Google Website Optimizer

Google Website Optimizer – Improving Conversion Through Testing

For many of us, once we have a website or we are using Pay per Click such as Google AdWords we only tend to think about making changes to things when we have some new information or offers that we want to publish.  The fact is however that it is rare to get everything right first time form a conversion point of view i.e. how many of your web page visitors or those who click on your AdWords adverts actually go on to buy something, download something, fill in your contact form, call you etc. How well your web pages and landing pages ‘convert’ is vital to your business and can be improved.

There are popular and relatively easy and ‘scientific’ ways to make, measure, and monitor these improvements.  If you have a Google Account and / or Google AdWords Account you could benefit from using the Google Website Optimizer.

What Is Google Website Optimizer?

This is free platform that allows you to optimize (improve) the conversion properties of web pages.  This is done by enabling you to run experiments which show different versions of your web pages to visitors, or different variations of elements within web pages, and then measures which are most successful in producing conversions (as defined by you).  These experiments are for conversion optimisation rather than search engine optimisation and don’t appear to search engine rankings or produce duplicate content problems for the pages involved.

Test Options

The Google Website Optimizer gives you 2 basic text options – A/B Split Testing or Multivariate Experiments.

A/B Split Testing

This involves creating different versions of a page, and rotating the page for every visitor. The success of each page in generating clicks to a specified conversion page is measured, and therefore it is possible to see which page design and layout is going to produce better results, and thus that page can be adopted and further refined over time.

Multivariate Experiments

This involves a similar process of rotation of different versions for visitors and measuring the results, but instead of using different whole pages, different elements of the same page are created and varied.  For example, different visitors could see (and react differently to) different headlines/ headings, images, paragraphs of text or ‘submit’ buttons.

For both A/B Split and Multivariate experiments it’s worth remembering that:

• You may want to conduct experiments on high traffic volume pages and landing pages, or pages which you believe are under-performing.  High amounts of traffic are likely to produce more obvious and more reliable data and information more quickly.
• Very small variations are likely to produce very little noticeable difference in conversion, so make sure that your variations really are ‘different’ in a considered way.
• Code needs to be inserted into pages for Google Website Optimizer to work, and for A/B Split Testing you are likely to need to build / have built for you some different versions of pages.
• It is possible to import the ‘Goals’ and ‘Transactions’ that you set in your Google Analytics account for conversion into your Google AdWords account to use as Conversion Actions thus saving you time.

How Do I Set It Up?

It can be set up with AdWords or a standalone site, and you can specify on set up whether you want to link to a Google AdWords or Analytics account.  Signing into AdWords for example, click on ‘Get Started’ go to ‘Reporting’ and select ‘Website Optimizer’.  Go to the ‘Experiments Links’ page, then ‘Create New Experiment’ and then select the type of experiment e.g. A/B Split Testing.  Google helps to guide you through the process using checklists and help along the way.  You will need to give your experiments names so you can easily distinguish between them.

For A/B Split Testing you will need to provide the URLs for the pages involved (original, text pages and conversion pages).  For Multivariate experiments you will need to name and create the different elements of the individual page that you want to test.  The more variations there are, the more the combinations increase for testing. For both types of experiments script tags need to be added to the pages – the original, variations, and conversion pages for A/B Split Testing, or to the individual page elements for Multivariate experiments.  The Google Optimizer does provide a guide for this process.

With this system you can preview the experiments, and you can phase out any particularly poorly performing pages or page elements as you go so as to narrow things down and get the best results.  Google Optimizer allows you limit the % of participants in the experiments (although it’s better to include all) and you can view the results in the Reports section of an ‘Experiment Dashboard’.