What Does a Cleaning Robot Cost? Every Factor, No Price Tag

"Cleaning robot" is not a product category with one price level. It spans compact scrubber dryers built for a few hundred to a few thousand square metres, dry sweepers that carry no water at all, and industrial large-area machines sized for logistics halls. Between the ends of that range, machines differ in size, tank volume, drive train, sensor stack, brush system and serviceability — several multiples apart, in every direction. Averaging across that range produces a number that describes no real machine.

The second reason is the area, and here the relevant question is not how many square metres you have but how many of them a robot can actually work: contiguous, largely free of obstacles, hard-floored, and laid out the same way tomorrow as today. Two thousand square metres of open sales floor and two thousand square metres of partitioned office with doors, carpet islands and chairs are two entirely different jobs — one is a straight route at full working width, the other is constant manoeuvring with a large share of edges left to a human.

The third reason is the service model. Purchase, rental and full-service do not merely shift when you pay; they change what is inside the number at all. A purchase price without wear parts, maintenance, spare-parts logistics and downtime cover is not comparable with a full-service rate that contains all of it. This is precisely why we publish no prices here: a figure quoted without knowledge of your floor, your cleaning frequency and your infrastructure is not a quote, it is a guess dressed as one.

Machine class is the largest lever, and it follows from the floor, not from a budget. Wet cleaning of hard floors means a scrubber dryer — the class covered in our scrubber robot guide, where working width, tank volume and manoeuvrability determine which machine fits which layout. Dry debris, dust and litter on large areas mean a sweeper instead, compared in our sweeper robot overview; these machines carry no water at all, which removes an entire infrastructure question. Within the wet class alone, a compact machine such as the CC1 with 15-litre fresh and waste water tanks and an official 700–1,000 m²/h sits in a different world from a ride-on-scale industrial unit, and both are correctly called cleaning robots.

Infrastructure is the factor most consistently underestimated because it is invisible in the machine spec. A docking station that charges and — in the case of the CC1 workstation series, which Pudu sells separately — automatically supplies fresh water and drains waste water needs a location: a parking spot with a socket, a water connection and a drain within reach. If that does not exist, either you build it, or you accept that a human fills and empties the machine every cycle. Both are costs; only one appears in an offer. Dry sweepers sidestep this entirely and instead need a place to empty the dust container.

Setup is a real one-off item: mapping the site, programming and testing routes, defining no-go zones, and instructing your staff. It is not large, but it is not zero — and it comes back whenever the layout changes. A shelving rearrangement, a renovation, a new seasonal display area: each can require the map to be reworked. In fast-changing retail floors this is a recurring item, not a one-off.

Finally, accessories and options: spare brush sets and squeegee blades, filters, mobile water tanks for sites without plumbing, dust-bag options, and integrations such as elevator or access-gate control for multi-storey buildings. None of these is decisive on its own; together they routinely account for the gap between an advertised base configuration and the machine that can actually do your job.

Wear parts come first, because they are the only running cost that is genuinely unavoidable. Brushes, squeegee blades and filters are consumables, and their service life is not a property of the machine but of your floor: abrasive unsealed concrete in a warehouse eats a brush at a completely different rate than sealed PVC in an office. Squeegee blades are the part that shows wear earliest, and worn blades do not fail loudly — they leave streaks, which is a cleaning-quality problem before it is a cost problem. Ask any supplier for the wear-part list and the replacement interval they assume for your floor type, and treat a vague answer as an answer.

Then the consumables in the narrow sense: detergent and water on wet machines. Both scale with area, soiling and dosing, and both are easy to size once you know litres per fill and fills per cycle — with 15-litre tanks and a documented coverage per fill, that arithmetic is straightforward. Electricity is the smallest of the running items on machines in this class, but the honest way to handle it is to compute it with the kilowatt-hour price in your own supply contract, not with an industry average someone else picked.

Software and monitoring are the item most often overlooked at purchase. Fleet management, cloud dashboards, reporting and, where the machine reports over mobile networks, the data connection itself are frequently separate contracts with their own term — the machine is yours, the ability to see what it did is a subscription. Clarify before signing what happens to reporting when the software contract ends.

