Google AdWords™ For Dummies®, 3rd Edition

Determining Whether Using Website Optimizer Is Worth the Effort

When someone asks us whether we can improve their website performance, we ask in return whether they've been testing page variations in a systematic and scientific manner. If their answer is No, ours is Yes. Smart testing improves performance. This attitude isn't boasting, or blind faith in our magical marketing abilities. Rather, it's based on probability theory and years of experience, plus a clever trick that Charlie Darwin discovered when he wasn't getting seasick on the Beagle.

Your website is the curtain you're sticking with

Begin with probability theory, specifically the infamous “Monty Hall problem.” This mathematical riddle asks you to imagine yourself on the game show Let's Make a Deal. Monty Hall, the host, shows you three curtains and tells you that two of them conceal piles of elephant poo, while the third hides a million dollars. You choose one of the curtains, and Monty then pulls it back (actually, the lovely Carol Merrill pulls it back) to reveal a steaming pile. Monty then asks whether you want to stick with your curtain or switch to the other curtain that hasn't been opened.

The mathematically correct yet intuitively wrong answer is to switch. Most people either say it makes no difference or that it's better to keep your first choice. But probability theory states that switching is twice as good as sticking — at least if your current outlook values money more than dung.

You can see the correctness ...