Search analytics: a window into how users actually search

Sometimes the best way to explain what something is involves first discussing what it is not, and that’s the tack that Otis Gospodnetic used to open his Day 2 presentation “Search Analytics: What? Why? How” at Lucene Revolution, Day 1.

Here are the slides for this session.

Search analytics” is not search engine optimization (SEO), nor is it exactly the same as Web analytics. Rather, It’s a complement to these functions; search analytics focuses on the mechanics of the search process itself. Given the fact that search is critical to the success of may Web sites, especially commercial ones geared at selling, search analytics can be a tool vital to an organization’s business success.

Performing such analysis requires instrumenting the search pipeline to enable direct observations of user behaviors including the forming of queries, resulting click-throughs, and subsequent user actions (or transactions, such as “Buy!”).

Search analytics are ultimately about search success and failure, with an eye to identifying underlying, but actionable, causes for outcomes and trends. In this context, failures include: zero or too few results, low click-through rates, high search exit (or bounce) rates, insufficiently relevant search results, and too much user refinement of search queries.

Otis continues by enumerating the kinds of reports required to deliver value, with some of the kinds of data including in each report. For example, the “zero hit queries” report might show that the root problem is misspelling in the query string, or a synonym table that’s too small or incomplete. Some fixes for these kinds of problems might include more extensive use of string auto-completion or “did you mean? (DYM)” options.

Cross-posted with Lucene Revolution Blog. Tony Barreca is a guest blogger.This is one of a series of presentation summaries from the conference.