For problems that keep coming back, three tools sit on the problem record itself: a fishbone, the five whys, and an A3. They are deliberately ordinary. What matters is that they live on the problem rather than in a folder somebody has to go and find.
They are also entirely yours to fill in. Kumiko does not write your analysis. Judgement about cause is the part you would not want a machine doing, so it stays with your people.
The fishbone: widening the search
Start here when you do not yet know where to look. You get six headings to think against:
- People. Skills, handovers, who was available.
- Process. The steps as written, and the steps as actually performed.
- Equipment. Tools, machines, software.
- Materials. Inputs, stock, data arriving from somewhere else.
- Measurement. What you check, how often, and whether the number is trusted.
- Environment. Conditions, layout, time pressure, anything about the surroundings.
Put candidate causes under each. The headings exist to stop you settling on the first plausible answer, so try to write something under at least four of them before you choose which one to pursue.
The five whys: going deeper on one
Take the cause you believe in most and ask why it happened. Then ask why that happened. Keep going until you reach something you can actually change.
The usual failure is stopping too early, at “the procedure was not followed” or “the file was wrong”. Those are restatements, not causes. Keep going until you reach a decision, a design, or an absence: nobody owns this step, the check happens after the point where it would have helped, the form allows it.
Five is a guide, not a target. Three is fine if the third answer is something you can change. Seven is fine too.
The A3: turning it into a decision
The A3 is the one-page summary: what the problem is, what you found, what you propose, and what you expect to change as a result. Write it for somebody who has not read the thread.
Keep it short. The whole value of a single page is the argument it forces you to have with yourself about what matters.
Countermeasures
Every action that comes out of the analysis needs a named owner and a date. Without both it is a wish, and it will read as one to whoever opens this problem next year.
Vague ones are worth rewriting before you save them. “Improve communication” can never be finished. “Add the handover check to the closing routine, Priya, by 14 March” can.