A marketing strategy to improve the performance of a website. Within one experiment two versions of a website are randomly delivered to visitors. User behavior in terms of time of interaction, activity with call-to-action-elements and customer conversion into recurring users is measured. By using the results of the experiment certain parts of a website can be optimized. It's important to limit the versions on either a certain element (e.g. call-to-action button) or style or content. After finishing one experiment, a new one can be started on another topic.