They rewrote it from scratch as a sluggish web app, failed to address issues that have been annoyances since day 1, piled on dubious new features that just made it harder to open the damn thing up and write down the idea that just hit you, or to open up the note you're looking for at this very instant.
Evernote had multiple chances to get back on track. Its decision to rewrite its multiple apps into a single web-based one was one of them. In theory having only one codebase to fix bugs and build new features in makes a ton of sense.
Unfortunately the hard product work of simplifying the feature set and prioritizing server-side tech debt was overlooked. Therefore the new app was weighed down from the beginning because it had to support all these legacy systems.
IMO the best move for Evernote is to rethink the UX and tech architecture from first principals and have existing users opt-in to migrate their data over. It's a tough decision to make but I don't see any meaningful progress being made otherwise.