Best Practices for Landing Pages Testing

[stupidpoeticjustice]

Hello,

I am interested in deploying a concurrent landing page test for one of my major paid search clients. Essentially, they have three distinct landing pages that they would like to test. I would like to be able to concurrently test the landing pages in order to eliminate any time-based confounding variables.

I have thought that the best way to do this would be to send users from the engines to a page which, based on some sort of random code, serves up one of three landing pages to test. Has anyone here utilized a similar method? If so, what was your experience? Also, any resources on how to implement such a test? How did you handle tracking?

Lastly, how sound is this method? What would you consider to be the best practices for implementing a landing page test?

At this point, I have many questions and few answers. Any help would be greatly appreciated!

[AussieWebmaster]

What you can do in Google is create three exact same ads with the three different landing pages as the respective destination urls... do this with tightly grouped keywords or better yet adgroups of only one keyword in each.

Turn off the optimzation option so the ads rotate equally through impressions... you will need to weight slightly for the differences in CTR that may occurr but you track the Conversion rates tied to the tracking and you will discover which landing pages has the best conversion for what terms - I would think there is the chance that some keywords will convert better for one lkanding page versus another if they are done decently.

[stupidpoeticjustice]

Thanks for the reply AussieWebmaster. Unfortunately, I currently use DART (doubleclick) to track all my paid campaigns. Accordingly, each keyword has a unique URL. As a result, that solution would not work, at least utilizing my current tracking system.

[AussieWebmaster]

Thanks for the reply AussieWebmaster. Unfortunately, I currently use DART (doubleclick) to track all my paid campaigns. Accordingly, each keyword has a unique URL. As a result, that solution would not work, at least utilizing my current tracking system.
Okay then you have to set them up in separate adgroups with individual ads and possibly use exact, phrase and broad matches with a healthy use of negatives.

[tomslick]

What are the specific goals of your landing page testing? Are you trying to reduce drop-offs? Increasing click-thrus? Looking for better conversion? You might be able to use data from your web analytics reporting to analyze user behavior on the various pages depending on how complex your test is.

Here's an example:

Send all of your traffic to the same destination page but do the randomization that Aussie suggested. This way, all keywords and match types are distributed equally across the three versions. You can append parameters (or parse them) from DART to tag users and what versions they see.

Next, use your web analytics or in-house tracking tools to analyze results. Many web analytics tools allow you to parse parameters for reporting purposes. Or, you client could parse the parameters through their app and write this data to a log or database.

[AussieWebmaster]

You can also use the tracking solution offered by Google or Overture (though Google offers the free one!)