Beware the Outliers.
If you have relatively low traffic to your website, a few outliers can greatly skew the numbers. Let’s say you want to understand whether users are spending significant time on a given page - an indication of how engaging it is. If 10 users look at a page for 10 seconds, but one accesses the page and then goes to a half-hour meeting, leaving it on the screen, the average time shown will be almost three minutes. By excluding the 30-minute outlier, you’ll see that most people are actually coming to the page and leaving quickly, so the page is apparently not satisfying your community. To exclude it from your calculations, just export the data (to a spreadsheet, for example), delete the 30 minutes reference from the list, and calculate the average of the remaining figures.
You’re Measuring Computers, Not People.
Measurement software looks at a computer, not a person. While you can for sake of simplicity assume that one “user” is one individual, you’ll have situations where one computer is used by multiple individuals (such as at a school or library) or one person uses more than one computer (such as home, work, laptop, Internet cafe). You can determine what fits your community and adjust your measures accordingly. If you know you’re attracting a lot of students who use both their own laptops and a school computer, you can estimate those numbers and decrease your uniques for a more true measure. By contrast, if you believe your traffic is mostly coming from local schools’ computer labs, you can presume that each “unique” in your analytics software represents more than one individual.
Watch Trends.
Remember, too, that one day does not a data set make. While you can start getting useful information almost immediately, you’ll need at least three months, preferably six months or a year, before you can start really identifying trends and traffic flows. It’s very useful to compare year-on-year trends. Real data geeks correlate for differences in holidays and weekends when comparing months or quarters.