Intuition leads to misunderstanding: "If you are a traffic analysis product manager of renrenren, how can you explain that the total traffic of Renren declined by July 1-7 every day from January 1, 5%?" Feelings

I have a very good blog, but I have benefited a lot from my blog. I just subscribed to the blog and published some ideas to encourage you to exchange ideas.

I just read "if you are the traffic analysis product manager of Renren.com, how can I explain that the total traffic of Renren.com fell by July 1-7 every day from January 1, 5%?" This blog Http://piaolin.sinaapp.com /? P = 28
The original blogger thought, "an intuitive answer is given in an instant: Because Renren's main users are students, and early July may be the day to take the exam or go home from the summer vacation. Such a product manager is very valuable ."
In fact, I think it is not so valuable to give an intuitive answer in an instant, because sometimes many problems may be due to a mix of reasons and many aspects are full of external factors.
In the book "fast and slow thinking", this book provides examples of intuitive thinking on misleading projects.
"A study of the incidence of kidney cancer in the 3,141 counties of the united a> <H has States reveals a remarkable pattern. the counties in which the incidence of kidney cancer is lowest are mostly rural, sparsely populated, and located in traditionally Republican states in the Midwest, the south, and the West. what do you make of this?"
In fact, the only reason is that "the law of small numbers" is that if we rely on intuition to provide some analysis similar to "political environment", isn't it done?
I appreciate the fact that at first I could not give a conscious answer, but I analyzed the data from multiple dimensions.
For example, after clustering analysis is performed on people with a decrease in the chain, it is found that the school-based distribution is re-associated with whether it is caused by the school test week.
Assuming that the decline caused by different exam weeks in different schools is not due to diving, but rather a gradual decline, then modeling to verify the performance of students and non-students in the model. Other reasons are ruled out because of the exam weeks.
In the end, I think that the comprehensive analysis to overcome the intuitive interference will have a positive and negative impact on all the project subjects, as well as the depend on product factors. All the assumption, risk, and external factors will be given in detail and then the assumption will be verified repeatedly review project objective is the right path
This is also the core idea of quantum investment.
Intuition allows us to respond faster and better to things in our lives, but it is worth thinking about how to use intuition to guide projects.

//////////////////////////////////////// //////////////////////////////////////// //////////////////////////////////////// ///

Text of the original blog

If you are the manager of Renren's traffic analysis product, how can you explain that the total traffic of Renren's network fell by July 1-7 every day from January 1, 5% to seven?

The above is a problem in zhihu. I tried to interview many product managers, and many of them did.

How to analyze and investigate this problem from the data perspective is a test of product manager's understanding of the data, whether it is comprehensive, and whether it has internal logic, and whether a good summary and sublimation have been made in actual work.

The answer is as follows:

• Is there a similar trend in the data of the same period last year?
• Is there a similar trend last month or last month?
• Is the overall website traffic decreasing? Or is traffic on some pages (or subdomain names) falling?
• Is access in all regions decreased? Or is access in some regions declining?
• Is access by all users degraded? Or is access from new or old users declining?
• During the same period, do competitors have any actions?
• During the same period, what products were launched?
• During the same period, what have been launched for Operation promotion and what have been deprecated?
• Does UV decline at a rate of 5%?

The above non-standard answers can be further considered. In fact, during the interview, many product managers who claim to be focusing on data have not been able to systematically think about this question. They often fail to answer questions in two or three ways.

If this is the case, it is an interesting data interview question. However, there is a relatively small proportion of product managers (about 10%), giving an intuitive answer in a flash:Because Renren's main users are students, early July may be the day for the exam or returning home from the summer vacation.Such a product manager is very valuable.

Product managers need data guidance, but better product perception or product intuition. The data will lie because there are not enough people to interpret, and the data on the films is misleading. Data never lie, because every data is real.

Most of the products developed by large companies place great importance on data. However, there are not many people who can think about and guide product optimization design through data. The products developed by small companies ignore the data and it is good to go online, but without the core data guidance, they will also be lost. One is the ocean lost in massive data, and the other is the desert lost in data.

Data capabilities are the basic capabilities of product managers, but they are not omnipotent keys.

After a real example, the data shows that the conversion rate of a complicated registration and authentication process is only 6%, So we plan the corresponding process optimization and make the entrance more eye-catching. After two months of discussion, one month of development, the conversion rate did not increase but declined within one month after the launch. After one month, the conversion rate was increased to the original data, but it was still the same. Do you know the reason?