And the award for sweeping generalizations goes to...
In other words, the best person to analyze a company's data is the person who understands the company's business, and not necessarily the person who understands "data science."
You don't think this might depend just a tiny bit on the data and applications in question? And in particular on the type of analysis desired?
It's not surprising there's a generic app to do generic sentiment analysis. I haven't looked at it, but it's probably the sort of simplistic naive-Bayesian bag-of-words thing that students regularly throw together for class projects. It's just barely possible that some people would want something a tad more sophisticated.
Take a glance at Vincent Granville's list of "worst predictive modeling techniques". When you understand why he's arguing against those - and under what conditions he might be wrong - then you might know enough about "data science" to be able to formulate a coherent argument about when you need a domain expert, and when you need a data scientist, and when you need both.
(Whatever that is.)
Hi, I'm Matt Assay, and I'm going to make bold claims about things even I admit I don't understand.