Receive new blog posts immediately direct to your email inbox!
The ROI Revolution Blog
« Google Analytics Report Enhancer Updates | Main | Website Optimizer Wednesdays - Optimize Your Call to Action! »
Website Optimizer Wednesdays - Excluding Internal Traffic
October 22, 2008
Google Website Optimizer provides a handy preview tool that allows to check how your experiment pages will look before you start your test and while your test is running. Now there are some cases where the preview tool does not work.
While there is a manual way to preview tests even when the preview tool fails, there's nothing like being able to just go to the test and page and see that it's working, maybe even clearing your cookies a few times to see all the different variations.
To make sure that visits like these don't muddle up the results of your experiment, you should exclude yourself from the experiment. Here's how:
Google Website Optimizer experiments use 4 different types of scripts. In a nutshell:
- Control script - Determines which combo to serve up
- Section scripts - Determines which areas to swap out
- Tracking script - Registers visit
- Conversion script - Registers conversion
To exclude ourselves from the reports, we only need to modify the tracking script and the conversion script. By only modifying these sections, we can see the page exactly as our visitors see it without skewing our test results.
Here's the way your tracking script looks "out of the box":
If we're already using Google Analytics, we'll want to properly integrate our Google Website Optimizer code with our Google Analytics Tracking Code. Check out Shawn's excellent post on the subject if you want to know exactly how to do this. For simplicity's sake, we'll use the above code as our base.
Now let's say we are excluding our own traffic by setting the user defined value to "internal". When we do this, the user defined value is stored under a cookie name "__utmv".
Using regular expressions, we can check for the existence of this cookie and its value and only run the Google Website Optimizer tracking script when the cookie exists with the correct value:
Note that our conversion script should be the same as above, only replace "/test", with "/goal".
Now if you have a static ip address, you can also exclude based on that ip address, but this will take some server side code. We'll use php as an example:
This would exclude all traffic from Google Website Optimizer tests coming from the IP address 12.34.567.890.
As a footnote, these same ideas that we explored above can be used to do even more with our Google Website Optimizer experiments. We could, for example, modify the scripts so that only AdWords traffic shows up in our Google Website Optimizer reports. For a heavily AdWords-driven business, this would help tailor experiments to more closely match the most often used Google Analytics reports.
To learn more useful testing tips sign up for our Google Website Optimizer Training Series starting January 8th. This two session training series will encompass landing page principles, an intro to testing and advanced testing. Join us for the GWO Training Series!
Want more of Website Optimizer Wednesdays? Check out the rest of the series!
Exclude Internal Traffic from GWO | Optimize Your Call to Action | Landing Page Relevance | Choosing the Right Test Page | GWO and GA Renew Their Vows
Posted by Jeremy Aube, Analytics and Optimization Specialist at 9:20 AM
Permalink | Comments ( 2 ) | TrackBacks ( 0 )
Filed under: Google Website Optimizer
TrackBack
TrackBack URL for this entry:
/mt/mt-tb.cgi/327.
Comments
Good post, though I'm wondering what's the point of tracking the dummy visit instead of just doing nothing?
Also, readers should keep in mind that it might not really matter much in terms of the actual results. For details see:
http://www.websiteoptimizerguide.com/How-Do-I-Exclude-Certain-IPs-from-the-Experiment_8.html
Posted by: Ophir at November 4, 2008 10:42 AM
@Ophir: Tracking the dummy visit isn't essential, but depending on what types of tools you are using to evaluate your Google Website Optimizer implementations, it may be handy to see some sort of a hit register.
As far as whether or not excluding internal traffic from Google Website Optimizer experiments matters, again, it depends on what you're doing. If you're checking out the Google Website Optimizer experiment and want to see several different variation live without using the preview tool, you may end up clearing your cookies several times and register several "conversionless" visits in the process.
Or you might be testing to see whether conversions are registering correctly, in which case you are artificially increasing the conversion rate for a particular variation. Of course, you can't do this kind of testing without seeing the dummy hits to the conversion page.
For a site with a lot of traffic and a decent conversion rate, it probably matters a lot less, but methods like these can help out those who are testing with lower volume traffic, especially if they are bringing in a third party to help with the testing process and setup.
