data is a very good thing that can help us make relatively objective product decisions. However, there are a lot of traps in the data, if we do not have a good logic analysis ability, is likely to be playing with data and make stupid decisions. Today, let’s talk about the pitfalls and how to avoid falling into the trap.
Before the start of
, we first look at a few of you may have encountered in the case of work it!
buy goods 80% a user is a user, so when the user was found to be a user, should give users a more general than recommended goods
when the b value increases, the conversion rate of B function is reduced, so it should be limited to a certain value of
user survey found that 90% users like C features, so we need to increase the C function
according to the existing data found that D value is relatively high commodity has a higher click, so the D value of the goods should be ordered in advance
users using the E function than the user does not use E function conversion rate is low, so the E function should be offline
you should guess, this is certainly a few examples fall into the trap of data description, so you can think first the description of where is the problem, if you do not want to understand this article suggest you see three times.
1 users have 80% class a buy goods are a user, so when the user was found to be a user, should give users a more general than recommended goods
inverse proposition trap
the first case of the trap is very confusing, in the actual work, I often encounter with this description to ask for a certain kind of commodity in some situations to give more traffic demand. From the data, it can explain the purchase of class a commodity a user side, but this can explain a user preference than the overall user A goods? Obviously not, the case as follows:
assumes that the demand for a class of goods and non class a goods is 50. On average, the flow rate of a class a commodity and non class a commodity should be as follows: 50%. Although the user a user in the purchase of class a goods accounted for 80%, but the user needs a commodity only 45%, not to the normal average case 50%, the reason is because the purchase of a commodity of a non user user ratio is higher than 80%.
This is called the inverse
trap trap, the inverse proposition should be the concept of mathematics in senior high school, if you forget we cite a simple example, if there is a judgment statement "if Yu Qian is the father of Guo Qilin, then Guo Qilin is the son of Yu Qian," if Guo Qilin is the son of Yu Qian, the "