It's possible for tech companies to collect more data on users than ever before: from phones, smart watches, browsing habits and connected homes. The question, however, is should they?

There are two real motivations for companies to collect user data, to better understand their customers and so improve the service, or to better monetise their users by serving targeted ads or just selling on the data.

Understandably many users don't like companies to collect their data just because they can. Privacy is an important, but increasingly elusive, concept in this hyper-connected age. The onus is on the company to set users' expectations about what data they'll be collecting and why. If it's a free product that's supported by ads, or a paid product where data is used to improve the product, that's fine. But if you position it as data only being used internally, then by selling on data you've created a ticking PR timebomb.

Real user data is a crucial input when thinking about how to improve your product. Real users discover those edge cases you haven't thought of, or end up using your product in novel and unexpected ways, and if you don't collect data on them you'll never know. The data by itself isn't a magic silver bullet, it can only be used to answer questions, so you need a strong product vision and a set of theses you want to test. User perception of what data it is reasonable to collect depends on the product in question, if they can see a clear link, and if they trust you not to misuse their data.

Uber have been in the news again recently for collecting more user data. The change in data collection is accompanied by a perfectly reasonable explanation of how they'll use it for product improvement. The problem is that users don't always trust these explanations and, in this case, the link isn't clear enough for many. The lesson here is that you shouldn't just collect user data because you can, but only when you have a valid need to that users will understand and accept.