Website Optimization

Economic efficiency – in this case the website conversion ratio – is derived from the proportion of visitors that perform a desired action (aka conversion) to the total number of users that visit the website. Most often, conversion is the purchase of products.

Example: A website has 10,000 visitors during a certain time period. About 150 of them buy a product, so the number of conversions is 150. The conversion ratio is therefore 1.5%. The goal of optimizing a website is to increase the number of conversions. In this example, increasing the conversion ratio to 3% would mean the number of orders would rise to 300 in the period.

A website should be optimized if it fulfills the following criteria:

• Fulfillment of the website’s goals is quantifiable – typically, this means raising the number of orders (or their value), or it can also be increasing the number of advertisements shown or growth in the number of registered users.
• The website has a high level of traffic – when a website with tens of thousands of daily visitors sees a fraction of a percentage point increase in the conversion ratio, it can mean a significant increase in revenue.
• Optimization must be viable – the benefit from the changes made must be higher than the cost of carrying them out.

We use several techniques for optimizing a website, including testing the website on users, a survey using a questionnaire or A/B testing (or multi-variable testing). Tools for evaluating the behavior of users on the Internet called web analytics are then used to propose changes and measure results. Some typical tools include Google Analytics, Urchin or Omniture SiteCatalyst.

A/B Testing

We use A/B testing to evaluate a website during normal operations using real users. The subjects view two different versions of the website to see which one fulfills the goal best. The versions can have different texts, functions, different graphic elements or even a completely different design. Some users see one version and others see another.

We set quantifiable criteria at the beginning in order to evaluate the success of the websites. After beginning the tests, we monitor how the different versions fulfill the goals. In the ideal case, testing lasts until statistics confirm which one works best. Then the test can conclude and the most effective version becomes the regular presentation.

How is it technically possible for different users to see various versions of a single website and then evaluate the results? We have experience using two tools:

• Website Optimizer is free tool by Google which can easily be used for A/B testing on almost any website. Usually, we take care of the experiment from start to finish. We design, prepare, launch and monitor the experiment and in the end evaluate the results. When requested, we are able to help with a selected part of the experiment and we also can prepare training on how to use Website Optimizer.
• Our content management system jNetPublish supports displaying various versions of a website to different users and the results can then be evaluated with the help of Google Analytics. In this way we are able to perform more complex tests.

Besides A/B testing, we can also use multi-variable testing, which compares more than two versions, but the principle is the same. Kromě A/B

We don’t just use A/B testing to evaluate websites. The same kind of testing can be done with newsletters or advertising, such as in PPC campaigns.