Web Analytics

In my role in working with technology in the e-commerce business, many people want to view analytics usage of people viewing the web store, where people exit, how long they visit, etc.  This is nothing new and it seems that more and more people are moving toward Google Analytics.  In the past, a very nice albeit expensive solution was Web Sense.

The way that these external analytics services work is that bits of web (HTML) code are embedded onto the web page that report back to your analytics provider with information about how you got to a certain page (from a link, keyed into the address bar directly, etc.).  So what happens is if you view a web site with an outsourced analytics service, you're also sending your request information to the analytics service.  This has been going on for years and it's nothing special or surprising.

What we've been seeing for a while on the business security systems but increasingly on consumer security systems (both higher grade routers as well as software firewalls) is blocking of sending the information to analytics services.  Earlier today I was at a workshop discussing network security and over half of the businesses represented are already or beginning to block the data being sent to analytic services.  While this small sampling may not be entirely representative of the world, it does reflect the trend.

Now what this means to those who are using external anayltic services, such as Google Analytics, Web Sense, et al. is that the accuracy of the numbers in the reports is waning.  I believe that as time goes on and the operating systems of computers and security services proliferate, less and less data will be provided to the analytics services.

So, for those of us who wish to be able to review our data and turn that into useful metrics and graphs, we'll have to retool how we analyze our data.

Virtually all real web servers log the exact same information that the analytics services provide, but few people actually review those logs.  There are commercial and open-source software that can live on the same server or network as the web server and can capture all that data and turn it into something useful.  Another possibilility is that we may start to see 'log shipping' taking place where the web server logs are shipped or accessed by the external anaytics services.

In the mean time, if you, as someone hoping to garner useful data from external analytics services, you need to start thinking about an alternative immediately.  Already there is less and less web usage information going over and that's going to continue to decrease in the not-too-distant future.