Why Process Improvements Fade After the Pilot — and How to Make Them Stick

The pattern is familiar to anyone who has led an improvement project. The pilot goes well: cycle times fall, errors drop, the team is enthusiastic and the results are presented with confidence. Six months later, someone pulls the numbers again and finds that performance has drifted most of the way back to where it started. Nobody decided to abandon the improvement. It simply wore off.

This is not a rare failure. Researchers studying Lean and Six Sigma programmes note that early gains often “simply disappear over time”.¹ A survey of experienced improvement practitioners found the most common causes of failed projects were resistance to change, a lack of commitment from senior management and teams without the skills to carry the work.² None of these is a technical problem. They are problems of ownership — and they appear after the pilot, not during it.

Why pilots succeed

A pilot is an unusual environment. It has a project lead whose job is to make it work, a small and willing group of people, extra attention from management and a temporary tolerance for things going slowly while the new way settles in. Every one of those conditions helps the improvement — and every one of them disappears at handover.

There is also a measurement effect. During a pilot, performance is watched closely, so people behave as if it is being watched. When the measurement stops, so does that effect. Part of what the pilot measured was the attention, not just the new process.

Why improvements fade

The project ends, and nobody owns the result. The improvement belonged to the project. When the project closes, its lead moves on and the process returns to a manager who was never asked to own the new standard.

The new way is not written down. If the improved method lives in the heads of the pilot team, it erodes as people leave, new staff are trained by colleagues who learned the old way, and small shortcuts accumulate.

Measurement stops at the finish line. The before-and-after comparison that proved the pilot worked is usually a one-off exercise. Without an ongoing measure, drift is invisible until it is large.

The old way is still available. If the spreadsheet, the manual workaround or the old form still exists, people under pressure will return to it. Every alternative path left open is a route back.

Victory is declared too soon. John Kotter identified declaring victory too soon, and failing to anchor changes in everyday culture, among the most common reasons transformations fail.³ A celebrated pilot can signal that the work is finished when it has only started.

What makes improvements stick

Sustaining an improvement is a design task in its own right. It should be planned during the pilot, not after the results are presented.

Name a process owner before the pilot ends. One person — usually the manager who will run the process day to day — accepts responsibility for the new standard and its performance. Involve them in the pilot so they inherit something they understand, not something handed to them.

Write the standard down, briefly. A one-page standard: the steps, who does each one, the inputs required and what “done correctly” looks like. It becomes the basis for training new people and for checking whether the process is still being followed.

Keep one or two measures running. Choose the measure that proved the improvement — cycle time, error rate, first-time-right — and keep reporting it at a regular interval, with a threshold that triggers a review. A simple run chart is often enough to show whether performance is holding or drifting.

Close the old paths. Retire the old template, remove the duplicate spreadsheet, change the system default. Where the old way cannot be removed, make the new way the easier one.

Review on a fixed rhythm, then respond. A short monthly check by the process owner: is the standard being followed, is the measure inside its threshold, and if not, why? Research on sustained Lean and Six Sigma improvements found that the ones which lasted were repeatedly cycled through experience, reflection and adjustment over an extended period, and documented as they went.¹ Sustaining is not a single control step; it is a loop.

A sustainment checklist for the end of any pilot

Before a pilot is declared a success, check that:

• A named process owner has accepted the new standard and its measure.
• The new method is written down on a page and used for training.
• At least one measure keeps running after the project closes, with a threshold and a review date.
• The old way has been removed or made harder than the new way.
• A first review is in the calendar three months after handover — and the result of that review, not the pilot, is reported as the outcome.

Closing thought

A pilot proves that an improvement is possible. Only the months after it prove that the improvement is real. Organisations that treat sustainment as part of the design — with an owner, a standard, a measure and a review rhythm — keep more of what they gain, and spend less time solving the same problem twice.

Footnotes

1. S. S. Chakravorty and D. N. Hales, “Sustainability of process improvements: an application of the experiential learning model (ELM)”, International Journal of Production Research, 55(17), 2017. Abstract. ↩ ↩²

2. J. Antony et al., “A study into the reasons for process improvement project failures: results from a pilot survey”, International Journal of Quality & Reliability Management, 36(10), 2019. Abstract. ↩

3. J. P. Kotter, “Leading Change: Why Transformation Efforts Fail”, Harvard Business Review, 1995. ↩