The seasonality of search demonstrated
I spent some time on Saturday reviewing the past year of search queries performed at MSU.edu, the web site of my employer, Michigan State University. It proved a useful way to distract myself while my co-author's alma mater destroyed my Spartans in football.
For quite some time, I've known how seasons affect search needs among an academic institution's searchers. After some seat-of-the-pants manual clustering of queries by topic, the seasonality of search seems to be validated once more. I've color-coded the clusters in spreadsheet snippet below; you can download the entire spreadsheet here (942Kb Excel spreadsheet; each column represents a single week's worth of queries from each month, September 2005 through September 2006):
...and here's the legend to help you make sense of those colors...
When you scan through this data, you might pick up on the following trends (or perhaps some others, which we hope you'll share as comments below):
- No matter your site, many popular searches will be impermanent—people will search for some content no matter when, because that's what defines their introduction to your enterprise.
- But much of your searching will be seasonal. People will seek certain items based on time of year.
So we can predict, for a university:
- At the beginning of the school year, students seek bookstore sites. At that time and prior, they will seek housing.
- In the middle of the semester, students seek sites that let them complete assignments, take tests, etc.
- At the end of a semester, students seek sites to plan the next semester, or to look up grades, or to see if they are on academic probation.
In January of 2006, we see very few academic searches. Why? School started the next week.
Bottom line is that a student portal could adapt its top 10 or so links to match what students seek at that time of year. The seasonality depicted here is probably extreme for most schools, but it may apply to many other contexts, such as fashion, cars, travel...