A/B Testing with Google Website Optimizer

Tuesday, May 19th, 2009
0 Flares Filament.io 0 Flares ×

Have you ever wondered how a design change to a webpage might affect conversion? If you are a serious online retailer, the answer is likely yes. Often times even small page modifications like simply changing the color of a button can have an immediate impact on your conversion rate. One of the greatest things about having a website is that almost everything can be measured scientifically. Although detailed measurement is possible, it’s not surprising that most people still take a fairly subjective approach to their websites. With the help of Google Website Optimizer, there really is no longer an excuse for “guessing” that certain changes will improve the key metrics of your site.

Google Website Optimizer is a free tool found within Google AdWords. Google Website Optimizer makes it very easy to run an A/B test or even a multi-variant test. In the case of running a simple A/B test, you would simply create two different versions of the same page with just one variation. Google then asks you to put two snippets of JavaScript code on the original or control page (one snippet at the top and one at the bottom). You also need to put a snippet of code at the bottom of the new version of the page and a different snippet at the bottom of the conversion page (this is the page that someone visits when your site has successfully achieved its objective). Once those pages are successfully loaded on your site, you tell Google Website Optimizer to start running the experiment.

Google Website Optimizer is very clever. You do not need to change any navigation components on your site. Once the experiment is running, all of your visitors will still visit the original control page. However, half of them are going to be directed to the new version of the page (the top JavaScript snippet enables that redirect). The other half will stay on the control page.

Through the web-based Google Website Optimizer interface, you can decide how much traffic is going to be used for the experiment. Naturally, if you use 100% of your traffic, you’ll be able to determine sooner which option is working better. Google also gives you the ability pause the campaign at any time.

Once the experiment has begun, be sure to regularly check on the results of the reporting. If the reporting tells you that Version B of your test is the high confidence winner, congratulations….your intuition that a change was necessary was correct. If version A prevails, stop the experiment and rethink your approach. Making even slight changes to your site can have a positive impact and Google Website Optimizer gives you the means to prove it.

After receiving a degree in Industrial Engineering from Cal Poly San Luis Obispo, Alex began his career as a technology consultant with Accenture. In 1999 he left Accenture and founded Nexternal – a cloud-based eCommerce Platform company. Alex is passionate about eCommerce and online marketing. He is now a Senior Vice President of HighJump, managing the Nexternal business unit.