The 5 things we should all test using A/B testing techniques

On May 19 and 20, within the framework of the Practitioner Web Analytics 2009 and in one of his presentations, Avinash Kaushik He spoke to us about the importance of A/B Testing when it comes to increasing the performance of a website.

He A/B Testing consists of testing different versions of a web page or an area of the web page, and deciding which of these versions is the one that obtains the best results. Although we offer this service from GeaIPC, few companies decide to use it even when the results are really spectacular. Precisely, my presentation at the PWA09 It consisted of explaining a case in which the booking engine of one of the hotels in the AC Hotels chain doubled the number of reservations thanks to the changes implemented in said engine, testing it with A/B Testing techniques.

And it was also in the framework of PWA09 that Avinash commented on the 5 things that we should all test on our site:

1. All the “call to action” on our website. That is, all the buttons and texts that encourage you to take action. For example: the typical “Call us now and ….”, or the “Click here to contact", either "Subscribe now and get…”. We should try different colors, fonts, phrases, etc…
2. The steps from the home to the interior areas that interest usThat is, it is assumed that when conceptualizing our page we have determined different navigation paths… so we must test whether the page is designed in such a way that these paths are followed by our visitors.
3. The “pain pages” (Avinash called them that and I think it's a very apt name.) These are the pages that have a lot of traffic, but also a high bounce rate (people who go directly to them and leave the website without having browsed). You have to review them and test different versions of them to get the bounce rate down.
4. Test whether page performance improves with or without scrolling. This point is interesting. In the study of Usability with Eyetracking In our January article, we already indicated that very few of the under-25s we tested scrolled through the Google results pages (which were the ones we tested in that study). If our website has this type of user as its target audience, this is a good thing to test.
5. “Dramatically different” designs (It sounds better in English than in Spanish, but oh well). What Avinash meant was that we should try completely different designs than the ones we are using. Maybe we will discover one that also has “dramatically positive” results.

The good thing about A/B testing is that it is easy to perform and if we use Google Website Optimizer, it may not even cost us any money. Google offers the Website Optimizer for free, which allows us to test as many versions as we want in real time, and then tells us which of the tested versions has the best results. The image we can see next to these lines shows the Google Website Optimizer control panel.

• Learn more about Google Website Optimizer here: Google Website Optimizer
• Information about our A/B Testing service: A/B Testing at GeaIPC
• Avinash Kausik's Blog: Avinash Kausik
• Web Analytics Practitioner 09 Website: PWA09