An idea that is not tried is an opinion. The step from a short list to a measured result is where most idea work stops, and it is also the only step that produces a number anybody outside the room will believe.
What a trial card holds
- The change, in one sentence that a person who was not in the session can carry out.
- The measure, the figure of the question, counted the same way the baseline was counted.
- The period, long enough for the figure to move and short enough that somebody still cares.
- The person, one name, not a department.
- The decision after: keep and standardize, change and repeat, or drop and say why.
The five trials
| Idea | From which tool | Cases per year | EUR per year |
|---|---|---|---|
| Corner protectors cut from foam offcuts, placed at the paint line exit | Change questions | 48 | 14,400 |
| One rack height instead of three, loading pattern printed on the rack | Reverse question | 36 | 10,800 |
| Marked transport lane, no cross traffic, speed limit for the tow tractor | Reverse question | 24 | 7,200 |
| Handover card at the gate, signed by both areas | Other field | 20 | 6,000 |
| One rack type used only for housings, never under other parts | Morphological box | 18 | 5,400 |
Together that is 146 of 264 cases, 55 percent, EUR 43,800 a year against a baseline of EUR 79,200. The one-time cost is EUR 6,000: a cutting jig, the rebuild of eight racks, floor marking, cards and labels, and the training that goes with them. The handover check costs 3 minutes per handover, 1,200 handovers a year at EUR 60 an hour, so EUR 3,600 a year of running cost. Net benefit EUR 40,200 a year, payback 1.8 months.
Count what the measurement shows
Six months later the five trials were closed. Three were kept and written into the standard, two were changed and repeated. The measured result was 131 cases avoided per year against an estimate of 146 and a target of 132.
That is close enough to keep going and not close enough to call it done, and reporting it that way is the point. A method credited with its estimates survives one management review. A method that reports 131 when it promised 146 survives the next five.
Write it into the standard
A trial that is kept has two weeks to reach the document that governs the work: the loading instruction, the rack drawing, the transport rack list. Until then the improvement lives in the memory of the people who ran the trial, and the next shift has not heard of it. A status list in one file does the rest: open, in trial, in the standard, dropped, each with a date.
Make it a routine
One good session is luck. Three routines make it a capability.
| Routine | How often | What |
|---|---|---|
| Question board | Weekly | Open questions with owner, baseline and planned session date |
| Session and trial review | Monthly per area | One question, and the measured effect of the running trials |
| Method review | Twice a year | The 25 statements of the check, the key figures, the next facilitators |
Four numbers tell you whether the routine is healthy: ideas per session, selection rate, days from session to first trial, and confirmed effect per question. A selection rate above 40 percent usually means the screen is too soft. Below 5 percent it means the limits were never said out loud.
Train facilitators, not groups
Two sessions as a participant, one as a co-facilitator, one alone with an observer. One page of rules and one agenda, because a facilitator who improvises the agenda loses the clock. Rotate the role, so that the method never becomes one person’s property, and the 118 cases still left in the example become somebody else’s question.