When shopping for window cleaning robots for sale in 2026, check seven things before paying: hold-time spec, tether anchor rating, wet-glass incline, battery replaceability, pad and seal availability, service location, and a filmed power-loss test. A listing without those is a photo, not a product.
There are more brands on the market this year than at any point I can remember, which should be good news and mostly is not. The same chassis shows up under four names, with different badges and the same manual translation errors. So the checks below are about the machine, not the brand.
What should the product page tell you before you contact anyone?
Weight, suction pressure in pascals, cleaning width in millimetres, runtime, tank capacity, IP rating, and the maximum incline – with the wet or dry condition stated. If incline is missing, assume 0 degrees from vertical and nothing more.
A trustworthy listing also names the safety standard it was tested against, or at least the certification body. CE marking alone is a self-declaration for most of this category. It tells you the paperwork exists, not that anyone tested the drop behaviour.
Why does the battery question decide the product’s lifespan?
Because a lithium pack in weekly commercial use is at 70-80% capacity after two years, and the pack is also the safety reserve that holds vacuum if the machine slips. A non-replaceable battery therefore puts a hard three-year ceiling on the whole machine, regardless of how good the motors and seals are.
Ask the price of a replacement pack before purchase. If the answer is “we don’t sell them separately”, you have found the end-of-life date.
How do you test a used unit?
- Seal ring: press it flat against a clean pane and look for light gaps; a hardened ring will not seal at 30 degrees
- Tracks: run the machine dry up a mirror; any hesitation or hunting at the edges means worn treads
- Recovery pump: fill the tank, run a spray cycle, confirm dirty water returns rather than pooling
- Hold test: unplug mid-climb on a low window and count the seconds before it detaches. Under 60 seconds is a fail
- Battery: full charge, then run to empty and compare against the original spec
Twenty minutes and one mirror settle most of it. If a seller will not allow the hold test, walk away, whatever the discount.
Is a used unit actually a good deal?
Sometimes. A two-year-old commercial unit from a facilities company that upgraded to a facade system can be excellent value at 40-50% of list, provided the seal ring and tracks are still available as spares. That last condition is the whole ballgame. A discounted machine with no parts channel is a doorstop with a battery.
Be sceptical of “barely used” listings with a scuffed chassis. Roof robots get scuffed because they work. A machine in cosmetically perfect condition after three years on a commercial estate is more likely to have been a failed experiment, which is also fine, but ask why it stopped being used.
What about the accessory list?
Accessories matter more on a roof than on a window. A tether long enough for your actual height. A second pad set so a first-pass pad never touches the finish pass. A remote or app that works when the machine is on the far side of a glass roof, out of line of sight.
And a carry case. Not for aesthetics: a roof robot carried loose in a van gets its seal ring crushed under a toolbox, and that failure shows up as mystery grip loss six weeks later.
Who should buy new instead of shopping the resale market?
Anyone with a building where a fall would injure someone on the ground. The warranty, the service contract and the manufacturer’s incident support are worth more than the discount, and a secondhand unit gives you none of them. Same for a first robotic purchase: learning the workflow on a machine with known provenance is cheaper in the long run than learning it on a bargain.
For measurement of what “good” looks like on each model, our skylight and window robot review archive covers wet-glass incline behaviour and true runtime. The Lingfeng S1 review is worth reading if you are leaning toward new, because it shows the step and bar handling that a spec sheet almost never quantifies.
What do the review sites leave out?
Almost everything about roofs. Community threads on window robot subreddits are useful on consumer units for flat panes and near-useless on glass roof work, because most posters have never run a machine on an incline. Treat them as a source of failure reports rather than buying advice.
The other gap is service geography. A robot with a European distributor and a robot with none are the same product on day one and different products on day 400.
Does the price band tell you which building it fits?
Roughly, yes. Under 300 USD you are buying a flat-pane consumer unit for reachable glazing, and that is a legitimate purchase for a shopfront or a small office. The 400-1,200 USD band is where prosumer machines sit, usually with better sensors, a smarter path pattern and a serviceable pad system, and it is the band most small commercial buyers should look at first.
Above 2,000 USD you are in facade or roof-class territory, and at that point the purchase is a procurement rather than a shopping decision. Ask for a site survey, a training session and a spares list. A seller who quotes a 5,000 USD machine over chat without seeing the roof is selling you a machine that will disappoint you in a specific, predictable way.
Key Takeaways
- Demand the incline rating on wet glass and the hold time after power loss, in writing.
- Non-replaceable batteries cap machine life at about three years of commercial use.
- On a used unit, test hold time by unplugging mid-climb on a low window. Under 60 seconds is a fail.
- Spare seal rings and tracks available? If not, the discount is meaningless.