Data Mining Fun at AOL’s Expense

AOL’s little search data leak is producing all sorts of interesting stories. This article from Lee Gomes sheds a bit of light on some of the more interesting aspects of the not-so-private data. Here are some of the more enlightening tidbits:

• The word “free” is the most common term found in the 17 million+ searches. It will obviously remain quite challenging to convert all these free seekers into paying customers for things like content, for example.
• In 47% of the searches conducted, users didn’t click on any of the results provided. Not a single click. Yikes. When you consider this statistic, it would seem that the search industry has nowhere to go but up, at least regard the quality of search results.
• Related to the previous item, 28% of all searches were refinements of earlier searches. So more than one in four searches were the frustrated results of not finding what someone was looking for the first time. That sounds about right. I’d say I wind up rephrasing my searches about a quarter of the time.
• In 42% of the searches the user clicked on the very first item in the search results. Yes, it pays to be #1 in the results, but how many people are misled into thinking their appearance further down the first page is a success? My guess is the click-thru rates drop precipitously as you go from #1 to #2 and so forth. In fact, I’d be willing to bet that anything below slot #3 is fighting for single-digit percentage click-thru results at best.
• How well do the search engines answer queries structured as a question? According to Gomes, 35% of those questioners never clicked on an answer. No wonder Ask.com changed their philosophy.