One-Shift Floor Robot Pilot Plan: Zones, Metrics, Go or No-Go

What a one-shift floor robot pilot answers

A one-shift pilot answers one question for leadership: does a floor robot clean the zones we picked, to the coverage we need, without adding work for the crew? It does not prove a fleet case. It shows how a single unit behaves in your building, on your floors, with your traffic.

The plan runs in four moves. Choose two zones, define coverage and fault metrics before the first run, hold a fixed test window, then score the result against thresholds set in advance. None of this needs a long study.

The market is moving in this direction. According to IFR, professional cleaning robots grew 34% to more than 25,000 units sold in 2024, and floor cleaning is the main application. IFR's figures come from sample data from 294 companies, so treat them as a trend signal, not a forecast for your site.

How to pick the two zones

Choose one zone that is forgiving and one that is hard. The forgiving zone has open lanes and predictable traffic. The hard zone has racks, tight turns, a threshold or two, and a crew that already complains about it. A pilot that only tests the easy zone proves very little.

Keep each zone on a single hard floor type, so the results are not blended. Avoid any zone that will see construction, a seasonal stock change or a layout reset during the window. Those events will swamp the data you are trying to read.

Before the run, record three baselines for each zone:

• Mapped area in square meters, with obstacles and fixed equipment marked on a plan
• The crew's current hours and the exact tasks performed in that zone
• A named person who will sign off on handover cleanliness using one checklist

Which coverage metrics to set before the first run

Coverage is the share of the mapped floor that the robot actually passes over, compared with the zone's full area. Set the target before the window opens. A target written after the numbers arrive is a story, and leadership will treat it as one.

Track four things per zone: percent of mapped area covered, time from start to completion, passes missed, and whether the zone passes handover inspection. Use the same checklist the crew uses today, so the comparison is fair.

Keep the baseline honest. Time the crew on the same zone with the same checklist, on a comparable shift. If the crew's baseline is inflated or the robot's route is tuned by hand, the pilot measures the tuning, not the machine.

Which fault metrics predict trouble

Most pilots fail on faults nobody logged. Count every stop, every manual rescue, every blocked-route event and every time someone had to clear the path by hand. Record the time to recover each one, because a 20-minute stall at 2 a.m. costs more than its duration suggests.

Sort faults into a few plain categories: obstacle stops, sensor or navigation faults, low-battery returns, routes blocked by stock or carts, and human interventions. A fault that repeats in the same spot points to layout, a cart staging habit or a threshold, which means the fix belongs to the facility, not the robot.

Log every event in one shared sheet with a timestamp, zone, category and a sentence in plain words. The final report should come from that sheet, not from anyone's memory of the week.

Running the window without changing the rules mid-test

Hold the fixed window steady. Freeze the map, the schedule and the staff roles on the first day. If you change the route midweek, you have run two pilots, and neither one answers the question leadership asked.

Hold a ten-minute log review each day with the facility lead and whoever supports the unit on site. Agree on what counts as a fault before it happens, so the team does not argue about categories at the end of the week.

If a fault is genuinely a layout problem, fix the layout and note the change in the log with a date. Leadership can then see that the result reflects a corrected floor, which is more useful than a clean number that hides the problem.

The go or no-go template for leadership

Write the decision rules on one page before the window opens. Keep the thresholds specific enough that the call is mechanical once the data is in. A usable set looks like this:

• Go if both zones meet the coverage target on the agreed share of pilot shifts, with no zone missing it two shifts in a row.
• Go if manual interventions per shift fall below the crew baseline and do not climb in the final third of the window.
• No-go if one fault category dominates the log, or if a person must stay with the unit for most of the shift.
• Conditional go if the failures trace to layout, cart staging or a threshold, a fix is scheduled, and the zone is re-tested inside the same window.
• No-go if handover inspection fails in either zone on more than one shift without a documented, layout-based cause.

Where a lease fits after a go decision

Service Robot Co. leases robots only, through a 3-year lease with one fixed payment quoted after a site assessment. A go decision leads naturally to that assessment, which maps the zones you tested so the quote reflects your real floor rather than a brochure.

IFR's 2025 report ties this kind of arrangement to demand. It says RaaS fleet grew 31% to more than 24,500 units, and IFR president Takayuki Ito said companies are choosing subscription or rental agreements to adopt automation without a heavy upfront investment. Staff shortages are also named as a key driver for professional service robots.

Our US engineer network handles deployment, training and service for every unit, so the crew that ran the pilot does not inherit the fixes alone.

How to take the pilot result to leadership

Keep the memo to one page: the question, the two zones, the metrics, the results against thresholds, the decision and the next step. Attach the raw log. Leaders tend to trust a dated log more than a polished slide.

Name the one number that drove the decision, and the one that would change it. A clear no-go with a documented cause is a useful outcome. It saves the next budget cycle from repeating the same test.