The contents of the log file enable site search analytics: the entries provide the evidence needed to deduce how your users are searching and how well the site search is helping them. Cherish the logs or at least keep an archive: you may need to go back someday.
[7] The NCSA combined/extended log format is documented at http://publib.boulder.ibm.com/tividd/td/ITWSA/ITWSA_info45/en_US/HTML/guide/c-logs.html#combined and http://httpd.apache.org/docs/2.2/mod/mod_log_config.html#examples
Summary
SSA offers a unique treasure trove of data worth tapping because it’s the one place where users tell you in their own words what they want from your site.
SSA provides different information about users than insights you normally get from SEM (Search Engine Marketing) and SEO (Search Engine Optimization). Think of people searching the Web as people you want to attract to your site, while people searching your site are customers you want to retain. SSA is concerned with the latter.
Query data can be captured in search engine logs or by analytics applications that harvest information on users’ actions on your site.
When users search your site, they typically will have more specific needs (and queries) than when they search for information on the Web.
As the Zipf distribution shows, a little SSA goes a long way. Start by improving the performance of your site’s most common queries; they will account for a huge portion of your search activity.
Use pattern analysis, session analysis, failure analysis, and audience analysis to analyze your query data, diagnose problems, and determine ways to improve your site’s content, navigation, and search system.
Use goal-based analysis to determine new ways to measure your site’s performance and connect search to your organization’s KPI (Key Performance Indicators).
Chapter 3. Pattern Analysis
Getting Started with Pattern Analysis
Finding Patterns in the Long Tail
Anti-Pattern Analysis: Surprises and Outliers
The next five chapters (including this one) cover various ways to analyze and derive new insights from your query data that can directly improve your site’s user experience. We’ll start by looking at the patterns and oddities that emerge from your data if you play with it (and stare at it long enough).
Analysis as a Form of Play
In pattern analysis, we look for what our queries have in common: tone, length, topic, type, and more. We also explore what’s odd—are there queries that don’t fit with the rest? We then study those groups and misfits to see if we can learn something new about our searchers and the content they want and need. Does their language match the tone of our content? Are they requesting certain types of content more than others? Do searchers demonstrate certain kinds of information needs at particular times of the year? Or day?
How does pattern analysis work? We simply play with our data and see what emerges. Yes, it’s essentially that simple: we play. And it’s as fun as it sounds.
The Simple Part
The simple part about pattern analysis is that as humans we’re naturally built to detect patterns, especially semantic ones. Don’t believe me? Well, take a quick look at this list of words in Table 3-1.
Table 3-1.
| Google Common Queries | |
|---|---|
| #1 Indiana earthquake | #11 earthquakes today |
| #2 isabelle caro photos | #12 laura govan |
| #3 candace cameron bure | #13 moshe katsav |
| #4 lily shang | #14 indystar |
| #5 amazon eve | #15 happy new years |
| #6 isabelle caro before anorexia | #16 new year quotes funny |
| #7 new years eve 2011 | #17 brie iarson |
| #8 billy taylor | #18 Christine O donnell |
| #9 jamie foxx | #19 billy boyd |
| #10 2011 predictions | #20 feliz ano nuevo |
| A set of common queries logged by Google Trends, December 30, 2010. |
What did you notice? Any patterns emerge? Any outliers?
You may have noticed many queries that were people’s names (for example, laura govan and billy taylor), while others were related to the end of the year (new year quotes funny and 2011 predictions), and the rest were an odd mix of stuff pertaining to mostly earthquakes.
Or you might have divided the people into different categories (for example, politicians, musicians,