Once again, the figures from the annual improvement round are positive. For the eighth year in a row, production reports productivity gains; the track record shows hundreds of implemented suggestions; the programme is considered a model within the group. Only one figure appears on no slide: profit in the first year was six percent; last year it was 0.8. With the same effort. While the organisation completes its eighth round, a competitor has not improved the same process but replaced it with a digital method with a different cost structure. No ninth round will help against that move.
Continuous improvement is not a substitute for renewal. The more mature a process is, the more expensive each additional percentage point becomes—and the closer the moment when only a fresh start will work.
At a grid operator I supported on this issue, a sober analysis of the long-running maintenance programme showed: the last three improvement rounds together had cost more than they delivered. No one had noticed, because the organisation reported its cumulative successes, never the return from the most recent round. The total looked impressive. The trend told a different story.
This pattern has a familiar shape—and those who know it recognise the right time to change course. Three levers turn it into leadership practice.
Richard Foster described the relationship between improvement effort and performance gain as an S-curve. At the start of a new method, each unit invested yields little; the organisation is still learning. Then comes the steep phase, where effort and impact are in an excellent ratio. Finally, the curve flattens: the method is exhausted, and each additional percentage point costs more than the previous one. Foster’s key point is aimed at defenders of the status quo: switching to a new curve always seems unattractive at first by comparison, because the new curve starts below the level the old one has long since reached. That is precisely why established organisations rationally keep investing in the old until an attacker proves that the new curve is steeper.
| Characteristic | Further optimisation | Fresh start |
|---|---|---|
| Return per unit of effort | declines with each round | rises after the ramp-up phase |
| Risk | low and familiar | high and unfamiliar |
| Impact on the status quo | reinforces it | calls it into question |
| Typical mistake | continued for too long | started too late |
One misunderstanding needs to be cleared up: this is not an objection to operational excellence. Day-to-day operations that are not under control first need stability and discipline—and there, continuous improvement is the method of choice. The trap begins only after mastery, where an organisation continues to refine a mature process because the routine of improvement feels more familiar than the question of whether the method itself is still the right one. Those who align themselves with the industry’s best practices also move only within the existing curve—at best, along its upper edge.
Most improvement programmes report their successes cumulatively: amounts saved since the programme started, total measures implemented, productivity in multi-year comparison. This presentation flatters the programme and obscures the decisive trend. Whether further rounds are worthwhile is revealed only by the marginal return: what did the most recent round cost, and what did it deliver? Economics describes this pattern as the law of diminishing marginal returns: the more mature a method is, the more input the same increase requires.
Therefore, for each improvement round, set effort and impact side by side—including internal capacity, which few people ever calculate. If the ratio declines over several rounds, that is not a reason for greater effort, but the signal that the curve is flattening. Beyond the numbers, there are early warning signs every experienced manager recognises: suggestions become more granular, discussions more often revolve around measurement than the measure itself, and the programme increasingly needs self-promotion to generate participation. Missing this signal is costly. Ignoring it because the programme is politically successful is even more costly.
As soon as marginal returns fall, a second option belongs on the table—one that is structurally absent from improvement routines: not refining the method further, but rebuilding it from scratch. This assessment requires a different question than the usual one. Not: How do we make the process better? But: How would we solve the task today if the process did not exist? Only this question opens the view to methods with a fundamentally different cost structure—from automating entire workflows to eliminating the process altogether. Those who only refine will, at best, find the best version of what already exists—never the fundamentally different solution. You are polishing the perfect combustion engine while the market switches to electric drive.
A fresh start is not an end in itself and is not automatically the better choice. It ties up attention, creates resistance, and takes time to reach the level of the established method. Whether it is worthwhile is decided by comparing both curves over an honest time horizon—not the next quarter. Often, such a fresh start leads to a far-reaching structural change that can be carried out thoroughly and at the same time swiftly, at the right pace.
When improvement rounds and fresh starts compete for the same resources, improvement wins—because its numbers are familiar and its risks are known. An organisation that runs both in one programme will postpone the fresh start year after year, each time for good reasons. The result is a perfectly optimised past.
Therefore, establish a deliberate ratio: a defined share of improvement capacity works on the current curve, a protected share on the next. The CEO of an energy service provider, whom I supported as a sparring partner, set aside one fifth of his optimisation budget for experiments with fundamentally new methods—explicitly with no return requirement in the first year. Two of these experiments failed. The third replaced a process that had previously been optimised for seven years and undercut its costs by a third. No further improvement round would have achieved this result, and the resources freed up were available for investments in new growth.
First: In your largest improvement programme, how has the return from the last three rounds developed, measured against the effort in each case? If you do not know this figure, then probably no one in your organisation does.
Second: Which process in your area of responsibility would you no longer set up today the way it currently exists? And for how many years has that very process been improved instead of replaced?
Third: Before the next budget round, have both alternatives calculated for your most mature improvement programme—another optimisation round and a fresh start—each over three years. Decide only afterwards.
The most dangerous feature of continuous improvement is that it can always be justified. There is always another percentage point, another suggestion, another round. It is precisely this endlessness that obscures the moment when the method is exhausted and only switching to a new curve will move things forward.
Optimisation answers the question of how well you have mastered what already exists. It never answers the question of whether what exists is still the right foundation. You must ask that question yourself—before a competitor answers it for you.
Efficiency is not effectiveness – Before the question of how well something is done comes the question of whether it is the right thing.
Quick wins vs sustainable transformation – Rapid results and long-term renewal require different governance.
All Insights can be found in the overview.
Proven tools and models for self-application are available under Solutions.
If you want to take these thoughts further for your company, a no-obligation initial conversation is worthwhile.
