Quick Tips

• 7-Step Process for Effective Web Analytics
• 6 Do's and Don'ts for a Successful Web Analytic Implementation
• How to Select a Web Analytic Tool
• Analyst Skills for Effective Web Analytics
• Web Site
  Pre-production
• Google Analytics Multivariate Testing Website Optimizer
• Personas: Designing for People
• Know Your Customer: Learn Their Words
• Bridge the marketing results gap
• Integrate Online and Offline Tactics
• Help Your Customers and Search Engine Crawlers

Updates & Help with Google Analytics

Home  >  Quick Tips  >

Google Analytics' Multivariate Testing Website Optimizer

Want a to get started with A/B split-testing?
Looking to find out what multivariate testing is all about without a sizeable budget?

Try out Google Website Optimizer.  Sign up for a Google Adwords account.  Google Website Optimizer is free to all Google Adwords customers.  All Adwords customers will see a box highlighting the availability of Google Website Optimizer upon logging into Adwords. 

A multivariate testing platform, Google Optimizer allows you to test landing pages or conversion events for not only search engine marketing, but also email, ads or offline drive-to-web campaigns. Sure, this will help Google increase Adwords revenue but it's a definite win-win for Adwords customers as well.

However, there's nothing more dangerous than an invalid test. Because people tend to have more confidence in test results, they tend to risk more after a test is completed. Taking action on invalid test results can be deadly.  So do make sure you keep the following "safe testing" tips in mind:

1. Ensure your sample sizes are adequate for a quantitative test. The appropriate sample size is a function of the confidence you require for the test, the level of difference you are trying to detect and the base number you are assessing.  Estimating the sample size required for your test will ensure that you can estimate the duration of your test.  If you require more assistance in this, contact us.
    
2. Monitor external events during your test.  Outside influences can skew your test.  Ensure you have adequate resources to monitor for external influences during the duration of your test.
    
3. Ensure that your sample is representative of your total population, especially if you are only exposing a fraction of your total audience to the test.

Google Website Optimizer will declare the winning combination for you, based on statistical calculations.

If your site already uses another Web traffic tracking tool, you may still be able to use Google's Website Optimizer. Just make sure that you talk to your information technology department before you start making any plans. And if you do run a test in this multi-tool situation, don't try to reconcile the Web traffic numbers between Google Analytics and your existing tool. I guarantee the numbers will be different. Create a baseline. Compare your baseline to your test results. That's what you're after, right?  Experimental results that help you improve!