The Finding
Nelson Repenning and John Sterman of MIT Sloan School of Management, examining why process-improvement effort so often fails to produce lasting gains, found that organisations under pressure consistently default to “working harder” — applying more effort at the point of visible strain — rather than “working smarter,” addressing the structural condition generating that strain. Their broader finding, the capability trap, describes how this pattern becomes self-reinforcing: effort applied at the visible point of pressure relieves that specific pressure, is rewarded because it is visible, and is therefore repeated the next time pressure appears — without ever addressing why the pressure appeared there in the first place.
What It Means in Practice
A familiar and often frustrating experience for operational leaders follows directly from this. A bottleneck is identified. Resources are added at that point. The bottleneck clears — and within a period of time, a bottleneck reappears somewhere else in the same process, often in a location that was previously running smoothly.
This is not evidence that the fix at the original bottleneck failed. In most cases, it worked exactly as intended. What it reveals is that the original bottleneck was not the sole constraint on the system’s overall capacity — it was the tightest constraint at that particular moment. Once it is relieved, the system’s overall throughput rises until it meets the next-tightest constraint, which then becomes visible as the new bottleneck. The bottleneck has not failed to be fixed. It has moved, because the underlying capacity of the system as a whole has not changed — only the location of its current limiting point has.
The capability trap dynamic explains why this cycle so often repeats indefinitely rather than converging on a stable, well-functioning system. Fixing the visible bottleneck produces an immediate, measurable, creditable win. Examining the system’s overall capacity — asking what would need to be true for no bottleneck to reappear anywhere — is slower, less visible, and produces no equivalent immediate credit. Under sustained operational pressure, the visible, creditable option keeps winning, which is precisely why organisations can spend years chasing a bottleneck around a process without the underlying pattern of behaviour ever changing.
Where It Varies
By how tightly resourced the whole system is. A system with capacity distributed close to evenly across every stage will show this pattern most dramatically — relieving one constraint reveals the next almost immediately, because there is little slack anywhere else to absorb the change.
By visibility of the constraint. Bottlenecks in highly visible, easily measured parts of a process attract fixing effort quickly. Bottlenecks in less visible parts of the same process can persist far longer, simply because nobody is watching that part closely enough to notice.
By how success is currently measured. Where success is measured by “did the last known bottleneck get resolved,” rather than by overall system throughput, the moving-bottleneck pattern is far more likely to go unrecognised as a pattern at all — each instance looks like a fresh, unrelated problem.
By organisational memory. Organisations that track where bottlenecks have previously appeared are better positioned to notice the pattern of movement itself; those that treat each instance independently tend to keep rediscovering the same underlying dynamic without ever naming it.
There is a compounding consequence worth naming directly. Each cycle of fixing the visible bottleneck and watching it reappear elsewhere consumes real resources — time, attention, sometimes capital investment — that could, in principle, have been spent examining the system’s overall capacity once, rather than its current tightest point repeatedly. Because each individual fix is locally successful, the cumulative cost of chasing the bottleneck around the system rarely gets tallied as a single figure; it is distributed across many separate, apparently unrelated projects, each justified on its own terms, none of them evaluated against what a single structural investment might have achieved instead.
A related misconception deserves attention: a bottleneck that has moved is often read as evidence that the previous fix “wasn’t the real problem” — implying the earlier diagnosis was wrong. This usually is not the correct reading. The earlier diagnosis was very often right, for the constraint that existed at the time. What changed is that relieving it revealed a different constraint, which is a normal and expected consequence of successfully fixing a bottleneck in an interdependent system, not evidence of an earlier misdiagnosis.
This connects to a further point about how success itself is judged in this territory. A team that repeatedly and visibly resolves the latest bottleneck can, over time, build a strong internal reputation for effectiveness — precisely because each individual intervention genuinely works. The reputation is not unearned. What it obscures is the question of whether the same effort, directed instead at the system’s overall capacity, would have prevented the need for the next several interventions altogether. Judging operational effectiveness purely by the visible rate of successful fixes, without asking whether the underlying rate of new bottlenecks appearing is itself falling, risks rewarding exactly the pattern this research describes.
Looking for your own sector? The Other Sectors page points to the closest match.
Questions worth considering
- Has a bottleneck in this organisation been “fixed” more than once, in different locations, within the same overall process?
- Is current success measured by whether the last known constraint was relieved, or by whether overall throughput has genuinely improved?
- What would it mean, in this specific process, to ask about the system’s overall capacity rather than the location of its current tightest point?
More on this topic: Operations.
Your team is facing similar challenges?
Subscribe to The Riddle Ledger
Perspectives are sent as they are published, with no promotional email, and you can unsubscribe at any time.