Mark Zuckerberg describes how Facebook’s Newsfeed evolved from user behavior around profiles: people were clicking many profiles to see day-to-day updates, so the team first made a product showing which friends updated their profiles, then a simple Newsfeed that listed posted content on the homepage. He emphasizes a product-development loop: use data and qualitative community feedback to identify problems, apply intuition to propose solutions, and then treat those solutions as hypotheses to test via rollouts and further feedback, enabling incremental learning and growth.
Chapters
You want to actually evolve in a way where you're working with your community and making steps and learning. So take newsfeed for example. There was a relatively big shift, but it actually had been a couple of years in the making by watching how people were using the service, right? So when we started off, we didn't have anything like Newsfeed that showed you updates from what people were sharing. We just had profiles. And what we found were originally one of the big behaviors was people would just click around, right? They'd click on different profiles, hundreds of them, and they'd go through all their friends to see what people had changed, to see what the update was in their friend's day. And we learned from that that people were not just interested in looking up and learning about a person, but also understanding the day to day changes. So first we made this product that just showed in order which of your friends had updated their profile so that at least told you whose profile to click on. And. And then the first version of Newsfeed was really simple. All it did was it basically took the content that people were posting and put it in order on your homepage. So I think when things are working well, you use data and you use the qualitative feedback that you're getting from listening to how your community is using your product to tell you what problems to go solve. And then you basically use intuition to figure out what solutions to those problems might be. And then you test those hypotheses by rolling them out and getting more data and feedback on that. And then that gives you a sense of where to go. But I think when stuff is working well, you're learning incrementally and growing that way.