Timing a Supermarket Robot Setup Day, Step by Step

What does a supermarket setup day really involve?

According to Gausium's October 2026 blog, a 3,000 m² supermarket can go from zero to running in about 20 minutes, where the same work previously took one to two days. Gausium's own page adds that actual time varies with site conditions, and that caveat is the one to read first.

No independent measurement of that 20 minute figure turned up in our research. Every timing claim here comes from a vendor or distributor, and the one to two day baseline is the vendor's own comparison, not a measured industry average. Treat the 20 minutes as the mapping and configuration step, then add the walk, the dock and the crew time around it.

A realistic day has four phases: a survey before delivery, the physical setup, the map and first task, and a supervised run that ends with someone signing off on the route. The sections below take each in order.

What should happen before the robot arrives?

The survey decides most of what follows. Sproutmation's grocery guide tells buyers to measure real pinch points at end caps, checkout approaches and stockroom turns before requesting a quote, and to treat the narrowest passable section as the number that matters most.

Tennant's process guide starts the same way: identify where the robot will be housed and charged, then walk the areas to be cleaned and agree on the scope, including routes and optimal times. Robolist's deployment math adds that facilities should factor in dock time at the start and end of each run, a line item that is easy to forget on a spreadsheet.

Service Robot Co. quotes one fixed lease payment only after a site assessment, so this survey is also the step where a store's layout gets written down in full. A walk done properly before the truck arrives saves the most time on the day itself.

How long do the dock and the first map take?

SoftBank Robotics says its Scrubber 50 Pro can map a 20,000 sq. ft. site in about 30 minutes. That is a vendor figure for one model at one site size. Mapping speed matters less than what happens around it, because the dock has to be ready before any task can start.

SoftBank notes that a deployment needs a station for recharging, a way to empty the waste tank and a supply to refill water. Gausium describes its Smart Deployment as site staff building the map while the robot handles annotation, task creation and parameter settings. In shelved areas, only the main aisles need covering, and the shelf zones fill in automatically.

Gausium also says sites that marked the dock during the survey can run a first task the same day, while sites that did not usually stop to resolve power or drainage. That one point explains much of the gap between a 20 minute demo and a long first day.

Which store features stretch the clock?

Layout is the variable no vendor can fix for you. A distributor page lists Gausium's Phantas as cleaning up to 1,180 m² per hour, and its Omnie as up to 3,000 m2 per hour. Those are throughput figures, and they say nothing about how long a crooked end cap or a stacked pallet takes the robot to route around.

Runtime matters too. Robolist says most mid market autonomous scrubbers run three to six hours on a charge, and that a full tank supports 45 to 90 minutes of continuous operation at typical solution rates. A supervised first run has to fit inside those limits, with a recharge plan built in.

How much staff time does day one take?

Staff time is the variable a vendor timeline rarely prices in. Gausium's case study with AB Vassilopoulos describes a rollout carried out between January and May 2026 with 9 Omnie units and 16 Phantas units, and reports a 20% reduction in the time staff spend on manual floor cleaning. That is a vendor-reported result over months, not a single afternoon.

Sproutmation advises that the highest value retail deployments usually reserve the main route for windows before opening, after close or low traffic periods. Those are often the hours when a store is thinnest on staff, so the day one plan should name the person who walks the first run and the person who answers if a route stalls.

How do you lay out your own setup timeline?

Build the day backward from the first supervised run. Work out who is free to walk the floor, then place the dock, then map, then run. The list below is the order we would check, not a promise about minutes.

• Confirm the dock position, power access and drainage before the crew arrives.
• Assign one staff member to build the map with the robot's guidance and one to shadow the route.
• Mark no go zones and check that the robot blocks them without manual edits.
• Run the first task in a low traffic window and watch the narrowest passages.
• Log the actual minutes for each phase, so the next store starts from a real baseline.

What do other vendors' timelines say?

Vendors describe setup in broad strokes. Robotlab says deployment is typically days, not months, covering mapping, route configuration, staff training and a handover. ToDo Robotics says most installations are completed within days rather than weeks, and its scope includes docking planning for charging, water access and drainage.

These ranges are wider than the 20 minute figure, and they describe the full project rather than the mapping step. Robolist warns that most cleaning robot deployments underperform because facilities managers skip the pre deployment math. The gap between a demo minute and a finished store is where that math lives.

How should you judge the claim after go live?

Robolist's illustrative example uses 60,000 square feet of compatible hard floor split across three separate zones, with only three hours a night available for autonomous cleaning. That window arithmetic, not the mapping minute count, decides whether a store is a fit.

Gausium says actual time varies with site conditions, so a store manager should ask for the conditions behind any timing quote, including dock access, aisle widths and the hours the robot can run. Compare staff hours on manual floor work before and after, using the same measure each time, and judge the claim on that record.