At Facebook, this is pretty standard practice. We test many different interesting ideas that come out of hackathons or team brainstorms, like the ability to save posts for later, useful if you’re in a hurry and see a post that you want to respond to or an article you’d like to read later. However, we only wanted to launch the feature if enough users used it and found it valuable. If not, then it wasn’t worth taking up space as yet another action link on every story. A test helped us verify that it was, in fact, something only a small group of people used, so we decided ultimately to not launch it.
Nice story about how Facebook takes into account the extra cognitive cost of having a feature added in the UI, after deploying, and then using that decide if they want to launch a feature or not.