Why good management policies make things worse
Many management failures happen with plans that were executed well. What failed was the reading of how the system would respond to the plan.
John Sterman, who teaches system dynamics at MIT, gives this a name in Business Dynamics: policy resistance, the tendency for interventions to be delayed, diluted or defeated by the system’s own response. His point is uncomfortable. Competent, well-intentioned decisions routinely make worse the very problem they target.
His simplest example is a sales gap. Sales came in at $80 million against a $100 million goal. You cut prices to win share. Sales rise, and the problem looks solved. Then competitors cut their prices, and sales fall again. Yesterday’s solution has become today’s problem.
Every executive has seen some version of this. Headcount added to a late project that slows it further. A cost cut that raises costs two quarters later. A productivity target that teaches people to game the metric. Each one looked reasonable on the day it was approved.
These failures follow a few recurring structures: feedback loops nobody drew, delays nobody priced in, and accumulations nobody measured. This essay describes those structures, two patterns in which they show up inside management teams, and a short list of questions to ask before approving a decision.
Why good policies backfire
Most management runs on an event-oriented view of the world. We see a gap between goal and reality, pick an option, and expect a result. Sterman argues that reality works through feedback instead. Our results change the situation, other people react to restore the balance we disturbed, and the new situation shapes our next decision. As he puts it, we are not outside the system pulling strings. We are inside it.
From there he makes a claim worth putting on the wall of any leadership team: “there are no side effects, there are just effects.” What we call side effects are the effects we did not anticipate. Their existence says less about reality than about how narrow our understanding of it was.
Four mechanisms explain most of the damage.
Feedback loops we did not draw. Every dynamic comes from two kinds of loops: reinforcing loops that amplify what is happening, and balancing loops that resist change. A price cut that triggers a price war is a reinforcing loop. A competitor restoring parity is a balancing one. A decision memo that shows only the arrow from our action to our goal is missing the loops that will decide the outcome.
The decision memo usually shows only loop B1. Loop B2 is slower, so the plan looks like it worked before the competitors’ answer arrives.
Delays. Cause and effect are often distant in time. When a corrective loop has a delay, decision makers keep correcting after enough has already been done, and the system overshoots. This is the hiring binge followed by the layoff, and the inventory cycle in every supply chain.
Accumulations. Many things that matter are stocks: capability, backlog, technical debt, trust, customer base. Stocks change only through their flows, and slowly. Building them takes much longer than draining them. Trust lost in a quarter can take years to rebuild.
The trade-off between short and long run. This is the most practical idea in Sterman’s work. Policies with real leverage often get worse before they get better. Policies with little leverage often bring a short improvement before the problem returns, larger. The dashboard will therefore often reward the wrong decision and punish the right one, at least for a while.
Why experience does not fix it
Seniority does not protect anyone from these errors. Sterman calls the pattern the misperception of feedback: our mental models think in chains of events, ignore delays and confuse stocks with flows. In the Beer Distribution Game, a four-stage supply chain simulation whose rules take fifteen minutes to explain, participants ranging from high school students to CEOs produce average costs more than ten times the optimum. The errors persist with experience and with money at stake.
Two more traps sit inside the leadership team itself:
- We blame people instead of structure. We tend to explain behavior by character rather than by the system people work in. Yet different people placed in the same structure tend to behave in similar ways. Replacing the VP of Sales changes little if the incentives stay the same.
- Our metrics filter what we can see. Measurement is an act of selection. Sterman’s example is NASA software that discarded very low ozone readings as instrument error, which delayed detection of the ozone hole by as much as seven years. Most management dashboards make a similar choice quietly: they show flows and outcomes, rarely the stocks that will drive next year.
Two patterns worth learning to recognize
Intuition alone cannot handle these dynamics. Sterman’s MIT colleague Peter Senge offered a practical shortcut in The Fifth Discipline (1990): a small set of recurring structures he called system archetypes. They do not replace analysis, but they are the fastest way for a leadership team to notice which story it is in. Two of them explain a large share of the failures above.
Get Daniel Braz’s stories in your inbox
Join Medium for free to get updates from this writer.
Fixes that fail. A fix relieves the symptom quickly. The same fix also triggers a consequence that comes back after a delay and makes the problem worse. Cutting maintenance to hit a cost target is the classic case. Costs fall this quarter. Breakdowns rise later, and costs rise with them. Because the delay hides the link, the usual response is to apply the same fix again.
