I was a researcher at a reasonably big UK University until about 6 months ago, and I used SciHub fairly regularly.
Sometimes I used it because we didn't have the right subscription for that publisher to get me the journal I needed, but mainly I used SciHub to access papers I already had the "right" to access. The layers of logins screens, buggy single-sign-on systems, having to be in the right IP address range, not being able to download a PDF, meant that I almost always attempted to get the paper through SciHub first, and only went through the library systems when I couldn't find what I needed on SciHub.
Sometimes I used it because we didn't have the right subscription for that publisher to get me the journal I needed, but mainly I used SciHub to access papers I already had the "right" to access. The layers of logins screens, buggy single-sign-on systems, having to be in the right IP address range, not being able to download a PDF, meant that I almost always attempted to get the paper through SciHub first, and only went through the library systems when I couldn't find what I needed on SciHub.