Part of the idea of Google Analytics is to take the guesswork and hunches out of website traffic analysis. Instead of your feeling about what visitors will do, you have the actual paths and click numbers. But data in a vacuum isn’t particularly helpful. If your response to all this information is to build new site pages and features based on hunches, then the analysis isn’t as useful as it might be.
The point of the Website Optimizer is to try to determine the benefit of one set of pages as opposed to another objectively rather than by hunch, thus better matching the intelligence provided by Google Analytics.
To run Website Optimizer, you need to be prepared to sort through the data provided by multiple A-B tests of specific pages with specific features to see which performs best. In other words, Website Optimizer provides a testbed for comparing sequences of pairs of pages. In a well-designed test, each pair deviates from one another in only one way. Once you’ve determined the best choice, then you can move on to the next deviation. Obviously, it takes plenty of visitor data to make Website Optimizer statistically significant. Google Website Optimizer advises that for pages that receive less than 1,000 page views, A/B tests will be the most effective for testing variations.
Website Optimizer can be started from a number of places within Google Analytics: from the Reporting tab, from the Opportunities tab within AdWords, or by direct navigation. See Figures 13-24 and 13-25.
