Websites and Analytics
The statement “There are three kinds of lies: lies, damned lies, and statistics.” was popularized by Mark Twain at the turn of the last century. Some things never change… I spend some time every day going over analytics for several dozen websites. Over time you get used to the weekly and seasonal ebbs and flows of the traffic. Most every site has its own rhythm; some are slower on the weekends and others busier. Some during the day and others during the evening. Some overseas and some domestic.
It is interesting to watch over time as improvements occur. An increase in search rankings, an improvement in bounce rate or pages viewed or repeat visitors. Changes in search traffic, since that is primarily driven by how many people search for related terms, tend to be gradual especially with higher volume sites; although, you may get a short term spike to a related news event. Referrers can radically change the amount of traffic they drive since your placement is driven by other factors. My favorite is Stumbleupon.com. You can get thousands of visitors per day for a week or two if you happen to be in their selected groups but then have that traffic disappear just a quickly as they rotate others into those spots.
One of the most powerful features of Google Analytics is the advanced segments tool. With it, you can select (or deselect) traffic that meets a particular criteria and then analyze that particular segment. One site I have has a couple of pages on jobs that went up years ago but attracts a fairly large number of daily visitors. Most of these visitors don’t help this local business, so we separate those visitors out for a different review allowing us to focus on just the visitors they are targeting. Another usage is to separate out a group coming in on a particular search term and review and contrast their experience on the site with others. Sometimes it becomes painfully clear where your errors lie.
Without segments you could make some mistakes. For example, bounce rate is a very important metric, but in most places in analytics it is summarized at a page or source type level. You may find a page with a high bounce rate and then decide to make drastic changes to it. The root cause may have been that the page draws a high volume of traffic from one search term that isn’t really related and that term experiences almost a 100% bounce rate, but the correct terms could have a very low bounce rate. Just using the averaged number creates a bad result.
Another place to pay attention is geography. If your site attracts worldwide viewers but your target audience is solely domestic, then you should segment that group out. Overseas visitors often do not have the same types of bandwidth, browsers and screen resolutions that domestic visitors do. If you are using your stats to help drive your web design (and why wouldn’t you???), it could very like alter your requirements.
What can you change?
I’ve always thought one of the easiest things to improve on an individual landing page is the bounce rate for a given term. It is easy to determine this value and often easy to understand why you have a higher than average bounce rate for the term once you review the context of the page. Then, just test a few different modifications that you think would better connect with the users of that term and keep the best result.
Google Analytics is free, simple to use, powerful and constantly improving. There is no reason not to have it on your site. Knowing what you visitors are doing on your site and how they got there is a critical piece of information for running a successful website.