for the goal, not the goal itself. Agents in the system, responding rationally to the rules (leverage point 5), optimise for the proxy and diverge from the actual goal. Leverage (Systems) leads to Goodhart's Law by explaining the structural mechanism: the further the intervention is from the system's actual goal, the more likely that agents will find ways to satisfy the intervention without satisfying the goal. The remedy is to intervene at the goal level — ensuring that the system's purpose is clearly defined and that metrics serve as feedback signals for the goal rather than substitutes for it.
The most underappreciated leverage point in the hierarchy is number 6: the structure of information flows. Most organisations spend enormous resources on strategy, culture, and incentive design while paying almost no attention to who has access to what information and when. Yet information architecture shapes every decision made within the system. An engineer who cannot see customer complaints makes different product decisions than one who sees them in real time. A sales team that cannot see product usage data sells differently than one that can. A board that receives financial metrics quarterly makes different governance decisions than one that receives operational dashboards weekly. Changing information flows is a mid-range leverage intervention that is dramatically more powerful than the parameter adjustments most organisations default to, and dramatically more feasible than the goal or paradigm changes that require the rarest form of leadership vision.
The practical framework: before every intervention, locate it on the twelve-point scale. If you are operating below level 6, ask whether a deeper intervention is available. If it is, the deeper intervention will almost always produce more durable change with less ongoing effort — not because deeper is magically better, but because deeper interventions reshape the feedback structures that generate all surface-level behaviour. The founder who adjusts a price is pushing a parameter. The founder who changes the metric the organisation optimises for is changing a goal. The first change lasts until the next pricing review. The second change lasts until someone changes the goal again. The durability of the effect is proportional to the depth of the leverage point — and that proportionality is the most actionable insight in the entire systems dynamics literature.
The AI era is exposing leverage-point blindness at an unprecedented scale. Organisations deploying AI as a parameter-level intervention — automating individual tasks, reducing headcount for specific functions, generating content faster — are operating at levels 10 through 12. They will realise marginal efficiency gains. Organisations deploying AI as a structural intervention — redesigning information flows so that every employee has real-time access to synthesised intelligence (level 6), rewriting organisational rules to enable human-AI collaborative decision-making (level 5), or redefining the company's goal to include capabilities that only human-AI systems can achieve (level 3) — will produce nonlinear competitive advantages that the parameter-level adopters cannot match. The technology is identical. The leverage point at which it is applied determines whether it produces incremental improvement or structural transformation. Meadows's hierarchy, written before the internet reached mass adoption, is more operationally relevant in the age of AI than at any previous point in its history.
Scenario 4
A tech company's engineering team is consistently missing deadlines. The VP of Engineering responds by adding weekly status meetings, requiring more detailed project plans, and instituting a mandatory code review process. Delivery times improve slightly for one quarter, then degrade further as engineers spend increasing time in meetings and on documentation rather than coding.
