Bugs (including, but not limited to Yosemite) and some significant UX issues. One marketing problem I have is that although it's better designed than the competition, in practice it's not quite as intuitive as it looks. I characterized a "revolt", but in reality there are two camps of users:

* People who come from traditional GIS, who love it (except for the bugs). These are the revolters.

* People who come from MapPoint, who think they're going to love it but find some things confusing and give up. These people were always discontented with the software at some level.

The frustration from the latter group stems a few unintuitive workflows in the software (e.g. how data gets imported and linked), some of which are due to design flaws in the data model IMHO. So although the software looks good in screenshots, and many people get a lot of value out of it, Magic Maps needs some work beyond basic bug-fixing to be a "5-star app" that pleases everyone.