Beware those derivative numbers

Editor and Publisher reports  'Time Spent' at Top Sites Still Declining." This is another case where numbers can fool you.The Nielsen Online data "tracks the average time spent per person at a site during October." As the story notes, unique users "soared.""Average time spent" is calculated by dividing the total online time by the number of unique visitors. When the uniques go up, the derivative "average" will drop unless the new users collectively match the behavior of the old/regular users.Often they don't. There were a lot of election-related stories in October that may have attracted new, casual users who may have had no natural interest in anything else on the website. They may have been referred by the Drudge Report, Daily Kos or some other specialty site. Or they may have come from search engines. If you "get lucky" with a story that scores huge out-of-market traffic, you're going to "get unlucky" with your averages. No way around it.Search-engine optimization is a change many sites made between 2007 and 2008. On a news site, SEO may drive up unique users by bringing in out-of-market visitors with specific interest in only one story. These folks will generate only one pageview. The result will be depressed time-on-site averages.If you're creating an incentive program for a site manager, you have to be very careful of issues like this. The time-on-site average is a useful metric in site management, but as always, it has to be considered in context with other measures.