But what fraction of people are actually going to take their device in to get a fix or a refund? In the EU your right to a refund for a malfunctioning product only lasts 2 years, so only the fraction of users who notice a problem before that time limit would even have the opportunity to bring it in. Then there is a decent chunk of the population that doesn't want to go through the effort of turning it in (which also incentivizes overly burdensome return workflows). Then there might be some people who notice the problem and would be willing to go through the effort of returning it, but mistakenly believe that the problem was their fault instead of a defect (incentivizing arcane error codes and unintuitive maintenance procedures). The real refund percentage could easily be low enough that the increase in sales from bad design practices is still worth it. And of course that's assuming the company selling it is a single rational entity - if the product development head gets rewarded based on short term sales and won't suffer the consequences of returns months or years down the road, then even a high refund rate won't necessarily stop bad behavior.