by Aaron Aders
Analytics in business is unfortunately something that is not typically a standard process for the average small to medium sized company. Applying analytical segmentation to date can solve business problems and provide answers that directly effect your bottom line. No place is easier to implement analytics in your business processes than on your website.
The Internet is the most trackable form of media in history. A wealth of data about user traffic is provided by even the most rudimentary hosting companies. However, gathering raw data alone cannot provide answers. Segmenting the data in clever ways is what provides answers. Data segmentation answers questions as simple as “what is the most popular page on my website” and as complicated as “what is my top converting product during the workday lunch hour in each time zone?”. Here are a couple must-have web applications that will allow you to just get your feet wet or take a dive down the rabbit hole into web analytics.
Must-Have # 1: Google Analytics (FREE)
Every website owner should be using the website tracking system that provides analytical segmentation. My personal favorite is Google Analytics if your website receives less than 500.000 unique visits per month(1). The Google Analytics interface is extremely user friendly and comes pre-packaged with many useful data segments. The savvy internet marketer can use regular expressions with filters to segment the data in nearly any way imaginable. It is also possible to create tasks that automatically generate and email analytics reports on a monthly basis. Avinash Kaushik is a Google Analytics wizard who writes books and blogs about segmenting data and finding answers through data segmentation website. Visit his website or purchase his book to get an education by an expert in web analytics.
Must-Have # 2: Google Website Optimizer (FREE)
So now you know which page is most popular on your website, thats great. But what if you had a different tagline? Would that page would be even more popular? Would users purchase more frequently from my version of the tagline rather than my colleague’s? (common battles here!). For answers to questions like these you must use testing software. This is where the Google Website Optimizer comes into play. What once was an argument is now a lunch bet, because your users will tell you which tagline converts better.
The Google Website Optimizer displays different versions of a particular section of your website in equal numbers according to an A/B or Multivariate test. This may be a graphical ad, the tagline, a paragraph of content, an application, the shopping cart experience, etc. This data is then compiled to determine the best performing version, the % chance of improvement, etc. We should have had this done yesterday, but wait! There are a few necessary ingredients you need to run successful experiments.
1 – A large sample set of visits OR a lot of time. If your site receives 1,000 + visits to your experiment per day you will get statistically significant answers in a relatively short amount of time. Less traffic means the longer experiment (a low traffic website may take a month or longer to generate statistically significant answers). Also a more complicated experiment (multivariate) means much more time or traffic.
2 – An experiment plan. This could be an A / B test comparing different versions of one module (eg which version of our tagline H1 converts the best) or could be testing multiple versions of multiple modules in a multivariate test (eg which combination of H1 tagline and graphical the ad works best).
3 – Dedication. Optimization does not end after one test. Testing is ongoing if you have not reached 100% conversion rate (which is impossible, of course).
I hope this peek into the world of web analytics has been helpful. For more information visit the Google Analytics support center.
(1) In cases involving our clients with 500,000 + unique visits per month Google Analytics seems to cut off anywhere from 25-75% of individual user tracking data after the half-million unique visits mark. If you MUST have perfect data for every user you can purchase very expensive web analytics packages from a company like Omniture which will charge different pricing tiers for more or less traffic. However, the half million fully tracked visits typically provide a large enough sample set to trust most segmentations.