I was listening to Andreas Mattsson presenting how Spotify's uses machine learning to recommend music. A fascinating topic, but what struck me was how interesting and challenging and rewarding must have been for Andreas and the team to work on this problem. Successful companies are made successful in large part by the people who work there. In many cases, employees-led initiatives result in the biggest breakthrough for a company. Just let people follow their curiosity and see magic happen.
Sergei Brin pointed to the same magic when he told the story about how Google's Brain team became a fundamental part of the company: "Jeff Dean would periodically come up to me and say, 'Look, the computer made a picture of a cat,' and I said 'OK that's very nice, Jeff' … and fast forward a few years and now Brain probably touches every single one of our main projects."
I think of this as the fun-to-work-on effect. Similar to other types of network effects, it is a self-reinforcing loop that makes companies and teams stronger. A small group of people starts working on something. That something grows and begins touching more and more interesting problems. The volume of data available increases and even more problems can now be solved and explored. People start joining your team because they want to solve interesting problems. Once they are there, their curiosity leads them to uncover even more possibilities, and so on.
The biggest part of any company's success comes from the drive, curiosity and ingenuity of the people working there. Create a company where it is not only fun to work IN, but where there are enough problems that are fun to work ON. The rest will probably take care of itself.