"To prevent damage ... please remove the device from your pants pocket before sitting down"

-- HTC / Google Android phone packaging

In the late 80s, my family built a house in Canada for which we bought fancy nitrogen-filled double-pane insulating windows. Each window came with a big red sticker, somewhat hard to remove, proclaiming a 10-year warranty. The not-so-fine print also said the warranty was void if the window wasn't opened for 30 minutes every day.

Perhaps somewhere in Canada there's a dutiful person who actually opens his house to the wind, rain and snow for 30 minutes every day, and  arranges for a neighbor to do it when he's on vacation. But most people who buy either Android phones or that brand of window ignore the warnings and take their chances. 

Building reliable products is hard. At Anybots we're in the middle of this process. We're developing mobile telepresence robots that have to drive around offices and factories. We know people will throw them in the trunk of their car, pack them in luggage, drop them, kick them, crash them, drive them off curbs, spill coffee on them, and 1000 other abuses we haven't even thought of. We're doing our best to make them survive as much abuse as possible, but they're physical objects made of plastic and metal containing optics, electronics, and bearings. They can't survive everything. They won't be falling-anvil-proof.

Actually, places with falling anvils are a great application for telepresence robots. If you have to look around somewhere dangerous, you'd much rather risk a $10,000 robot than yourself. So we'll definitely have customers who drive them into lava fields or unstable mining tunnels or fires or chemical spills. There are military-grade robots costing $100,000+ that would survive more of these situations than ours, but we think 2 of our robots will outlive 1 military-grade robot. That is, for every 2 incidents that would destroy our robot, there'll be at least 1 that would destroy any robot.

So, some customers will use our robot in air-conditioned offices and some will drive them into burning toxic waste. How do we provide a warranty that makes sense for everyone? Robots that fail in office settings didn't meet reasonable customer expectations and we should replace them for free. A robot hit by an anvil or blown up by an IED or melted by flames has already provided fair value to the unharmed customer, and they should pay for the replacement. At either extreme, I think reasonable people will agree on what's fair. It gets trickier in the middle. 

A typical middle scenario will be a user who drives it between buildings in light rain. The robot should survive that sort of thing most of the time, but it's outside our spec and it may eventually fail. Depending on who that user is and how his company budget works, he may offer to pay for repairs or may try to claim warranty work. Auto makers have this problem all the time. People must bring in cars with blown engines caused by racing or never putting oil in, and the dealer has to decide whether to replace it under warranty. It can be an unpleasant, confrontational process that leaves bad feelings all around. I want to avoid that.

The current plan is to offer 3 levels of service contract for different uses: office, dirty environments, and hazardous environments. They'll be priced according to our estimated repair/replacement cost. I want to err on the side of being nice to customers, but we may occasionally have to insist that people upgrade to the next contract if, say, they send us back melted robots for replacement under the "office" service.

I'm open to other ideas that meet our goals. To state our goals explicitly, they are:
  1. Make customers happy by exceeding their expectations.
  2. Collect honest data about why robots fail. Don't force customers to cover up the real reason for failure.
  3. Don't lose large amounts of money on outlier customers.