product operations require constant attention to the user’s movements, and need to seize all possible opportunities to actively find users to communicate. We understand them, and then change the product and operation strategy, but in the process we will find the user "duplicity".
the above paragraph is small team partners often chat with us, indeed, although the user research is very important, but after the investigation we need to go through layers of screening after another round of testing, which can really restore to the essence of our research, we call the "A /B test". Before writing this article, Mr. head special search wiki, but Wikipedia introduction is not professional, but in reality is not such a test.
first, the user is always "duplicity" behavior, we need to understand the real nature of
users through the test
users do not care about the product market, design, research and development, operation and so on when using the product. They are concerned about the product experience is not enough, the product service is not really meet their needs, this is the key. Therefore in the user interview, when we tried to know when the function of the product design and operation strategy which is insufficient from the user’s mouth, we often hear is an answer, when you put the user needs improvement, the actual operation of the user is another answer. So you feel the user "duplicity", but in fact it is the actual nature of behavior you don’t have a real understanding of the users. One advantage of the A/B test is that it has a very complete and reliable answer, and the answer is, of course, screened from numerous research interviews. Let the user choose between the two answers, which is what they really need, which is what they really need to operate, and that’s the point. As in the American election, the two candidates are chosen by many of the party’s elite, and then the two most essential answers to voters are either A or B.
second and A/B tests are not just for single purposes, but can be used for multiple target selection,
if you search for the theoretical diagram of the A/B test, most authors will tell you that the answer to the test is two. But in fact, this "two" like China in ancient Chinese literature, "three" is a number. In our actual tests, we can fine out many of the fine results based on the selected answers. And classify the tests according to the results. This may be easier for a single target, but for variables, it can improve the operator’s ability to mine.
for example, the first official Mr. blog to start advertising, what kind of advertising is the user really loved? Put on the PC side and mobile terminal in the end what is the difference? It is a single target on the surface of the test on advertising. But actually this >