Oh yeah, I am familiar with this trend. It all depends on what the question is and whether the methods you’re using will give you the answer to that question.
Sometimes you do just need to measure and optimize certain interactions, but a lot of time, that won’t give you what you need. I’ve seen people make some really weird leaps of logic based on analytics. There are a lot of offline influence that won’t show up. Measuring specific interactions won’t tell you why people are doing one thing or whether there are whole audiences you aren’t reaching, for example. You won’t learn about anything you didn’t think to ask, or don’t have the means to measure. Why limit the scope of your potential learning like that?
I’ve been hired to come in and help companies who went down an analytics rat hole and ended up with a lot of data, but still lacked understanding.