Service robots in 2026: what to ask before you sign

The service-robot conversation is split in two. On one side, cleaning and floor-service robots: they’re already working every day in plants, malls and hotels. On the other, humanoids: impressive on video, but not a product a company can buy and put to work. Mixing the two conversations is the fastest way to make a bad decision.

Six questions before you sign

1. How much does it clean in practice — not on the spec sheet?

The square-meters-per-hour figure on the spec sheet is theoretical. Manufacturers that publish both numbers show real-world throughput several times lower, once you account for turns, obstacles, recharging and emptying tanks. Ask for the practical number, from the same manufacturer, on a floor like yours.

2. How many hours does it run between charges?

Real-world runtime for most commercial units is a few hours per charge, not full shifts. That determines how many units you need and how they’re scheduled.

3. What infrastructure does it need?

Some robots navigate without markers; others need tags installed around the building. And the station that fills and drains water automatically almost never shows up in the first quote, even though it changes the total cost.

4. What certification does it carry?

For commercial cleaning, the reference standard is IEC 63327. A manufacturer not publishing it doesn’t mean they don’t have it — but it does mean you should ask in writing.

5. What does a person still do?

Field reports agree that, after the robot passes, edges and corners still need a manual touch-up. The robot doesn’t replace the cleaning team; it changes what the team does. Plan for that from day one.

6. Who runs and maintains it locally?

The most common reason robots get abandoned isn’t that they fail on day one. It’s that nobody runs them well by month four. In ISSA’s survey on autonomous cleaning, a share of companies that tried them had already stopped using them, and there are public cases of chains pausing entire rollouts over missing ROI. Local support, spare parts and someone accountable for the outcome matter more than the brand.

What about humanoids?

In 2026, no general-purpose humanoid can be bought for productive commercial use. Several manufacturers acknowledge that their factory units are mainly there to learn, and one manufacturer’s CEO has talked about more than a decade before the residential market arrives. They’re worth watching closely and piloting on narrow tasks. They’re not worth budgeting as a replacement for an operation.

The bottom line

You buy a service robot for what it will do in your facility, with your team — not for its best video. Ask for real-world throughput, real runtime and the total cost including the station and support, and decide on day one what a person keeps doing.

Want to know if your facility is a good fit? Talk to Luz.