Aussie Web Analyst

One of the tasks that can be performed using web analytics is to evaluate the effectiveness of a particular page on a website. The ideal method of doing so would be for every single viewer of the page to tell you if it was useful or not. For a variety of reasons, you are not likely to be provided with this data. Instead alternatives methods are used which indicate whether visitors found the page useful.

I have thought of an alternative view of a report which I believe would be a useful tool in evaluating the effectiveness of a page. This reporting view does not exist in Sitestat or Google Analytics and I suspect it probably does not exist within other web analytics packages either (please let me know if I am wrong). On the assumption it does not, can I please request that someone designs this reporting view for me.

Quickly first, some current methods of approximating the effectiveness of a page are:

• rating of the page by a sample of visitors (there can be issues with sample size and potential bias?)
• actions taken on a page (but what if there was no way to interact with the page?)
• time on page (only valid if didn’t exit from page but even then, was the visitor actually viewing the page the whole time?)
• actions taken by visitor during rest of visit or within specified time period (did viewing that page actually influence future actions?)

Another method of judging the effectiveness of a page is to base it on what the visitor did next. This can be done using a website navigation report (Navigation Summary in GA or Clickpath Explorer in Sitestat) or a Site Overlay. However it is currently only really practical and useful within a multi-page process e.g. the 3rd page of a 5 step checkout process. At this point, you know where the visitor should have come from and what they should be doing next - therefore can see if visitors are behaving as you expect.

For most other pages on the website, what is happening looks something like this diagram. There are a range of pages viewed immediately before and after the current page with one option in each category being ‘Entered Site’ or ‘Exited Site’. What can’t be easily seen is which of the previous pages link up to which of the next pages. And it is this kind of detail which could be useful in evaluating the performance of a webpage.

Imagine that your (small) site had been visited 9 times with the following being the clickpaths on your site.

1. A -> B -> X
2. A -> B -> X -> C
3. A -> B -> X -> B
4. A -> B -> X -> A
5. A -> B -> X -> X
6. A -> C -> X -> C
7. A -> C -> X -> B
8. A -> X -> A
9. X

Using the current reports for the page X, you would be able to see the following breakdowns for the previous and next page viewed:

Previous pages

• B - 50%
• C - 20%
• Entry - 20%
• A - 10%
• X - 10%

Next pages

• Exit -30%
• A - 20%
• B - 20%
• C - 20%
• X - 10%

This report is nice but doesn’t really tell you that much. Visitors came from a variety of pages and left to a variety of pages. However, if the next pages could be grouped, an alternative view of the ‘Next pages’ could be shown as follows:

Next Pages

• Exit - 30%
• Previous page - 30%
• Previously viewed page - 10%
• Current page - 10%
• New page - 20%
• Specified page - 0% (no page specified)
• Internal search results - 0%

This report now shows that the visitor went on the view a new page only 20% of the time, information that was not readily available in the current reports. The visitor returned to their previous page 30% of the time, possibly they are clicking on the back button due to unappealing content or because there is no clear indication of where to navigate to next on that page. For expected breakdown would vary for different types of web pages but this report could be useful in evaluating the performance of pretty much any page on a website.

Of course, for any report to be really useful, it needs to be segmented. Besides the usual segmentation options (traffic source, etc), I can imagine this report would be improved by first of all eliminating bounces visits for that page and then segmenting on whether the page was the landing page or not. Also, assuming there was an option for specifying a page or set of pages viewed next, this could be useful in understanding if the visitor clicked through to either target pages or the homepage from the page being evaluated.

While it is hard to definitely say without the report being in front of me, I believe a report of this nature would be useful in evaluating the performance of a webpage. Most importantly, I believe the report would help in making business decisions that lead to a company achieving its business objectives. And given that is (or at least should be) the requirement for a web analytics report existing, I think this would be a useful addition to the reporting suite for any web analytics vendor.

Tags: Performance Reporting, Reports, Web Analytics, Webpage