Features · Gemba walks
Walk every aisle, every day, with a photograph of what you saw.
A Gemba walk shows you what is actually happening on the floor, and its weakness is coverage. What a recorded pass adds is not observation but reach, and a supervisor who arrives at the aisle already knowing which positions are wrong spends the time on the people rather than the search.
The problem with a walk
Observation does not scale, and memory is not evidence.
- It samples. A supervisor covers the aisles they can reach in the time they have, which is usually the same aisles, and the problems hide in the ones nobody reaches.
- It is not repeatable. Two people walking the same aisle notice different things, so a trend across weeks is a trend in who was walking.
- It produces notes, not proof. When a finding is challenged three weeks later, there is nothing to open.
- It costs supervisor hours that could go to the exceptions rather than to the search for them.
What the walk becomes
The supervisor still goes to the floor. They go with a short list of exceptions rather than with a clipboard and a route, so the hour is spent on why a position was wrong instead of on discovering that it was.
It changes who does the walking, too. The coverage comes from a driver who was already in the aisle, and the review is a screen reachable from any device, worked through before anyone leaves the office. Nothing about the drive is special, which is why two to three days of driver training is enough.
| What you want from the walk | Walking it | Recording it |
|---|---|---|
| Frequency | Weekly, if the week allows | Daily, inside a break |
| Consistency | Varies by who walked | Identical route, identical method |
| Evidence afterwards | Notes | The recording, per line |
| Supervisor time | Hours per walk | Minutes reviewing exceptions |
| Empty locations found | Only where someone looked | Reported automatically |
Continuous improvement
Four findings a walk is looking for, reported without walking.
Pallet in the wrong location
Put away drift, visible as a pattern across aisles rather than as one incident.
Pallet not in your WMS
Stock in a location your system does not know about.
Empty location where stock is expected
Capacity you think is full, and a picking failure waiting to happen.
Expected pallet missing
A targeted search instead of a recount, with the last known position attached.
What this does not replace. A Gemba walk is also about people, method and the conversation on the floor, and a camera on a forklift does none of that. Sentispec Inventory replaces the part of the walk that is checking what is physically in the racking, which is the part that samples badly and takes the most time. It does not observe how a task is performed, it does not count cases inside a partial pallet, and it will not tell you why a pallet moved. Keep walking for the rest.
Ten live sites hold 99% or better stock accuracy after go live, with the fastest recorded break even at one month.
See the ten published deployments and how each figure was measured
Do we stop walking the floor?
No. You stop using the walk to find out what is in the racking. Supervisors arrive at the aisle already knowing which positions are wrong, so the time on the floor goes to the people and the process rather than to the search.
How often can we run it?
As often as a forklift is free. A 20 minute run covers 500 to 1,000 pallet positions and fits inside a break, which is how sites move from a weekly walk to a daily record.
Who reviews the output?
Whoever runs the walk today. The discrepancy list is a short list of exceptions with an image behind each one, reachable from any device, and it exports for a report or an audit file.
Does it work in narrow aisle racking?
Yes. Coverage is limited only by where a forklift can go, and both wide aisle and narrow aisle sites are running today. Height is limited the same way.
Next step
See a recorded pass of an aisle like yours.
A 30 minute demo on recordings from a site with your racking and your label format, ending with what a daily run would show you.