Here it is: The first web analytics expert system for troubleshooting a drop in web site conversions

A number of folks and I had recently lamented the lack of an expert system for guiding web analysts to ask the right questions in the right order. Our web analytics tools can answer almost any question. But practitioners had been left to their own devices when it came to figuring out what series of questions to ask.

Click here to start the wizard Click here for a bird’s eye view summary flow chart

So, each practitioner has had to reinvent “analysis” on their own.

Lots of books and helpful consultants are available to educate practitioners on how to do that. But that doesn’t make it less silly that thousands and thousands of analysts have had to reinvent the wheel every time.

So, it is with great excitement that I get to publish the (apparently) first “Web Analytics Question Support System” (QSS) today.

What is covered in this QSS today?

This first release is only very limited in scope.

It only addresses the situation where there has been a drop in the number of conversions on the site. You, as the web analyst, have been tasked with identifying the cause of the drop in order to remedy the situation.

The multi-step wizard will guide you through investigating whether the problem is in various details of your acquisition tactics (PPC, direct traffic, etc.). If the problem isn’t acquisition related the wizard will also guide you to investigate your conversion process and check on external effects such as competitors or seasonality.

What is not covered today?

If your web site is doing the same as it was during the last period and your question therefore is what you can do to improve things further, then you won’t find guidance in this wizard today. That will be my next project.

How does the QSS work?

The wizard walks you through a series of questions in logical order. Based on your answer to each question, you branch to a series of next questions and eventually to suggested next steps.

In order to determine your answer to each question you will be querying your web analytics solution, whichever you may be using. Down the road, I would like for such a QSS to be directly integrated into Unica’s web analytics and interactive marketing solutions. Some of the answers could then be automatically retrieved by the solution to save time.

But much more important than automating the wizard today is to validate and improve the content with your feedback. Therefore, please treat this as an RFC release, i.e with kind request for your comments and feedback.

What else will you need besides web analytics?

It is common knowledge among web analysts that they have to juggle a series of analytics tools in order to do their jobs. Besides web analytics, these include competitive intelligence or audience measurement, performance management, voice of customer, etc., etc.

But while crafting this wizard I was flabbergasted just how absolutely essential it is for analysts to use these tools in conjunction with their web analytics product. A logical analysis of the situation very quickly reaches points where the data coming from web analytics / page tags is simply insufficient for making informed decisions.

For example, I’d like to take my hat off to Bill Gassman from Gartner who has been pointing out for years the importance of integrating your site’s response performance data into your analysis.

Copy and improve this content!

Since I work at Unica, the company holds copyrights to the content that is published in this wizard. But my colleagues and I at Unica encourage you to copy, improve and reuse the content to your heart’s desire as long as you clearly refer to the original source, i.e. this blog and Unica. All of us in the industry stand to gain. Thank you.

Credits and additional comments

• Great minds think alike, they say. I am indebted to James Dutton at insight for the chance to discuss various options for shaping a QSS. Tradeoffs that we discussed ranged from the kind of strict wizard that I created to a searchable database of Q&As that James thought of. Both approaches have their merits.
• The good folks of Wripple had replied to my original blog post indicating that they are working on a very similar idea. But they are taking it one step further and making integrations with web analytics possible  Can’t wait to check them out as they go live!
• I am also indebted to Dylan Lewis for a quick review over the weekend and constructive recommendations for improving usability.
• As I said before, this QSS won’t eliminate the need for having well trained, well read, and well supported web analysts. Just like a clinical DSS doesn’t replace the need for doctors. Besides, analysis is just one of many things that a web analyst does.