The use of A/B analysis is becoming popular for websites as they look for the best way to interact with their customers. It basically provides different versions of the website to people and the way that the customers interact with the site and if they make a purchase is tracked. The data is then reviewed and the information can be used to choose a final design or to make further changes.
It is relatively accurate in finding the way that customers respond to changes on the website and which achieves the better response. The number of times that various changes are made can start to impact the site negatively.
Positives That A/B Analysis Provide
Allows step improvement validation
If you are administering or developing a large website then you can conduct tests using A/B analysis on specific parts of the site. This makes it much simpler to implement changes and measure them using a step by step process.
It is also great to answer specific development questions. An example would be ‘do customers and visitors prefer larger buttons or smaller buttons?’ The data about how groups that used the web page or pages can then indicate which is preferred. This can then be implemented site wide with a high degree of certainty that it will have a positive impact.
One of the reasons that this form of testing is popular is because it gives clear and easily understood data on which version of the page or site is preferred. There are three possible results site A is better, site B is better or both are the same. It is also from data of real users of the website giving it real validity.
Negatives That A/B Analysis Do Not Show
Does not give qualitative data
A major drawback of this sort of analysis is that it gives no information on the qualitative aspects of the website. This means that it does not help in detecting problems of usability it just measures how many people completed a task or transaction in two competing designs.
This important because both designs might be poor and negatively viewed by site visitors but one has a higher rate of them persevering to the end. This needs to be kept in mind so that other tests or analysis is used to compliment this type of testing.
Higher work load
This type of testing can result in a much longer and more expensive development process. The need to produce two versions of each change that is proposed can lead to a huge amount of extra work and care needs to be taken that it is only used when it can be valuable.
There is a danger that it can be overused and become a way to try and check on every change as a site is being developed. This is not what A/B analysis is designed for and it should only be used to answer specific questions during the design process.