Gemba Walk in Manufacturing: Observe, Ask, Improve

Home » Resources » Gemba Walk in Manufacturing: Observe, Ask, Improve

Estimated reading time: 10 minutes

A machine is idle, an operator is waiting and a customer order is behind schedule. From the aisle, the answer looks obvious: get that job moving. A conversation at the workstation may reveal something different. The part is awaiting an inspection decision, the fixture is being used elsewhere, or the next operation has no room for it anyway.

A gemba walk in manufacturing helps you understand that situation before choosing a response. You go to where the work happens, observe it and ask the people doing it what helps or prevents progress. The purpose is to learn how the process works, not to catch someone doing it wrong.

The value comes from what happens next. A useful walk ends with a shared understanding, an owner for the unresolved question and a way to check the response. When that response changes production capacity or work priorities, SkyPlanner helps turn the agreed decision into a connected schedule instead of another round of manual queue changes.

What is a gemba walk in manufacturing?

A gemba walk is a management practice of observing work directly and asking questions before taking action. Gemba means the actual place. In manufacturing, that might be a machining cell, an assembly line, an inspection area or the handoff between two operations.

The walk is more than a tour of the factory. You follow a question about how value reaches the customer: how an order moves, what a worker needs to complete the next step, or why work stops between operations. Seeing the process at its source can expose a gap that a summary report leaves unexplained.

It is also different from a compliance audit. An audit checks work against specified requirements. A gemba walk seeks to understand the work and the conditions around it. You may discover a problem with a standard, but scoring employees is not the purpose of the visit.

Keep related improvement activities distinct. A 5S review focuses on workplace organization and its upkeep. Root cause analysis tests an explanation for a problem. A gemba walk can reveal a question worth investigating through either method; it does not prove the cause simply because a manager saw the event happen.

Prepare one question, not a factory-wide inspection

Choose a question narrow enough to investigate during the visit. “Why does this product family wait between machining and inspection?” gives the walk a purpose. “Find all the waste in the factory” invites scattered observations and a long list nobody owns.

Follow the work rather than a convenient route past every department. If the question concerns a handoff, visit both sides. A queue that looks unnecessary to one team may be waiting for information, acceptance or capacity controlled by another.

Before going to the floor:

• Tell the team why you are coming. Explain the question, the learning purpose and how findings will be followed up. Do not ask people to stage a better-looking version of normal work.
• Bring the people who can explain the process. Keep the group small enough for a conversation. Include the relevant supervisor or process owner, and involve the operators doing the work rather than talking only to their manager.
• Review enough context to ask intelligently. Look at the job, its route, planned timing and recent interruptions. Treat these records as a starting point, not a verdict.
• Arrange a safe time and place to talk. Follow site access and protective-equipment rules. Wait until an operator can speak without compromising a task.

SkyPlanner’s Job Report is useful preparation when you want to understand an operation’s timing. It shows actual logged time against a job, who worked on it and the workstation used, supporting comparison with the planned duration. A repeated difference gives you a focused question for the walk: what conditions made this work take longer than planned?

Do not arrive with the report as an accusation. A time difference can reflect the job mix, setup, interruptions or how the work was recorded. The conversation helps you understand which explanation deserves checking.

Observe the work, then ask the operator to explain it

At the workstation, first watch enough of the activity to understand what you are looking at. Notice the movement of the part and the information that accompanies it. What starts the operation? What must be ready? What tells the operator that the output can move on?

Look beyond whether the machine is running. Searching for a fixture, checking an unclear drawing revision or waiting for a release decision can shape the flow as much as machining time. A running machine can also be producing work that the next operation does not yet need.

Then ask the operator to help you interpret what you saw. Open questions make room for an answer you did not expect:

• “Can you walk me through this job and what needs to be ready before you start?”
• “What tells you which job to do next?”
• “What are you waiting for here, and who can help resolve it?”
• “Is this how the work usually runs, or is something different today?”
• “What makes this step harder than it needs to be?”
• “If we changed this, what might become harder for the next person?”

Follow an answer with curiosity, not a solution disguised as a question. “Why don’t you just run the other order?” assumes that the other order is ready and that changing sequence is harmless. “What prevents the next ready job from starting?” leaves room to discover the actual constraint.

Thank people for showing the awkward parts of the process. A workaround may be the only way they have found to keep production going. Understand the condition that made it necessary before deciding to remove it.

If you encounter an immediate safety or quality concern, follow the site’s normal escalation procedure. Learning about the process is not a reason to postpone a necessary response.

Separate what you saw from what you think it means

A useful observation is specific enough for someone else to revisit. Record the operation or job context, what happened, what the operator explained and what remains uncertain. Avoid labels such as “poor attitude” or “inefficient department.” They turn a process question into a judgment about people.

