Across three 2026 deployments, skylight cleaning robots cut cleaning cost per m2 by 45-62% at a shopping mall, 38% at an airport terminal and 55% at a factory. The factory gained the most because its sawtooth roof was previously cleaned least often.
A single case proves little. Three sites with different glass, access and dirt tell you where the method earns its keep and where it does not.
How did the mall atrium perform?
A 750 m2 atrium glazed roof, cleaned quarterly by rope crew before. A Lingyun Y3 with recovery tank now runs monthly at night. Cost per m2 fell 45%. More important, glass clarity held through trading hours because dirt never accumulated for a full quarter. The mall kept rope access on retainer for an annual deep clean of the high steep section.
What did the airport terminal gain?
A 1,800 m2 terminal skylight above a check-in hall. The gain was smaller, 38%, because rope crews already worked efficiently on a regular programme. The robot’s value there was disruption: no rope lines in a live terminal, lower passenger visibility of cleaning works, and a 2-4 hour night window instead of a full access mobilisation.
| Site | Glass area | Old cost / m2 | New cost / m2 | Change |
|---|---|---|---|---|
| Shopping mall | 750 m2 | USD 3.2 | USD 1.75 | -45% |
| Airport terminal | 1,800 m2 | USD 2.4 | USD 1.5 | -38% |
| Factory sawtooth | 1,100 m2 | USD 2.9 | USD 1.3 | -55% |
Why did the factory win biggest?
Because it was cleaned least often. The sawtooth roof had gone a year between cleans, so each visit started from heavy dirt and cost more in labour and water. Moving to a monthly robot cycle removed the heavy first pass entirely. This is the pattern across industrial sites: the robot’s saving scales with how neglected the glass was.
Who should not expect these numbers?
Buildings already cleaned monthly by an efficient crew on easy access. There, the saving shrinks to 10-20%, and the robot’s main value becomes safety and scheduling rather than cash. Also, sites with a single small skylight under 200 m2 do not reach these figures at all.
One trap from the airport job: tether routing past moving walkways and security lines took more planning than the cleaning. Sort the route before you buy. Case visits and layout reviews start at contact.
What did all three sites get wrong first?
Each site underestimated setup. The mall assumed the robot would run from the first visit; the tether route over two balconies needed a rethink after the first night. The airport lost a week to security access for the robot and its operator. The factory discovered its roof access hatch was 60 cm wide, narrower than the robot’s transport case, so staff had to carry the machine in sections.
None of these were robot faults. They were access and planning faults. The lesson repeats: survey access before you buy, and walk the full route the robot and operators will take.
How do you replicate the factory result?
Find the area you clean least often, then ask whether a robot can reach it on a regular cycle. Neglected glass carries the biggest saving because the first pass is the expensive one and a regular cycle removes it. The factory’s 55% saving came almost entirely from ending the yearly heavy clean, not from a faster machine.
What should be measured in the first 90 days?
Track four numbers: m2 cleaned per cycle, hours per cycle, water per cycle and any aborted cycles. Aborted cycles are the quiet cost, because they consume a window without delivering clean glass. If aborts exceed roughly one in ten cycles, the problem is usually cups, layout or wind, and it will erode the projected saving fast. Measure early, fix early.
What did the operators say mattered most?
In all three deployments, operators named the same thing: predictable setup. When the tether route, water point and anchor are fixed and known, a cycle becomes routine and the operator stops improvising. Improvisation is what turns a three-hour job into six. The technology was never the complaint; the logistics were.
The second theme was spare parts. A site that holds cups, a brush and a filter keeps working through a worn part. A site that orders on failure loses days. The factory, with the largest roof and the heaviest dirt, held the most spares and had the fewest interruptions.
Would these sites buy the same robots again?
The mall and the factory said yes without hesitation, mainly for consistency and cost. The airport was more measured: it valued the disruption reduction highly but found the tether routing and security access demanding, and would want that planned earlier next time. All three kept a rope crew on retainer for edge and complex work, which is the honest shape of a mature program.
The takeaway across sites is not that robots replace crews. It is that robots absorb the repeatable 80% and leave the crew to do the irregular 20% they are best at.
Key Takeaways
- Cost per m2 fell 45% at the mall, 38% at the airport, 55% at the factory.
- The factory gained most because its roof was previously cleaned rarely.
- Airports gain disruption reduction as much as cash.
- Sites already cleaned monthly may save only 10-20%.
- Plan tether routing around live traffic before purchase, not after.