Hmm, I disagree with the idea of collecting RUM data to assist in reaching the end goal of making a website faster.
Firstly, anyone with a website can assume that their users have a variety of network/processing capabilities. That’s a given.
So what more is RUM data going to tell me? And I mean real actionable information — something that will drive a site to be faster that wouldn’t have happened if I didn’t have RUM data.
The idea of “understanding your users’ experience” with the goal of improving performance via collecting numbers seems a bit theoretical and detached from reality for my tastes.
It’s all going to come down to HTTP/2, preloading, lazy loading, compressing, etc. I can’t think of a single improvement to site speed that I’ve made that would have been any different if I had RUM data. I’d be interested to hear of any instances where it has made a difference.
Capturing usage metrics for purposes other than improving performance is a whole other kettle of fish.
