Gemba Walk Planner: Keep the Improvement Visible Months After Handover

The Gemba Walk Planner does one plain job. It lets you plan recurring gemba walks so someone observes the process where it happens and talks to the people doing the work.

You reach for it in the Sustain phase of RDMAICS, usually at handover. The project team is winding down. The control plan is written. The dashboard is built. What is missing is a reason for a named person to stand in the work area on a given date and look at the process with their own eyes.

A gemba walk is not an audit. You are not there to score anyone. You are there to see what the data cannot show you: the workaround that has crept back, the queue that forms at eleven each morning, the exception that nobody logs because logging it takes longer than fixing it.

The planner turns that intent into a schedule with dates, areas, leaders, focus themes and participants. Then it counts what has actually happened against what you said would happen. That count is the useful part.

The tool opens with sample data loaded. The case below is fictional and is used here only to show how the tool behaves. Every number quoted comes from that sample data or from the result the tool calculated.

The scenario is an accounts payable improvement that has just moved into Sustain. The team has worked across the whole invoice lifecycle, from document capture through to payment posting. Rather than one broad walk, the plan splits the process into four areas and gives each one its own date, leader and focus theme.

The four walks run weekly through June. Invoice processing on 5 June, OCR capture on 12 June, approval routing on 19 June and payment posting on 26 June. Each walk has a leader, a focus theme and named participants.

That spacing matters. Weekly walks across four consecutive areas mean the whole process gets eyes on it within a month, and no single leader carries the entire load.

The input table has six columns. Walk date, area and status are required. Leader, focus theme and participants are optional, but leaving them blank wastes most of the value of the tool.

The area column should name the part of the process you will stand in, not the department. In the sample, the four areas are Invoice processing, OCR capture, Approval routing and Payment posting. These are process steps. Anyone reading the plan knows exactly where to go.

The leader column names one person who owns the walk. The sample spreads the load across three leaders: the AP Manager takes two walks, the Process Owner takes one and Finance takes one. A single leader on every walk is a warning sign. It usually means the sustain plan belongs to one person rather than to the business.

The focus theme is the question you are carrying into the room. Exception handling for invoice processing. Data quality for OCR capture. Cycle time for approval routing. Reconciliation for payment posting. One theme per walk keeps the conversation specific. Without a theme, walks drift into general chat and produce nothing you can act on.

The tool produces a summary block and then a card for each walk. The summary gives you five counters: walks, completed, leaders, areas and past due date.

In the sample, the counters read 4 walks, 0 of 4 completed, 3 leaders, 4 areas and 4 past due date. The date range is 5 June 2025 to 26 June 2025.

Read those counters as a pair rather than one at a time. Four walks across four areas with three leaders is a well spread plan. That is the design. Zero completed against four past due is the execution. That is the reality.

The tool also offers an AI review that explains the result in plain language if you want a second reading of the same numbers.

The cards below the summary repeat each walk in full, with status, date, leader, focus theme and participants. They are useful when you are deciding which walk to rescue first. In the sample, every card still reads Scheduled.

The first decision is coverage. Four areas and four walks means every part of the process gets observed once. If you had four walks all pointed at invoice processing, the counters would show four walks and one area. That tells you the rest of the process is unobserved.

The second decision is load. Three leaders across four walks is a reasonable spread. If the leaders counter read one, you would reassign before the plan starts, because the plan would collapse the moment that person took leave.

The third decision is escalation. Past due date of 4 with 0 of 4 completed is not a scheduling problem. It is a signal that the sustain plan has no traction. One walk slipping is normal. Every walk slipping, with none completed, means nobody has picked up ownership since handover.

The honest response is not to reschedule all four and hope. It is to go back to the process owner, confirm who actually owns the walks now, and restart with fewer walks that will genuinely happen. Two completed walks beat four scheduled ones.

The planner gives you three slip signals, and they appear in order.

The first is past due date climbing while completed stays flat. In the sample, past due date sits at 4 and completed sits at 0. That is the clearest version of the signal. Walks are being planned and then not done.

The second is leader concentration. Watch the leaders counter against the walks counter. The sample shows 3 leaders for 4 walks, which is healthy. If that drifts towards one leader for every walk, the improvement has become one person's hobby rather than a business routine.

The third is area narrowing. The sample shows 4 areas for 4 walks. If later plans show four walks and two areas, you have lost sight of half the process. Usually the areas that drop out are the ones nobody enjoys walking.

The most common mistake is planning walks for a quarter at a time and then never updating status. The status field is what turns the plan into evidence. If every row stays on Scheduled, the tool will keep telling you 0 completed, and it will be right.

The second mistake is naming a function instead of a person as leader. The sample uses role names such as AP Manager, Process Owner and Finance, which is fine for a plan template. In live use, put a name behind the role. Roles do not walk. People do.

The third mistake is overfilling the participant list. The sample keeps participants to one or two per walk: Green Belt and AP Clerk, Green Belt and IT, Approver and AP Clerk, AP Clerk alone. Small groups talk. Large groups perform.

The fourth is skipping the focus theme to save time. A walk without a theme produces no record worth keeping, and you will not be able to explain later what you looked at.

The Gemba Walk Planner schedules and counts. It does not tell you whether the walks were any good.

It has no field for what you observed, what you asked, what the frontline told you or what you changed afterwards. You can mark four walks completed and still have learned nothing. The counters would look perfect. Keep your observations and actions somewhere else, and treat this tool as the attendance register rather than the minutes.

It also does not know anything about your process performance. It will not link a missed walk to a rising defect rate or a slipping cycle time. The connection between walk discipline and process health is one you have to argue yourself, using your control charts and your audit results alongside this plan.

Finally, it cannot distinguish a walk that was cancelled for a good reason from a walk that was simply forgotten. Past due date of 4 is a prompt to ask a question, not a verdict. Go and ask the question. That is the whole point of gemba.