You have a spiffy website with engaging content organized in an easy-to-find manner. How do you know what is and isn’t working on your website? Where are your visitors coming from?
Analytics. Otherwise known as web statistics. There are three types of basic information for a website: page views, visits, and unique visitors. Page views are the number of web pages viewed on your website. Visits are the number times people visit the website. They can view multiple pages (page views) each visit. Unique visitors are the number of people who visit the website (usually as near as possible to ascertain).
There are a lot of packages out there: Omniture, Web Trends, Unica, Coremetrics, VisiStat, Google to name a few. The type of information they offer is wide ranging and so are the prices. How they gather stats can also be quite different, but there are two general camps: log-based, or tag-based.
Log-based packages:
Log-based packages take all of the logs from your web servers and crawl through them periodically to count up all the stats from your website. These logs usually contain all the traffic that comes to your website including search engine crawlers and bots. Search engine crawlers and bots are computers which visit and index your site, usually for purposes of adding you to the listings on their own website. This can be good for getting your website out in front of potential customers but it doesn’t tell you what the people who visit your site are doing. Most packages get around this by having a list of bots and crawlers to ignore in the stats – keeping this up to date can be a bit of work. Due to having to crawl through logs for stats, these stats packages usually cannot report ‘live’ data. They are usually at least a few hours behind.
Tag-based packages:
Tag-based packages work by placing a bit of code on all the pages on your site. When a visitor visits your site, this code activates and updates a count on your stats server. Most tag-based software is coded in such a way that automatically ignores all search engine and bot traffic. They may also ignore real traffic from people who have security software or a firewall that blocks the code – but this is usually a very small portion of your traffic. Many software providers have ways around this limitation by using tiny images instead. Tag-based packages can often deliver real-time stats.
Which to choose?
I’ve been using both types of packages for stats at the publishing company: Unica Web Analytics – log based and Google Analytics – tag based. My preference has been for the Google stats because we have a lot of difficulty keeping the search engine crawler and bots filter list up to date. Google analytics is easy to install and use and it is hosted so they keep the software up to date. The disadvantage to Google is that it is very difficult to get your stats out of it if you want to use them in another application for data mining purposes (we’re actually looking an Urchin which was bought by Google for this purpose). There are also some packages which use a combination of both methods. These can provide the advantages of both and negate most of the disadvantages, although they tend to be more expensive.
Technorati Tags: analytics, web analytics, web statistics, B2B, internet consulting, B2B internet consulting