Repairs and spare-parts availability are the item where the real risk sits, and it is not the invoice. A robot waiting three weeks for a part is not saving anything; it is producing cleaning costs, because the floor still has to be cleaned during those weeks — by hand, at short notice, usually at the worst rate. The questions worth asking every supplier: Are spare parts stocked in Germany? What is the committed response time? Is a replacement machine provided during repair, and at whose expense? Lastly, the labour that remains: tank service, daily visual checks, emptying and rinsing the waste-water tank, and the edges, corners and sanitary areas no floor robot handles. Without a docking station, tank service is a daily human task — which is exactly why the dock, an acquisition cost, can be the item that removes a recurring one.

Purchase is capital expenditure. You own the machine, you depreciate it, and you carry every risk attached to it: wear, repair, downtime, spare-parts supply, residual value and technological ageing in a product category that is still moving quickly. The advantage is control and a clean balance-sheet story once the machine is written down. The requirement is an organisation that can actually run a machine — someone responsible for service intervals, a stock of wear parts, and a plan for the week the robot does not work.

Rental is operating expenditure with a plannable monthly structure, and its decisive variable is not the rate but the scope. A bare rental contract shifts the payment pattern and leaves maintenance, wear parts and repair with you; a maintenance-inclusive one does not. This is where offers become incomparable without anyone lying: two contracts with the same label can allocate risk completely differently.

Full-service is operating expenditure in which the provider carries maintenance, wear parts and repair and, on failure, supplies a replacement machine. What you procure is not a robot but availability of a cleaned floor. Future Machines works this way — we operate our fleet under a full-service rental model, which is also why the running-cost items in this article are not theory to us; they are our own cost lines. The structural trade-off is the mirror image of purchase: less control over the asset, no residual value, and in exchange no exposure to the machine's bad days.

One question separates the three models more reliably than any comparison table: what happens on the day the machine fails? Who pays, who comes, how fast, and does the floor still get cleaned in the meantime? Ask it of every offer you receive, purchase and rental alike, and require the answer in the contract rather than in the sales conversation. Whichever structure you choose, that answer is the part you will actually feel.

A robot is never compared against nothing. It is compared against what you pay today to get the same floor clean, and in commercial cleaning that is almost entirely labour — your own staff or a contracted service provider. So the baseline is knowable, and it is the one number in this whole topic that you can look up rather than estimate.

The public anchor in Germany is the sector minimum wage in the building cleaning trade, agreed between the collective-bargaining parties of the building-cleaning trade and declared generally binding, meaning it applies to all commercial cleaning employees in Germany. Effective 1 January 2026 it rose by around 5.3 percent from €14.25 to €15.00 per hour in entry wage group 1, and by around 4.3 percent from €17.65 to €18.40 per hour for skilled staff in wage group 6. For orientation: the statutory general minimum wage has been €13.90 per hour since 1 January 2026, so the sector floor sits above it.

Read those figures for what they are: wages, not prices, and floors, not averages. What an hour of cleaning actually costs you is higher — employer social contributions, holiday and sickness cover, supervision and administration, and with a contracted provider their margin on top. We deliberately do not convert this into a "robot saves X euros" claim, and you should distrust anyone who does without knowing your shift structure, your cleaning contract and your floor. Any such model presented as a fact is marketing.

One part of the status quo resists being expressed as an hourly rate at all: whether the hours can be staffed in the first place, particularly at night and in marginal shifts. That is a separate discussion, and we treat it in our article on staffing shortages in commercial cleaning. For a cost model it matters because it changes the alternative you are comparing against — not "cheaper cleaning" but "cleaning that happens at all".

Seven inputs make any offer comparable, and without them none is. First: the robot-capable area — contiguous, hard-floored, drivable square metres, not gross floor area. Second: cleaning frequency, in cycles per week per area. Third: what that cleaning costs you today, taken from your own payroll or your service contract for exactly that area. Fourth: the available time window, and whether cleaning has to happen during opening hours or can run at night. Fifth: infrastructure on site — parking spot, socket, water connection, drain. Sixth: the residual manual work that stays regardless — edges, corners, sanitary areas, tank service without a dock. Seventh: your downtime scenario, meaning what you will do and pay when the machine stands still for two weeks.

With those seven values, a quote becomes arithmetic instead of a feeling — and vendors can be compared on identical assumptions rather than on which of them assumed the more flattering ones. If a supplier does not ask for at least four of the seven before quoting, you are being given a number, not a calculation.

Which machine class fits at all is the question before the money question, and we walk through it in which cleaning robot suits your operation. After that, the most reliable figure available anywhere comes from your own floor: a trial deployment on site produces real square metres per hour, real cycle times including docking and manoeuvring, and an honest measurement of the residual manual work — the three inputs every cost model is most wrong about when it is built from a data sheet.