Using Web Analytics To Create An Optimization Plan
This post was made Mar 27, 2009 by Carlos del Rio
On Tuesday the Google Analytics Blog tried to tackle using web analytics to create an optimization plan. I don’t think they really address the type of optimization that each example really needs.
Top Landing Page Analysis
One of their examples of performance indicators that indicate an opportunity is Top Landing Page. They point out that a high bounce rate on a top landing page is a loss of business. Yes, but you need to figure out why these people don’t engage your content. If you are optimizing a page that has high entrance value and high bounce rate the first thing that you should check are your referral stats. Social Media is notorious for serving low quality leads, look for Facebook, Twitter, StumbleUpon, etc. in your referrals. The other possibility is a slightly off search result. When you achieve ranking for a phrase that you don’t really fulfill you can get and influx less engaged users. Often this is the result of a nebulous concept, something that can be taken in multiple ways, like the word modern or bass.
If you are getting a lot of social media traffic you have two choices: add a special message for the refer or look for a link opportunity to capitalize on the reason the the social media portal is referring to you. Either way this is an occurrence that is largely outside of your control (you can steer it) but, for the most part, it is self-correcting or short lived.
If you are getting mis-targeted search traffic you should be either clarifying your content or providing a link to more appropriate content to the alternative definition. This may increase your exit percentage from the page, but it will also increase your search indicators by not bouncing back to the search engine, and if you support the alternative it can be an opportunity to drive a visitor deeper into your site.
Conversion Path Analysis
Google’s second example of a red-flag is a frequent exit point from the conversion path. In the specific example that is given in the Google post the exit point is the final step of a three step process. The likely culprit is a loss of trust. Any test should start with possible miscommunication on the page or elements that may introduce friction to completing the conversion.
The example they give is e-commerce 95% of people Login/Register –> 96% Fill Out Payment Information –> 61% Confirm. Clearly the people are well qualified at conversion start, but something breaks drastically. That usually indicates a loss of trust. The solution for this problem could simply be removing the third step– and replacing it with a user prompt.
High Exits
There are two types of exits: satisfied exits and unsatisfied exits. A satisfied exit is the result of a visitor finding what they want and leaving, an unsatisfied exit means they gave up. If we stick with the e-commerce example here are three reasons you may have a high exit rate from a popular item: low stock, back-stock, high price. The author of the Google post posits that your copy may not be persuasive, that is true, but it could be that your total offer isn’t persuasive– not in their size or not a competitive price. If you are in this situation check to see where your competitors are on price and stock of the offending item.
Pair Your Metrics
In each of the examples you can use a paired metric to get some more insight into your problem and choose a test that is in the right neighborhood of you potential problem– e.g. Bounces and Referrer, Exits and Time on Page, or Exits and New vs. Return Customers. Taking a little broader view on the problem will let you test more efficiently.
