Top 15 To-Do List for Multivariate Testing with Google Website Optimizer

Google Website Optimizer tips from SiPA’s 25th Annual Mid-Year Marketing Conference: Gain. Market. Share.

While Amanda was in Montreal this week at a Google Analytics workshop, Don and I were in Las Vegas at SiPA’s 25th Annual Mid-Year Marketing Conference: Gain. Market. Share. One of the sessions we enjoyed most was on using Google Website Optimizer, a topic Amanda also covered called “Google Website Optimizer 101“.

Take these lessons below from Kiplinger’s Greg Krehbiel, BLR’s Bob Brady and Josh Baker and watch your conversion rates soar.

• Have a techie in marketing – somebody who understands the Google technology and how your site works.
• Have your techie go through Google’s AdWords training program.
• Test big things first (page design, how many columns, long vs. short) before you test little things (color of buttons, size of text).
• Test high volume pages and apply those lessons to low-volume pages.
• Limit your tests to pages that get traffic from a common source – e.g., a Google AdWords campaign.
• Keep your tests simple.
• Test “distracting” links, like “free sample issue”.
• Consider the ROI of the entire process, not just a single page (e.g. one page might perform better, but Google might give it a bad rating and raise your cost per click (CPC) or it might improve one part of the funnel but hurt another).
• Keep records of your tests and take screenshots of everything relating to the test that you can including control and test panels, stats panels, analytics data, etc. in case you need to reference it at a later date.
• Share what you learn with the email marketers and web designers
• Use your web analytics data to uncover hidden or not so top-of-mind high impact pages or user paths to test. You may find that a high percentage of visitors travel certain paths and bounce or exit before reaching your desired destinations for them. These make great areas to test to improve.
• Choosing your test goal will help you plan what elements to test. A goal of increasing time spent on a page versus a goal of increasing clicks on a call-to-action may introduce different elements that you may want to test.
• Look for and document alternate learnings from your test results. Although you have your main test goal, you may acquire additional information or benefits along the way which may help in future ideas or elements to test.
• Know which metrics you will be using to declare a winner by and make sure you know how you will get to those numbers.
• Be on the look-out for major issues that might effect the validity of your test results such as:
  • Testing a page that receives traffic from various channels, by only using one traffic source for the test traffic and then making changes based on this data.
  • Outside variables affecting a test variable (i.e. a national news story breaks thus influencing a certain test panel over another that relates to the story).
  • Slow loading pages, testing platform problems, data collection problems, etc. that will effect how your results are measured.
  • Not having a large enough sample size to receive statistically significant results.

More from this event: 12 Online Publishing Tips from SiPA’s Marketing Conference

[text_ad]