Measuring your changes using Google Website Optimizer

Watch the screencast (14:00)

Lots of things can impact your site sales. The trick is to work out what they are, and which changes increase your sales. The problem is that the changes that seem obvious often aren't.

At a conference I spoke at I showed three versions of the same page to a room full of ecommerce industry folk - site owners, vendors, etc. I asked them to vote on which version of the page produced the most sales. About 80% of them chose the same version. I then showed them the data from a test that showed that the version the majority chose had the lowest sales of all three, 50% lower than the top selling one! I'd like to be all smug at this point and say I chose the best performing page, but my choice lined up with the 80% I'm afraid. The point is that outside of some well established best practices, we generally have little idea what impact a change will make.

Google have a tool that's free to Google Adwords customers called "Google Website Optimizer". It allows you to setup experiments on your website and test which ones deliver your goal, such as:

  • sales
  • click throughs
  • newsletter subscriptions
  • download of trial software
  • completing an enquiry form.

For example, you might want to test some copy changes to a landing page. You could make three versions of the copy - the original and two variants. You then set up the optimizer to test which of the three variants generates the most sales. You can test almost anything. Some ideas include:

  • copy
  • headlines
  • image size
  • image composition
  • layout
  • special offer (eg "30 day money back guarantee")
  • "Add to cart" button design/copy
  • anything else!

I've made a screencast which shows you step by step the process of setting up a test. It's not that hard, although some knowledge of how your site's HTML works does come in handy. It does allow you to send the instructions to your techie if that's how you prefer to work.

Give it a shot, I guarantee that you'll get some surprises, as well as some increased sales.

Watch the screencast (14:00)

Last updated on 6/27/2007