Data Analysis Framework 1.0: Three Steps from Golden Mind Circle to Universal

The author of this article starts with the pain points of interviews, introduces the "Golden Circle Thinking", and extends it to actions, and then introduces the "Universal Three-Step" framework of data analysis, hoping to be helpful to everyone.

Today, I will introduce to you a practical analysis framework, which I named: Qinsi Parallel Analysis Framework.

1. Starting point: Golden Circle thinking

Core: Why, what, and how?

1. Start with the pain points of the interview

How do you describe your project experience when writing a resume or interviewing for a job? You should use "problem thinking" to first talk about the problems you encountered at work, then talk about how you thought about and analyzed them, and how you solved the problems, and what results you got in the end.

Including the "STAR framework" that many people admire, it is also very suitable for describing project experience:

• S ituation: the situation in which the event occurred
• Task :
• Action : What actions did you take in response to this situation?
• R Result: What was the result? What did you learn from this situation?

Have you noticed that these excellent solutions have one thing in common: their narrative logic all starts from "why".

2. Golden Circle Thinking: Start with Why

The underlying logic behind it is the golden circle thinking.

Golden Circle thinking consists of three concentric circles, divided into the inner circle, the middle circle and the outer circle.

• Inner Circle: Why (Goal)
• Middle Circle: What (Current Situation)
• Outer Circle: How (Action)

Note: This is my revised version. The core of model thinking is "flexibility", so that the model can be used for our own purposes. Therefore, there is no need to worry about what the original version is.

This way of thinking can help us better understand and solve problems, as well as communicate better with others.

Small case: Use the golden thinking circle to explain the user segmentation clearly. Let's talk about an example of class division in a school. Before the class division, suppose the average score of all classes is 80 points. On the surface, this score proves that students have no problem understanding the knowledge points.

But in fact, this score is the result of the differences between the two groups of people with high grades and low grades offsetting each other. That is to say, if the teacher teaches according to the standard of 80 points, the group of people with high grades will feel that it is inefficient, but the group of people with low grades will not understand.

This is the problem, which is the pain point in the scenario of school class division, corresponding to why

After the classes were divided, the scores of the high-achieving students in the class were higher than 80 points, while the average scores of the low-achieving students in the class were lower than 80 points, and the differences were reflected again. Therefore, the essence of student class division and user stratification is to average out and magnify the differences between groups, so as to provide different marketing strategies.

The process of dividing students into classes is based on their grades. To be more professional, it is based on the index of grades to cluster people. In the process of user stratification, we can use a single index, such as price range, to divide people into groups with different price preferences. We can also use multiple indexes for clustering. For example, the RFM model is a multi-index clustering model.

So to sum up, the essence of user stratification is to de-average and cluster indicators.

The above explains the essence of user segmentation, corresponding to what

So, after all this discussion, how do we do user segmentation? Two elements are needed: segmentation dimensions and segmentation standards.

Let’s look at the school’s class division operation. Students’ grades are the dimension of class division, and 70 points is the standard for class division.

In the actual practice of user stratification, the stratification dimensions and stratification standards rely heavily on business experience. Indicators that meet the business scenarios are selected from a large number of indicators, and analytical methods are applied to calculate thresholds to divide the population.

The last part is how to do user segmentation, corresponding to how

3. How to turn Golden Circle thinking into action

To turn the Golden Circle into action and achieve better influence on others, the key is not to tell others "what" or "how to do it", but to give reasons "why".

The key to activating the Golden Circle thinking is to find the "why", that is, to find the deep inner motivation.

In the book "How to Start Golden Circle Thinking", the author introduces 7 steps to find "why":

The most critical step is at the end: use "one sentence" to clearly explain your mission, the problems encountered in the project, etc. If you can't explain it clearly in one sentence, it proves that your understanding is not deep enough.

2. Three universal steps to data analysis

Core: Identify the problem - Analyze the cause - Provide implementation suggestions

Back to the data analysis scenario, based on the Golden Circle thinking:

• Start with why. In data analysis, we need to clarify the background, problems, and goals of the project → Clarify the problem
• Start analyzing the problem, understand the business status from the data, and use appropriate methods to analyze the cause of the problem → Analyze the cause
• The analysis process itself is not very meaningful. What is important is to produce conclusions and provide implementation strategies → Implementation suggestions

Considering that we have to deliver a data analysis report in the end, we need to add the "general-specific-general" logic of writing the report to the above basis, so that we get a general data analysis framework 1.0: the "universal three-step" data analysis .

When doing data analysis, you need to have a structure and a main line, so that the results can be based on evidence. For example, the "general-specific-general" structure that my Chinese teacher often emphasized in composition classes in school is also followed in data analysis:

1. "Overall" - First conduct an overall analysis to understand the overall situation

This stage corresponds to the "clarification of problems" in the data analysis process: by analyzing the overall data, looking into the current business situation, and combining the analysis framework to clarify business problems

2. "Division" - Based on the observation of the overall data, find the problem points and conduct targeted analysis

This stage corresponds to the "analyze the cause" in the data analysis process: break down complex business problems into multiple small problems through multi-dimensional analysis methods, and select appropriate analysis methods and analysis models to analyze the problems one by one.

3. "Total" - Summarize and summarize the analysis of each part and make suggestions centered on business goals

This stage corresponds to the "implementation suggestions" in the data analysis process: the data conclusions obtained from the analysis need to be combined with business scenarios and converted into feasible strategic recommendations to help enable business growth.

Author: Brother Biscuit; Official Account: Brother Biscuit Data Analysis