↑ Click on the " Coke 's Data Analysis Road " above to follow + star ~
There is a high probability that it will be updated at 8:25 a.m. every day
This is the third article in the intensive reading series
Hello, everyone, I am Coke. Today is the third article of intensive reading "Data Analysis Thinking: Analysis Methods and Business Knowledge", which corresponds to Chapter 345 of this book. It mainly talks about the process of solving problems with data analysis and the two A practical case: business knowledge, indicators and cases of the domestic and cross-border e-commerce industry.
The first two intensive readings are as follows:
Intensive reading 1: Always talk about business, what are the commonly used indicators for business
Intensive reading 2: 11 data analysis methods, don’t say you can’t
01
Solve problems with data analysis
The content of this chapter actually tells us to combine the individual analysis methods of the previous chapter.
clear question
By observing the phenomenon, define the problem clearly, and clarify the data source and accuracy, which can usually be confirmed from the time, place and data source.
Secondly, fully understand the business indicators, including indicators and meanings , and who to compare with .
For example, the profit achieved by a store in the first half of the year is still far from the target of an average monthly profit of 5 million yuan set at the beginning of the year. The leader asked to find out the reason for the failure to meet the target.
First of all, it is necessary to confirm the accuracy of the data, clarify which department provided it, what is the data for each month, and is the data checked and correct?
Then understand the indicators, how profit is defined, etc. Compared with the first half of the year, there is a gap.
Note:
avoid subjective assumptions
Analyze the reasons
To analyze the reasons, two questions should be clarified:
what went wrong
why this problem occurs
First analyze the key factors, and use the multi-dimensional disassembly analysis method to disassemble each indicator, such as sales = sales revenue - sales cost - non-operating expenses, sales revenue can also be disassembled into customer unit price X number of users, as for disassembly To what extent, it should be flexibly grasped according to the understanding of the business and practical problems.
Then use the hypothesis testing analysis method to find out what went wrong.
After finding out what went wrong, further in-depth analysis can be performed, such as edible correlation analysis to analyze the deep-seated reasons.
make a suggestion
In the step of making suggestions, regression analysis and AARRR model analysis methods can usually be used.
The purpose of using regression analysis is to calculate the extent to which a certain cause can affect the target, such as predicting how much sales revenue must be achieved in order to achieve the target of 40 million profit in the second half of the year.
For regression analysis, refer to this article:
Come back strong! Let's talk about regression analysis
Next is the actual combat part, which talks about the business knowledge, common indicators, and case analysis of 12 industries, starting with the e-commerce industry
02
Domestic e-commerce industry
4 business models:
B2B : Enterprise sellers - enterprise buyers, such as Alibaba, with high order volume
B2C : Enterprise sellers - individual buyers, such as Tmall, Amazon, "XX Official Flagship Store"
C2C : individual sellers - individual buyers, such as Taobao
O2O : sellers online - buyers offline
The e-commerce industry has undergone an upgrade from traffic operation to user operation
business indicators
-
Proportion of old and new users
Proportion of new and old users
Repurchase rate : the proportion of repurchase users, usually measured over a longer period of time, reflecting user loyalty
Average repurchase cycle : the average time interval between repeated purchases by users
Repurchase rate : an indicator for analyzing the attractiveness of short-term promotions to users
The commonly used indicators of e-commerce can be divided into two categories: people and goods, which are user transactions and product management indicators
user transactions:
Number of visitors UV : the number of unique visits to the page where the product is located
Number of additional purchases : the number of users who put a certain item in the shopping cart
Number of favorites : the number of users who have favorited a product
Purchase stage:
-
Total turnover GMV
Payment conversion rate : number of payment users / number of visitors
Discount rate : GMV/total amount of tags
Receive and return:
Rejections : The total amount of rejections and returns
Chargeback Amount : The total amount of rejections and returns
Actual sales : GMV minus refund rejection
Stock indicators:
S PU number : model number, such as iphone 9 is a SPU
Number of SKUs : Refers to the specific product number of a certain model number, specific to the color and size, for example, the iPhone 9 has 3 SKUs, which are black, white and red
Stock value : tag price X inventory quantity
After delivery:
Sales ratio : sold out rate, GMV/stock value, used to see the circulation of goods and optimize inventory
Sales rate : number of SKUs with sales / number of SKUs on sale
Case: Analysis of Repurchase Rate Decline
Clarify the problem: find that the repurchase rate of Double Eleven users in 2019 has declined, and analyze the reasons
Comparative analysis found that both the number of users and the number of repurchases have increased compared with 2017 and 2018, but the repurchase rate has dropped. This is the Simpson paradox we mentioned earlier. The reason should be that the increase in the number of repurchases has not caught up with the growth of the base .
Using the multi-dimensional disassembly analysis method, first disassemble from the R value, that is, the R value in the RFM analysis, the time interval of the last purchase, and compare the two groups of R<=365 and 365<R<=730 over the years The base, the number of repurchases, and the repurchase rate are found to be higher in 2019 than in 2018, indicating that the R value cannot locate the reason for the decline in the repurchase rate.
Then disassemble from the F value, which is the purchase frequency, and disassemble it into F=1 and F>1. It is found that the repurchase rate of users who only purchase once has dropped significantly, resulting in a decrease in the overall repurchase rate. It can be obtained from F= 1 Start with base users for more in-depth analysis
Divide users with F=1 into multiple groups according to the R value, and find that the repurchase rate of the group 90<R<=180 has dropped the most year-on-year, which is the main reason for the decline in the repurchase rate of users with F=1. I found the reason. A large number of new users brought by the "618" promotion in 2019 did not survive. These new users were mainly low-priced users who were attracted by the advertisements on the promotion page of the platform.
03
Cross-border e-commerce industry
Cross-border e-commerce is an international business activity that delivers goods and completes transactions through cross-border logistics. 3 business models :
Platform type : Invite sellers to settle in cross-border e-commerce platforms, such as Tmall Global
Self-operated type : use cross-border e-commerce platforms by yourself, such as Xiaohongshu and Kaola Haigou
Hybrid type : both platform type and self-operated type, such as Amazon
business indicators
From the perspective of the funnel model, the business indicators of advertising have the following indicators to pay attention to:
Case: Member Analysis
The Amazon store membership activity must ensure that the order arrives within two days. The background data shows that the delivery rate is only 90%, which is lower than the standard of 100%, resulting in the failure of the activity.
Clarify the question: What is the reason for the lower-than-standard member delivery rate?
Reason analysis: sort out the business process from the buyer's order to receiving the goods, and use the comparative analysis method to analyze which link has the problem. Through the analysis, it is found that the main reason is the failure to deliver the goods in time.
Dismantling from the warehouse dimension and in-depth analysis, it was found that there were many abnormal orders in the warehouses in 06 and 07, which was the main problem.
The above is the whole content of this intensive reading, next time I will intensively read the case of the financial industry
Guess you like :
11 data analysis methods, stop saying you can't