How to run small continuous-improvement trials
A small improvement trial means changing one thing for a short time, measuring it against a starting figure and then deciding whether to keep it. This guide shows how a supervisor and a team can do that without a big project. It matters because most good ideas on a shop floor are never tested, and many tested ones never reach the written standard.
Why this matters
A packaging unit in Pune had a recurring problem: about 4 in every 100 cartons left the printing line with a smudged edge. Operators had several theories, such as ink thickness, roller speed and dust. Each shift tried something different, nobody wrote it down, and the defect came back after every change of supervisor. One trial with one change and a written count would have settled it in a week.
Before you start
- An idea log, which can be a notebook or a sheet where anyone can write a problem or suggestion with the date and their name.
- A way to count the problem today, such as defects per hundred, minutes per job or complaints per week.
- The current standard operating procedure for the process, printed where the work happens.
- One supervisor who agrees to review ideas each week and one process owner for each trial.
- A trial sheet with columns for date, shift, setting used, quantity and defects.
Step by step
- A team member, from any role, writes the problem or idea in the idea log with the date. For example, "smudge on carton edge, Line 2, worse after lunch shift." The supervisor reads the log every week and picks the one or two items worth testing.
- The supervisor records the baseline: the present measure over a few days, such as 4 smudged cartons per 100. The supervisor also names the process owner and writes what will count as better, for example "below 2 per 100 over five shifts."
- The process owner changes only one thing, for example slower roller speed on Line 2, and runs the trial for a set period on a small quantity. The trial sheet records every shift, the setting used and the counts. Line 1 runs as usual so there is something to compare with.
- At the end of the period, the team member who raised the idea sits with the process owner and compares the trial figures with the baseline. They also note side effects, such as slower output or more waste at start-up, before anyone decides to adopt the change.
- If the result is clearly better and the side effects are acceptable, the supervisor updates the written standard. The revised procedure gets a new version date, and the old copy is removed from the shop floor so only one version is in use.
- The supervisor briefs each shift on the change and checks the next two weeks of counts to see whether the improvement holds. If the defect returns, the supervisor reopens the trial and finds what was different.
- The supervisor records the review result in the idea log: adopted, rejected or needs another trial, along with the figures. Rejected ideas stay on record so nobody repeats the same test.
Who does what
| Role | What they do | Record they own |
|---|---|---|
| Team member | Raises problems and ideas, joins the evidence review | Idea log |
| Supervisor | Picks trials, sets the baseline, updates the standard and checks it holds | Baseline and review result |
| Process owner | Runs the trial and records each run | Trial sheet |
Common mistakes
- Changing two or three settings at once, so nobody knows which one worked.
- Running the trial for a single shift and declaring success on one good day.
- Skipping the baseline, which makes any result an opinion.
- Adopting the change on the floor but not updating the written procedure, so the next new joiner is taught the old way.
- Ignoring ideas that do not get picked, so staff stop writing them down.
How to tell it is working
- Trial completion. Count the trials that reached a recorded decision (adopted, rejected or retest) and divide by the trials started in the quarter. If this is low, trials are being started and left, which wastes the team's effort.
- Recurring defect. Pick the defect you tested and count how often it appears per hundred units in the four weeks after the standard changed. Compare with the baseline. It should stay below the baseline and not creep back.
- Adoption of standards. Walk the floor and observe a few jobs. Count how many followed the revised procedure and divide by the jobs you watched. A change that is on paper but not in use does not count.
Questions people ask
Who owns idea to improvement?
The supervisor owns the whole cycle, because the supervisor chooses what to test and updates the standard. The process owner runs each trial. Any team member can raise an idea, and their name stays against it.
What is the minimum record for idea to improvement?
You need the idea log entry, the baseline figure, the trial sheet, the revised procedure and the review result. If a manager asks why a step changed, these five let you show the problem, the test and the proof.
When should a case stop?
Stop a trial when the planned period ends, when it harms safety or quality, or when the result is clear. Stop a case entirely once the standard is updated and has held for the follow-up weeks, or when the idea is rejected with figures recorded.
Can a spreadsheet support this stream?
Yes. One sheet for the idea log and one for each trial is enough. The risk is untidy entries, so fix the column names, require a date and a name, and print the revised procedure rather than leaving it only on a screen.
What should the owner review?
Once a month the owner should look at how many ideas came in, how many trials finished, which standards changed and whether the defect figures moved. The owner should also ask about the ideas that were rejected, to be sure good ones are not being dropped.