In the marketing world, A/B testing has been a staple for many years. As you probably know, the idea is to change a product slightly so you have two versions--A and B--and then determine which version consumers prefer. But a recent Wired article questioned the efficacy of this esoteric marketing tool. It argued that A/B testing is limited and that experimental design might be a better marketing method.
According to the Wired post, A/B testing works well when hundreds of different tests can be run at one time. But when the number of tests conducted is limited, the variance in the sample is not great enough and the statistical significance in the testing is meaningless. It added that in many cases it is also hard to identify which variables elicit a response from consumers--all problems experimental design might have a solution for.
Experimental design works best with companies that market to a large group of customers--listing credit card companies, online retailers and telecommunications firms as good examples.
So, what is experimental design?
According to the story, experimental design uses “mathematical formulas use combinations of variables as proxies for the complexity of all the original variables," then it allows for adjustment based on responses to different variables. So what it does is increase the variance in marketing campaigns so as businesses can determine the best marketing campaign based on numerous changes and various combinations and then adjust to these more quickly.
However, the article pointed out that experimental design must be accompanied by other changes in the company. For instance, companies must be able to reach the right groups of consumers for the testing, that employees are properly trained in the new method and that there is a process of decision making set up around the shift to experimental design.