Robotics & Cobotics in UK Cleaning: Where the ROI Is Real

Most robotics demos are proving to be less than great. The machines work fine mostly, but the demos are shown in conditions that don’t exist in your contracts. For example, demos are often held on flat floors with no obstacles and nobody pushing a trolley through the path. But then, a robot is purchased and put out into the real world, but you spend the next 6 months going back and forth and eventually needing an operative to babysit it.

That experience has put a lot of people off the category entirely, which is a reasonable response to being oversold. But it's worth separating 'this robot was wrong for this environment' from 'robotics doesn't work in commercial cleaning', because those are very different conclusions, and one of them will cost you.

The cost context, briefly

The figures coming out of the industry in early 2026 are not looking shiny. A Facilities Management Industry survey published in February found that almost 90% of cleaning professionals are feeling inflationary pressure on operations, over 85% expect business costs to be a significant problem throughout the year, and 97% said the National Insurance increases had a serious impact on day-to-day operations.

On top of that, Statutory Sick Pay reforms that took effect this year remove the waiting day and lower earnings threshold, extending eligibility to more part-time workers. Analysts are putting the cost increase at around 20% for cleaning companies, costs that are nearly impossible to pass on in fixed-price contracts.

And labour availability hasn't improved. Brexit thinned the workforce and the gap hasn't closed. Around 30% of new cleaning jobs are proving hard to fill according to British Cleaning Council data, before you factor in the churn that comes with shift-based physical work. Recruiting and training are expensive, and when someone leaves after eight weeks, the cost doesn't show up anywhere obvious on a P&L, which is why it tends to get underestimated.

None of this is news to anyone running contracts. The question is where automation actually helps with it, and where it doesn't.

Where it works

Large floor areas, cleaned often

Autonomous floor scrubbers have a clear, well-documented commercial case in the right environment. The right environment means: high square footage, open floor plan without a lot of rooms and obstacles, and cleaning that happens daily or nightly rather than twice a week. (Warehouses, distribution centres, large retail floors, airports, hospitals, education estates with long corridors and sports halls.)

In facilities above 75,000 square feet with that kind of layout, independent analysis puts payback periods at 18 to 30 months. Across commercial applications more broadly, 18 to 24 months appears repeatedly in total cost of ownership modelling when the conditions are right. That's a real return, and it's been tested against actual deployments rather than vendor spreadsheets.

The calculation isn't difficult, but it needs to be honest. Full employment cost for the labour currently doing that floor scrubbing scope, not just base hourly rate, include employer NI, sick pay, holiday, recruitment. Against that: machine purchase or lease cost, maintenance, downtime. Then an honest assessment of whether the floor plan actually suits autonomous navigation. Run those numbers and you get a payback period. If you haven't done that before buying, you're essentially guessing.

One person, more ground, the cobotic model

Full automation gets most of the attention but the cobotic approach (a person and a machine working the same space simultaneously, each doing what they're suited to) is where a lot of UK businesses are finding more immediate practical gains. The machine handles the large, repetitive floor coverage. The operative handles detail cleaning, the areas the machine can't navigate, client-facing interactions, anything requiring judgment.

The output is one person covering significantly more square footage per shift without quality dropping. That changes the labour cost per contract. It also matters for retention, you're taking the most physically grinding part of someone's shift and handing it to a machine, which is a more compelling conversation with a cleaner than a pay rise in a lot of cases.


And it's a differentiator in a tender. 'We deploy cobotics on contracts of this type' is a different conversation from 'we have robots', the former implies a considered operational methodology, the latter sounds like something from a brochure.

Where it doesn't work

The failures are fairly predictable once you know what to look for.

Small or fragmented environments (anything under 20,000 square feet, or buildings with lots of rooms, tight doorways and irregular layouts) rarely generate enough savings to justify the investment. Navigation degrades and map-training takes time- the machine ends up creating overhead rather than removing it.

Low deployment frequency kills the economics. A scrubber running two or three times a week instead of nightly means the payback period can stretch to four to six years. That's not a sensible use of capital. The machine needs to run hard and often to earn its keep.

And anything that requires significant daily human intervention isn't automating anything, it's adding a task. If your operative is restarting it, redirecting it, untangling it from obstacles every shift, the time saving isn't there. This is often where the cheaper end of the market falls down. The machine looks the same in a catalogue as a commercial-grade unit. In a real contract environment, it's not the same machine at all.

Window-cleaning robots, for most commercial contexts, still don't make financial sense outside specialist high-rise applications. The technology exists. The environments where the maintenance overhead, access consistency, and scale all line up to produce a clean ROI are still limited.

The question that actually matters before you buy

What's the unobstructed floor area on this contract, and how many times a week does it get cleaned?

That number, run against full labour cost and realistic machine costs, tells you more than any demo will. A floor scrubber paying back in 18 months on a 50,000 square foot logistics contract is the same machine taking four years to pay back in a mixed-use office building with twelve rooms per floor. Same equipment, completely different commercial outcome, depending entirely on where you put it.

The vendors won't always tell you which category your contract falls into. That assessment needs to happen on your side, before the conversation gets to price.

Where Foremost fits into this

While supplying robotic and cobotic machines, we're also the first to say that the conversation shouldn't start with the equipment, it should start with the contract.

-        Is the floor area right?
-        Is the frequency right?
-        Does the floor plan suit autonomous navigation or would a cobotic deployment make more sense?
-        What does that change look like operationally for your people on site?

To get real commercial value from robotics an assessment needs to be made first and machines bought accordingly. Don’t assess first, and the machines will end up permanently parked in storage rooms.

If you've been dismissing the category because of bad early experiences, that's fair, there's been a lot of poorly matched equipment sold on hype. But the underlying economics, in the right environment, are solid enough now that the question is worth revisiting. As a margin decision rather than a technology one.

Facebook
X
LinkedIn