Just wanted to point out a curious observation from my blog's search stats...

By default, the stats.cfm page shows the 10 most searched terms for a blog. I modified mine to show 20 by default and all results greater than 2 hits if logged in as admin.

In addition, I added a most recent searches section just below that where I write out all searches for that day, or all searches for the past two days if logged in as admin.

I noticed an interesting trend when I started watching search stats by time. I would see search terms show in the list and I'd check my logs and they look like 'real people' making real searches, but then sometimes I'd notice that a sequence of recent searches showed up as a repeated sequence but in reverse.

Say I checked in the morning and I saw these recent searches:

view plain print about
1alpha
2beta
3gamma
4delta
5epsilon


Then if I checked later I would sometimes see something like

view plain print about
1epsilon
2delta
3gamma
4beta
5alpha
6foo
7bar
8baz
9alpha
10beta
11gamma
12delta
13epsilon


What I think is happening is that spiders are crawling the stats page and since these search terms are hyperlinked for convenience, the spiders just follow the links from top down, and your search term frequency has just been skewed. The spider fingerprint is a mirrored search term sequence.

To prevent this, I only hyperlink the search terms if logged in as admin, otherwise don't add links. I expect this will provide a more accurate picture of what folks are searching on without the spider artifacts.