A recent article in Reference & User Services Quarterly caught my eye today:
Finnell, Josh and Walt Fontane. “Reference Question Data Mining: A Systematic Approach to Library Outreach.” Reference & User Services Quarterly 49.3 (2010): 278-286. Web.
The authors describe a process developed at the library at McNeese State University in which reference statistics recorded at the desk included not just question type but also the subject of the question and the course (if any) connected to the question. Questions were later mapped to LC classification numbers, which then helped library staff to make collection development decisions. The data also led the librarians to refine their instructional offerings and to make new outreach efforts to specific departments.