A landing page has quite a number of elements which can be redesigned or given the go-by. Stuffing a site with a number of widgets might work out if the prime elements are identified with ease. Aiming to find which grouping of elements works the best is one of the defining functions of those tasked with designing it. There is no scope for trying out a layout in a random manner. This takes time and is economically unviable unless the design works out well early.
With A/B testing, two versions of a site are tested in real time with customers being equally directed or in different proportions to the two streams. At the end of the test, data on the most effective design is immediately available. The A and B versions will have different URLs with customers arriving at the A version being partially diverted to the B URL.
Implementing the redirection has to be done with care as the site might be reclassified by a search engine. The redirect has to be marked as a 302 redirect which means that it is a temporary diversion. Using a 301 redirect might lead to the page variant getting indexed by the search engine. Another important to-do is to close the text after the appropriate number of visitors has turned up and the required data has been collected. If the alternate versions stay live for a long while, it might result in a negative result by the search engine.
While conversions are the ultimate metric, other interesting metrics such as session duration can be optimized with an A/B test. Getting the best results from the available traffic is a resourceful approach when compared with looking to increase traffic. The number of variations which can be tried out can go up to ten. The higher the number of variations are, the longer the time taken for the test to complete. Try out different Calls to Action, photographs and headlines and find out which optimizes best for the desired response.