Keep three things separate:

Type of note	Example	What it supports
Direct observation	The operator left the cell to look for the fixture before the next setup.	A question about fixture availability and preparation.
Reported explanation	The operator says another cell sometimes uses that fixture without a clear return arrangement.	A lead to check with the other cell and the people managing the resource.
Hypothesis	The shared-fixture arrangement may be causing repeated setup delays.	Further checking before choosing a lasting change.

One visit shows one set of conditions. Ask whether the same difficulty occurs on another shift, with a different product or during a different workload. You do not need to delay every small improvement for an elaborate study, but the scale of the response should match the evidence and the risk.

That distinction matters for scheduling. Seeing one operator wait is not a reason to increase every job’s duration or make that customer’s orders the highest priority. First establish what changed and which planning input, if any, needs updating.

Close the walk with an owner and a useful next step

Before leaving, summarize what you understood and let the people doing the work correct it. Agree which question needs attention first. A short, well-owned follow-up is more valuable than a dozen instructions issued while walking away.

For the selected issue, record:

1. The problem in its job or process context.

2. The person responsible for the next check or action.

3. What will be checked or tried, and any approval needed.

4. When you will return and what improvement should be visible.

Choose the response that fits the finding. An unclear instruction may need the process owner to revise and test it with operators. A recurring failure may need a separate root cause investigation. A confirmed equipment problem belongs with maintenance. A priority conflict belongs with the people responsible for production commitments.

Bring the answer back to the team even if the proposed change cannot go ahead. Explaining the constraint and the next action keeps the conversation useful. Silence after a walk teaches people that showing a problem leads nowhere.

Example: turn a finding into a feasible production change

Imagine a machining cell where operators regularly walk to another area to collect the fixture needed for a changeover. During the gemba walk, the supervisor watches the next-job preparation and asks both cells how the fixture is shared. The fixture is available, but it has been left where the previous job finished. Conversations with both teams reveal that there is no agreed return location or handoff for the next setup.

The operators and process owner agree to test a revised arrangement. They also need a short period to reorganize the cell, during which its machine cannot produce. The supervisor approves that window with the planner rather than simply taking the machine out of service during an already committed job.

This is where SkyPlanner makes the follow-through easier. A Move/Organize workstation exception, with a description and time range, records the unavailable period. When Arcturus recalculates the schedule, it uses the changed capacity along with the configured routes, durations, priorities and other constraints. The Gantt view makes the revised timing of affected work visible.

The planner can review the effect on other orders instead of inserting the improvement locally and discovering the delivery consequences later. Priorities, rules and locks retain the team’s production decisions while Arcturus handles the scheduling calculation. That means less time repairing individual machine queues and a clearer view of which commitments need attention.

After the trial, the team checks whether fixture searching has actually decreased and whether the new arrangement works for both cells. If a lasting process change also changes the time needed for the operation, the process owner and planner review the relevant duration. They do not rewrite the planning assumption from a single observation.

SkyPlanner supports the work context and the resulting schedule. The operators and process owner establish whether the change solves the problem. Keeping those responsibilities clear lets you use AI scheduling confidently without mistaking it for a diagnosis.

A shared production picture makes follow-up easier

Mamec’s experience shows why the broader production picture matters. Before SkyPlanner, the manufacturer prioritized thousands of interdependent work orders using delivery dates on paper work orders. Management’s view of production load relied on subjective estimates, and keeping the whole sequence under control burdened production planning.

With SkyPlanner integrated with Oscar ERP, Mamec uses a shared planning environment where staff can look up an order’s status and schedule. Its customer account describes a clearly lighter planning workload and easier handling of production-sequence changes. The relevance to a gemba walk is practical: people can discuss a finding in the context of actual work and its place in the wider plan, rather than each department maintaining a different picture.

Return to see whether the work became easier

The next visit should test the response, not simply confirm that an action was marked complete. Is the fixture available for the next setup? Does the new instruction remove the ambiguity? Has the wait decreased, or has it moved to the next operation?

Use a measure that matches the original question. For a preparation problem, that could be time spent searching or the number of jobs held up for a missing item. Compare similar work and ask the operators what changed. A shorter recorded operation is not a success if it creates rework or makes the following step harder.

Set a repeatable cadence that fits the process and follow-up workload. There is no value in generating new findings faster than the team can investigate them. Return when the agreed response can be checked, and use what you learn to choose the next question.

A good gemba walk improves your understanding of production. SkyPlanner helps you put the resulting production decisions to work: shared job context before the conversation, a connected schedule after the approved change and less manual replanning between the two.

Book a SkyPlanner demo and bring one recent example of waiting on your floor. Start with what was observed, who owns the response and what the proposed change would do to the other orders. If you prefer to explore the environment first, try SkyPlanner with demo data.