This is the part people get wrong, and getting it wrong quietly ruins everything that follows. A target condition is not a goal for the process. It is a description of how the process will be working at a date a few weeks away, with one figure attached to it.
A description, not a number
| A target condition is | A result target is |
|---|---|
| A description of how the process works at a date | A number the process is supposed to deliver |
| Nine of ten cycles within 75 seconds at station 4 | Output up ten percent |
| Every part within reach, nobody leaves the station | Less waste at the line |
| Four to six weeks away | By the end of the year |
| Verified by standing at the process for an hour | Verified in the monthly report |
The difference is not academic. From a result target no experiment can be built, because nothing in it says what should be different in a cycle tomorrow morning. From a description you can read the next obstacle directly.
Three ways it goes wrong
It is a result target in disguise. Output up ten percent, scrap down by half, overtime gone. All worth wanting, none of them a condition.
It is too far away. Three months out, the next step is never obvious, and the daily conversation turns into a status report.
It is already solved. If the learner can list every step that leads there, this is a plan with a deadline. A target condition has an unknown in it, which is exactly why it has to be approached by experiment.
Five tests before you write the date
Close enough: four to six weeks, no more. Measurable: one figure, readable at the process in minutes, not a figure that arrives in a report. A process condition: it says how the work runs, not only what comes out. Out of reach today: the path is genuinely unknown. Owned: the learner wrote it and can say it from memory.
The first target condition on L3
The current condition said station 4 was the widest step of the line, 52 to 141 seconds, with 38 percent of cycles inside the planned 75 and 14 of the 23 daily line stops starting there. The pair wrote this, on 9 February, for 6 March:
By 6 March, station 4 completes nine of ten cycles within 75 seconds, every part is within reach of the operator, and nobody leaves the station during a cycle. Measured over five shifts and 60 cycles.
Four weeks. One figure a supervisor can check with a stopwatch in an hour. A description of how the station will be working, which immediately produced a list of things in the way. And an unknown: at 38 percent, nobody knew how to get to 90.
It was reached on 4 March, two days early, after seven experiments.
The gap that kills routines
The dangerous day is not the day a target condition looks impossible. It is the day it is met. The pressure is off, the next one feels like paperwork, and three weeks later the record board is out of date and the conversation has quietly stopped.
The rule that prevents it is simple: the next target condition is written within three working days of the last one being met or closed. On L3 the second was set on 9 March, five days after the first was reached, and it moved from one station to all eight: every station completes eight of ten cycles within 75 seconds, no cycle above 90 seconds, and the line keeps the customer sequence for four weeks without a catch-up shift.
The third, set on 13 April, added the thing that makes a routine permanent: a second learner, with a target condition of their own.