Comments on: How to Use Metrics to Improve Your Site – Part 2 http://blog.kissmetrics.com/how-to-use-metrics-to-improve-your-site-part-2/ Keep it simple... Thu, 09 Sep 2010 19:26:55 +0000 hourly 1 http://wordpress.org/?v=3.0 By: Silvia http://blog.kissmetrics.com/how-to-use-metrics-to-improve-your-site-part-2/#comment-1122 Silvia Tue, 29 Sep 2009 11:49:08 +0000 http://blog.kissmetrics.com/?p=248#comment-1122 Hey, Engineer, I was still reading the post when I thought of A/B split testing and now I saw your comment :) You are right: unfortunately split testing is only effective when you have enough traffic, or, for small businesses, if you can afford to wait until you get at least a 100 visitors before you run the test. As for Cost per Conversion, the best option is setting up a Sales funnel report. Check out www.logaholic.com to see how you can get the most out of web analytics. Hey, Engineer, I was still reading the post when I thought of A/B split testing and now I saw your comment :) You are right: unfortunately split testing is only effective when you have enough traffic, or, for small businesses, if you can afford to wait until you get at least a 100 visitors before you run the test.

As for Cost per Conversion, the best option is setting up a Sales funnel report.

Check out http://www.logaholic.com to see how you can get the most out of web analytics.

]]> By: tip http://blog.kissmetrics.com/how-to-use-metrics-to-improve-your-site-part-2/#comment-1106 tip Tue, 08 Sep 2009 01:56:30 +0000 http://blog.kissmetrics.com/?p=248#comment-1106 Hello, What a beautiful and awesome site. I adore what you've done with your setup and graphics. Thanks you so much. Hello, What a beautiful and awesome site. I adore what you’ve done with your setup and graphics. Thanks you so much.

]]> By: Engineer http://blog.kissmetrics.com/how-to-use-metrics-to-improve-your-site-part-2/#comment-504 Engineer Fri, 17 Apr 2009 09:52:45 +0000 http://blog.kissmetrics.com/?p=248#comment-504 <> The logic is sound, but the timescale, for me at least, is inconvenient. In order to have reliable results enough people have to view the altered page so that statistical variation is smoothed out. If you are getting a ton of traffic, no problem, but if not, every change has to be in place for days or weeks before it is safe for you to draw conclusions. An A/B split is a better idea, where you run variants of your page simultaneously and divide the incoming visitors between them equally. <>

The logic is sound, but the timescale, for me at least, is inconvenient. In order to have reliable results enough people have to view the altered page so that statistical variation is smoothed out. If you are getting a ton of traffic, no problem, but if not, every change has to be in place for days or weeks before it is safe for you to draw conclusions. An A/B split is a better idea, where you run variants of your page simultaneously and divide the incoming visitors between them equally.

]]>