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

What does a one-shift floor robot pilot actually prove?

A one-shift pilot answers one question: does a floor robot take real work off the night crew in one zone without making the rest of the shift harder? It does not need a facility-wide rollout to answer that. One robot, one zone and one overnight shift give you enough nights to see patterns, and little enough exposure that a poor fit costs weeks rather than a long commitment.

The timeframe is the first decision. GreenPoint, a Manhattan cleaning contractor, recommends a two-week baseline of labor logs, complaint tickets and appearance scores by zone, followed by a four-week pilot with the robot (vendor-reported). Sproutmation, also a vendor, suggests a 30-day pilot with one robot in one high-value zone. Either window can work. What matters is that the baseline comes first.

The wider market gives context. IFR reports that professional service robot shipments rose 24 percent to almost 250,000 units in 2025, and that professional cleaning robots rank third, with floor cleaning as the main application.

Which zone should the robot take first?

Pick a zone that is open, repetitive and hard-surfaced. GreenPoint says corridors and open hard-surface zones of 10,000 to 40,000 square feet per shift are often where autonomy shines (vendor-reported). Avoid zones with frequent furniture moves, loose cords or foot traffic that stops without warning. Those interruptions will dominate your first two weeks and tell you more about the site than the robot.

Do the arithmetic on hours before you commit. GreenPoint estimates that an operator spending 2.5 to 4.0 hours a night on repetitive scrub passes could move 60 to 80 percent of that driving time to edge work, detail cleaning and inspections, without changing headcount. The company calls this an estimate, not a measured result. Treat it as a hypothesis to test, and keep the existing night tasks in place so the pilot isolates the robot's effect.

What should you record before the robot arrives?

The baseline is what keeps the pilot honest. Capture the same measures you will track later: crew hours in the pilot zone, complaint tickets that mention the floor, and a simple appearance score by zone, taken on the same nights of the week. Without these, the first robot week has nothing to compare against, and the result becomes a debate about impressions.

Keep the baseline boring. Use the same people, the same checklist and the same time window. A pilot is a comparison, and comparisons break when the way you measure changes midstream.

How do you set success criteria without a vendor's numbers?

Vendor case figures are useful for deciding what to measure, but they are not your targets. Nabuurs Supply Chain Solutions in Oosterhout, Netherlands, has operated a Tennant T16AMR since July 2025 in a 60,000 square meter warehouse alongside conventional operations. The case report puts autonomous operation at an average of 5.57 hours per week, and usage data shows about 1,774 square meters cleaned per day on average. Tennant estimates routine floor cleaning labor fell by roughly 60 to 70 percent, a vendor estimate. According to logistics press coverage, the machine ran autonomously for almost 63 percent of its total usage time. The trial and ROI analysis were carried out by Nabuurs and Tennant before full use.

Gausium reports a different kind of gain. Two Omnie robots on a São Paulo metro line cut platform cleaning from about 3.5 hours to roughly 40 minutes, and water use from around 980 liters to 15 liters per operation (vendor-reported). Use these figures as prompts. Your criteria should be expressed as your own crew hours, your own complaint count and your own water and chemical use, each compared against your own baseline.

What is the go or no-go rule, and when is it written?

Write the rule before the robot runs its first night. A decision made after two weeks of results tends to bend toward whatever the manager already hoped for. Put the metrics, the comparison period and the threshold on one page, and have the person who will carry the decision upstairs sign it.

Build the rule around three outcomes. Go means the robot meets the threshold on the metrics that matter, and the crew can keep the zone clean without extra hands. Extend means the signal is promising but the data is thin, or the fault pattern looks fixable, such as a route that keeps stopping at one doorway. No-go means the robot is not reducing work, or it is creating new work that cannot be fixed on site.

What should the crew log every morning?

Morning logs are where the pilot earns its keep. A weekly summary hides the bad nights, so review each run the next day while the details are still fresh. Keep the form short enough that a supervisor can complete it in a few minutes.

Gausium says its workstation handles charging, water refill, wastewater discharge and self-cleaning automatically, so the robot can work overnight without anyone standing by (vendor-reported). Test that claim on your floor, and log every time a person had to step in.

• Whether the planned route finished, with the coverage figure the robot reports
• Each unplanned stop, with time, location and whether a person intervened
• Map or route errors, including doorways and zones that blocked the path
• Charging, water refill and wastewater events, and how long each one took
• Complaint tickets and appearance-score readings for the pilot zone
• Manager minutes spent on the robot, including restarts and cleanup

How do you judge fit and service before go-live?

Sproutmation argues that overnight success usually depends more on fit and support than on headline spec sheets, and that a deployment should include mapping, route tuning, operator ownership and a support path (vendor, 2026). Passable width, tank capacity, docking and refill friction, map stability and service response after go-live all belong on the checklist. A machine that is slightly slower on paper but refills without a fight can outperform one that wins the spec sheet.

Name one person who owns the robot each night, and agree on a fault response window before the pilot starts. Ask the vendor what happens on night three when a sensor fails, not only what the warranty says.

Where does a lease fit into a one-shift pilot?

Service Robot Co. leases robots only, through financing on a three-year lease with one fixed payment quoted after a site assessment. For a pilot, that means the site assessment comes first, so the zone, the routes and the integration plan are settled before the baseline period ends. We are OEM-neutral, so we select the robot that fits your floor across manufacturers, then handle deployment, integration, crew training and service through a nationwide US engineer network, with one partner to call.

A lease with a fixed payment makes the go or no-go decision easier to read. The pilot is judged on floor results, not on how quickly a deposit becomes a sunk cost.