I think you need to start with the problem you are trying to solve. The prototypical problem statement is a URL embedded in a seemingly trustworthy, but easily-compromisable source, such as unsolicited email. The real trick here is that users need training to better identify the risky scenario and to know the mitigation. Specifically, that URLs used to access important sites should come from sources you trust , rather than blindly clicking the link.
I can remember "www.discover.com", so there is no harm in me typing that one in, and when I type "mellon bank" into Google, pagerank and anti-fraud departments do a great job at ensuring that the first answer is trustworthy. Other viable options are to use browser history and as you suggest, bookmarks.
Another great trick is to use a password manager (lastpass is my personal choice) and to use random passwords for each site. If it does not auto-fill, the natural response is to say WTF, which will quickly expose a fraudulent URL. Plus it is "harder" to get tricked because one need to find the password and then copy/paste. In addition to being a PITA, it gives one more time to consider if there might be a reason it did not auto-fill.
Finally, don't get fooled by MFA. It is a great tool to combat a wide variety of attacks, but MFA is insufficient to combat URL fraud. Introducing MFA forces the bad-actor to work real-time, but it does not preclude them obtaining a complete set of valid credentials.
In direct answer to your request, I concur with @rslade that a case study recommending bookmarks is unlikely because it is far from the only way to skin this cat.
