Q: How can I conduct A/B testing or run experiments to improve the performance of my website?
A: A/B testing is a great way to learn how searchers interact with content on a website. It allows you to see what real estate on a website has the most value and where the searcher’s eye is attracted. One way to test a website is to run a paid search campaign using an analytics platform to measure results. For best practices, make sure you have an end goal established before you begin. Once this goal is established, you will want to create two distinct versions of the web page to be tested. You will want to change locations of graphics, types of text, and even colors to see which style brings in better results.
For example, if your website sold widgets and their was a horizontal banner image across the top and a call to action on the bottom of the page, another strategy could be to have a vertical image down the left side of the page with the call to action on the top right. Depending if the vertical strategy had a higher conversion rate, you could then test by switching them and having the call to action on the left, and the image on the right, etc.
Most search engines offer A/B testing within their paid search interface (like AdWords). You can easily enter both landing pages (the two pages that are being tested against each other) into the engine and see how they each perform.
Typically, you should run a test for 30-90 days in order to collect enough data to be sufficient and actionable. Once this data is collected, you can then use the test results to choose which web style is more efficient and converts at an accelerated level.