Key takeaways
  • The purpose is to see the work where it happens, not to inspect people. If your presence changes behaviour, you are not seeing the work.
  • Go with one question, at a fixed time, and follow one process end to end rather than wandering.
  • The walk is worthless without follow-up: what you find becomes tracked work or the next walk is theatre.
The signal you are doing it wrong
The area is unusually tidy when you arrive

If word travels ahead of you, you are seeing a performance rather than the work. Unannounced, regular and boring is the goal: frequent enough that nobody bothers preparing.

What a gemba walk is for

Gemba means the actual place. The purpose is to understand how work really happens, which is reliably different from how the procedure says it happens and from what gets reported upward. You are looking for the gap between those, and for the workarounds people have quietly invented.

It is explicitly not an audit. An audit compares against a standard and produces findings. A walk builds understanding and produces questions. Confusing the two makes people defensive and ends the flow of useful information.

Go with one question

A walk without a focus becomes a tour. Pick one: where does material wait, where do people walk furthest, where does the operator have to think hardest, what gets checked twice, what did we change last month and did it help.

Follow one process, end to end

Rather than visiting areas, follow one job from arrival to completion. This surfaces the handovers, and handovers are where most delay and error live. Areas look fine individually; the gaps between them are where the problems are.

Questions that work

  • What is the most annoying part of this job?
  • What do you do when it goes wrong, and how often is that?
  • Show me the last time this failed. What happened next?
  • What would you change if you could change one thing?
  • What do you have to remember that the system does not remind you about?

That last question is usually the most productive. Anything held only in someone's head is a risk that leaves when they do, and it is invisible from an office.

Questions that do not work

  • Anything answerable with yes. "Is everything alright?" always gets yes.
  • Anything that sounds like an audit. "Why is that not labelled?" ends the conversation.
  • Anything you already know the answer to. People notice, and stop volunteering.

What to do afterwards

Write down what you found the same day, decide which items become actions with owners and dates, and tell the people you spoke to what happened to their input. The second and third parts are what make the next walk productive. A walk that generates observations and no change trains people to say nothing.

Making it a habit rather than an initiative

Same day, same time, every week, on a schedule you keep even when busy. Consistency does more than duration. Putting the walk on a schedule with its own short checklist, and turning what you find into tracked actions with owners, is what separates a practice that lasts from one that fades in a month.