I agree it’s good data, but good data isn’t particularly useful if you don’t know how to interpret it. It seems to largely answer questions I don’t have, and finding relevant answers is a bit harder since the data is focused on datacenter use.

So I personally look for support quality first (very imprecise, but I look for anecdotes about good and bad customer experience) and avoid the capacities that seem to have consistently high failure rates and low average age in Backblaze data (e.g. 10TB drives). In the past, they largely used consumer drives (not even NAS drives), and now they largely use enterprise drives, neither of which I’m planning to buy anyway, so the main commonality between drives I’ll consider and drives they monitor are the platters, hence the focus on capacity.

I’m glad they publish it, I just think people misinterpret it more often than not.