Wednesday, February 24, 2010

Testing web analytics implementation with WASP

I stumbled on this thread on the Web Analytics discussion forum: "testing of webanalytics"

Q) Can you please suggest best practices around testing webanalytics implementation (Omniture or Webtrends). Any test plan or strategy document along with the test tools that can be used will be of great help.
A) There is no better test than randomly clicking your site to verify page names.

He? Excuse me! When has quality assurance become a random act of faith?

What should I test to increase quality of my web analytics data?

First, you should already have quality assurance test scenarios for your whole site. You can use the same ones to test your analytics implementation.

Here's how to proceed:
  • The home page is pretty unique (as are landing pages built for specific campaigns), so those should be tested individually.
  • Most sites uses templates. Identify each template and at least 3 pages using each of those templates. For example, category pages, product pages, article pages, etc.
  • Identify each process-driven conversion scenarios: shopping cart, subscribe to newsletter, contact us, internal search, etc. - each of those scenarios should be tested with at least 3 set of values. You will want to test extreme conditions, like entering wrong data or higher/lower range values. For example, when testing internal search, do you correctly track zero-search results and their search terms?
  • Identify how campaigns are going to be tracked and what are the parameters being used to track them
  • Especially with Omniture and WebTrends (and now Google Analytics), you can have several custom variables. You will want to identify them and test pages that are specifically assigned those values.

How do I test?

Don't do this by hand! The poor man's way would be do to a "View source", search for the tags, use some debugger and try to decipher the query string parameters. Chances you miss something are big and you will loose an incredible amount of time doing it.

WASP, the Web Analytics Solution Profiler was built specifically to ease quality assurance of web analytics implementations. When used manually, it will show and easy to read detailed breakdown of your tags in the  browser sidebar as you browse from page to page. There's a free version of WASP you can try, and if you plan on doing more serious quality audits on a frequent basis, the more advanced Market Research version will allow you to crawl your site or list a bunch of URLs in a text file (i.e. those using templates defined above!)

Disclaimer: I'm the creator of WASP - which was sold to iPerceptions a couple of months ago.

What should I watch for?

WASP for Omniture tags quality assurance

Check the following:
  • Are the tags firing at all? Does the WASP sidebar shows values?
    WASP shows data as it is being sent. If not, either the tag isn't there at all or there are JavaScript errors preventing them from firing.
  • Check for campaign parameters
  • Are the tag values the expected ones (check title and all custom variables)
  • Especially for processes, are the tags the right ones?
  • Are there multiple calls? If so, this might be normal, but you might want to double check you don't have duplicate tags or unexpected calls.

More information

You can get more information about WASP on the official site at WebAnalyticsSolutionProfiler.com as well as the following blog posts:
Post a Comment