What is so great about testing your website ?

Testing is arguably the quickest, cheapest and most guaranteed way of improving the performance of any web page/website.

  • Testing is quick, as you can easily test a multitude of different ideas.
  • Testing is cheap, as it involves hardly any development cost and you can literally have a new test running within minutes.
  • Testing is (nearly always) guaranteed to improve conversions. Even if, against all odds, you are not able increase the performance you can rest assured in the knowledge that your web page/site is a green, mean conversion machine (even though it might be blue and actually quite friendly).

How do you do the actual testing of the pages ?

We like to use Google Website Optimizer (GWO) for the testing of webpages. Although there are other tools out there, GWO is free to use and easy to set up.

Depending on the traffic your page/site receives, you’ll get the data back within a few hours - and it makes for quite addictive viewing.

Below is a screenshot from Google Website Optimiser. The table shows the likely high performance pages after the first live day of running tests on a set of landing pages we designed and set up for testing on behalf of a client on 'car insurance'.

Siteta increases conversion by 130% - 210%

The data should be fairly self explanatory: the top row ‘Original’ are the results returned from visitors being shown the original page (I said it was self explanatory). The following three rows are the three variation pages we designed to help the client increase their conversion. BTW we didn’t ‘cherry-pick’ this table.

Testing of web pages: A/B vs Multivariate Testing

A/B testing is generally used to test one page against another page (or pages). Usually the page will be different in several ways; one could have the form on the left hand side and a big image, the other could have the form on the right handside and no image.

Multivariate testing is generally used to test multiple elements on a page; there could be variations to the headline (e.g. ‘Buy your iPhone here’ vs ‘Special offer! Today only’). At the same time you could test two different colours on the button and maybe see if an additional picture of a student or a professional person will perform better. Multivariate tests usually need to run for longer due to the (normally) higher number of page variations available.

Sidebar