I like that you can get a holistic view of a conversation online. Typically to uncover valuable insights you will have to run multiple keyword sets, removing things that are skewing the conversation, but the final product can be quite insightful if you know how to use it.
You can't get a full picture without supplementing it with other tools. This is primarily because it doesn't automatically pull in all of the Facebook, LinkedIn, or content from all sources at once.
I also don't like that often times the most used terms, hashtags, and things are obvious. I'm pulling a report on Healthcare analytics & Healthcare data and the majority of the top terms, hashtags, and things are healthcare, #health, etc.
Also it seems like sentiment is a worthwhile metric, like demographics, but in my experience they typically make up such a small percentage of the total mentions or posts that it isn't valuable. Showing an end client the sentiment and then following it up with only 5% of all posts had sentiment period kind of kills that section completely.
We're trying to find hot topics of conversation to create content for, join existing conversations, and draw insights and recommendations from the data as a whole.