The capability trap. Nelson Repenning and John Sterman described this pattern in a 2001 California Management Review article whose title says most of it: “Nobody Ever Gets Credit for Fixing Problems that Never Happened.” When performance falls short, managers can push people to work harder, which raises output now, or invest time in improvement, which raises capability later. Working harder wins in the short run because it is fast and visible. But people find the extra hours by skipping improvement and maintenance, and capability erodes after a delay. Performance drops again. The pressure rises. The organization settles into permanent firefighting.
Two details make the trap hard to escape. Working harder produces “better-before-worse” behavior, the mirror image of real improvement. And managers misread the cause. Because the delays are long, they blame the people, apply more pressure, see a short-term gain and take it as proof. The authors conclude that managers rarely escape the trap because they rarely realize they are in it.
Before modeling: can the answer be known in advance?
Loops and archetypes assume the structure is stable enough to reason about. Some situations do not allow that. Dave Snowden’s Cynefin framework, presented to executives in a 2007 Harvard Business Review article with Mary Boone, separates situations where cause and effect can be worked out in advance, with analysis or expertise, from complex situations where we understand why things happened only in retrospect. In the complex case, the advice is to run small experiments that are safe to fail and watch what emerges, instead of committing to one plan.
What leaders can do differently
None of this requires a modeling department. It requires a few different questions in the meetings where decisions are made.
- Ask which pattern you might be in. Before approving a fix, ask what it relieves now and what it could cost in six or twelve months. If the answer to the second question is “nothing,” look again. That is how fixes that fail get approved.
- Ask who will react, and how. List the actors whose behavior the decision changes: competitors, customers, your own teams, regulators. Then ask what each will do to restore the balance you are about to upset. This surfaces most of the loops behind policy resistance.
- Write down the expected delay before you judge. If a policy has real leverage, it may get worse before it gets better. Agree in advance how long you will wait and which leading indicator will show it is working. Otherwise the first bad month kills the right decision.
- Put stocks on the dashboard. Revenue, tickets closed and features shipped are flows. Capability, backlog, technical debt, customer trust and team morale are stocks. Stocks predict next year. Flows describe last month.
- Protect improvement time. Treat time for maintenance, learning and process improvement as a stock to protect, not a reserve to raid when targets slip. The capability trap starts the first time it is raided.
- Blame the structure before the person. When performance is poor, ask whether someone else in the same role, with the same incentives and information, would act differently. Usually they would not. Change the structure first.
- Test before you scale, and review the mental model afterwards. Where the structure is understood, test the policy in a simple model or simulation first. Where cause and effect cannot be known in advance, run small safe-to-fail experiments instead of a program. Either way, review what you believed about the system, not just the result. Sterman, following Chris Argyris, calls this double-loop learning.
Behind all seven sits a cultural condition. When the pressure to perform overrides the need to learn, and when defensive routines make real assumptions undiscussable, nothing on this list survives. Teams that cannot say “our model of the market was wrong” will repeat the same policy and call the outcome bad luck.
Closing
Leaders are paid to act, so the practical goal is to act knowing that the system will answer, and to plan for that answer.
Most of the answer is predictable in shape, even if not in detail. Policy resistance, fixes that fail and the capability trap share one structure: a quick, visible gain, a slow, hidden cost, and a delay between them. The habit that helps most is to ask, before approving any decision, what it will set in motion and how the team will find out in time.
Sources
- Sterman, J. D. (2000). Business Dynamics: Systems Thinking and Modeling for a Complex World. Irwin/McGraw-Hill. Chapter 1, “Learning in and about Complex Systems.”
- Senge, P. M. (1990). The Fifth Discipline: The Art and Practice of the Learning Organization. Doubleday. Summary of the archetype: Fixes that fail.
- Repenning, N. P., & Sterman, J. D. (2001). “Nobody Ever Gets Credit for Fixing Problems that Never Happened: Creating and Sustaining Process Improvement.” California Management Review, 43(4), 64–88. Abstract
- Snowden, D. J., & Boone, M. E. (2007). “A Leader’s Framework for Decision Making.” Harvard Business Review, November 2007. Copy of the article
