In my inaugural You can't analyze data you don't have here, I argued in favor of capturing analytic data in general, and text data in particular. Mining transactional data is already good, but mining or otherwise exploiting non-transactional data can be a hugely valuable addition. It's very early, but to some extent this is already being done, most typically in the area of formal surveys. A couple of examples may be found at the link above, which is from my specialty blog Text Technologies.