What should I look for when I go and watch the work?
Plan what you will watch, what you will count and what you will ask before you go — because every failure mode of a Gemba walk is decided before you arrive.
Gemba Walk Planner · measure · White Belt · free, no account needed
Use this when
- You are about to go and watch a process for the first time
- A previous visit produced impressions and no data
- You suspect the documented process and the real one have diverged
Gemba Walk Planner
sandboxOne process, named. Otherwise the walk becomes a tour.
The physical place the work happens.
A process at 3am is not the same process.
What you will watch2
Things you could see happen. Not what you would conclude about anyone.
- Whether pickers understand the new layout
- Check that they follow the pick standard
What you will count0
Something simple and repeatable, so the walk produces data and not only impressions.
What you will ask2
Open, not leading. "Walk me through…", "show me the last time…".
- Is the standard being followed?
- Wouldn't it be quicker with a second bench?
no time or shiftNo time or shift stated. A process at 10am on a Tuesday is not the same process as at 3am on a Sunday — different staffing, different volume, different workarounds. Whichever you watch, say which one it was.
audit framing"Audit" frames the visit as an inspection. The moment the people doing the work believe they are being audited, you stop being told anything — and that does not recover within the visit. Go to learn how the work actually happens, including the workarounds, which are the most useful thing anyone can show you.
observation is a judgement"Whether pickers understand the new layout" — "understand" is a state of mind, which cannot be watched. Replace it with what you would actually see happen. Not "whether they understand the standard" but "what they do when the part does not fit".
too few observations2 things to watch. Fewer than 3 usually means you have decided what you expect to find, and a walk conducted to confirm something confirms it.
nothing to countNothing to count. Without a count the walk produces impressions, and impressions lose every argument against a spreadsheet — including the wrong spreadsheet. Count something simple: how many times a person walks away from the bench, how many items are waiting, how often the exception path is taken.
closed question"Is the standard being followed?" can be answered yes or no, which ends the conversation at the point it should start. Open it: "what happens when…", "walk me through…", "show me the last time…".
leading question"Wouldn't it be quicker with a second bench?" — "Wouldn't it" tells the person what answer you are hoping for. You will get agreement, and agreement is not information.
How this is calculated
No statistics and no score. Every check is about wording, because wording is where intent shows up and intent is the only thing that decides whether a walk is worth doing.
- A named process, a place, and a time including the shift.
- No audit language — a visit read as an inspection stops being told things.
- Observations you could watch, not verdicts you would reach.
- At least one thing to count.
- Questions that are open rather than yes/no, and not leading.
Source: Imai, M. (1997), Gemba Kaizen, Ch. 2 — the five golden rules of gemba management; Rother, M. (2009), Toyota Kata, on coaching questions rather than inspection.
How this is calculated
Structural and linguistic check, no statistics and no score. Verifies that a process, place and time-with-shift are named; flags audit language that turns the visit into an inspection; flags observations that are verdicts about people rather than things you could watch; requires at least one thing to count; and flags questions that can be answered yes or no, or that lead.
Source: Imai, M. (1997), Gemba Kaizen, Ch. 2, the five golden rules of gemba management; Rother, M. (2009), Toyota Kata, on coaching questions rather than inspection.
Learn the method
- Go and see: what a walk gives you that a report cannotWhite Belt · 12 min · free