Data analysis is sometimes not science, but human nature

The author of this article will share some experience in work processing based on his many years of experience in data analysis. The analysis of needs is often more meaningful than rigid data in the workplace. The author analyzes the reasons and gives suggestions, hoping to be helpful to everyone.

I have been in the field of data analysis for 8 years, plus working in self-media for 2 years, and have encountered many scenarios and cases.

Today, I would like to share some insights from the perspective of an insider with students who have just entered the industry.

Of course, every family has its own problems: product managers are often blamed, and programmers also have to compete with products. What I am going to say next is not to say that data analysis is worthless or to discourage everyone, but to expose problems in order to solve them .

This wave of data analysis in China began with the introduction of growth hackers from Silicon Valley in 2015, but the infrastructure of most companies has not kept up, which sometimes puts data analysts in an awkward position.

The following is about the current situation of small and medium-sized companies, which make up the majority.

First of all, there is the "contradiction" with various demand parties.

01 With the boss

Let’s not talk about whether the data analysis results are useful.

Is it just a question of logic? In fact, the boss often has the final say. If the boss says it is not true, then it is not true, no matter how rigorous the statistical probability method you use is.

As a result, data analysis is basically oriented towards the boss and caters to the boss's ideas.

02 With direct supervisor

Similar to a boss, but a superior has another attribute: office politics .

For example, a meeting is a performance in front of the boss, and the boss needs your data analysis to support your point of view and show your performance.

In addition, there is a strange phenomenon for independent data departments: the boss will prevent data analysts from talking too much to the business. Sometimes, repeated problems that can be solved by making a template are not solved. Why? It's very simple. If the business can do it by themselves after getting the template, and even learn the analysis logic, then what is the point of the data analysis department? The value is gone? So we have to manually build the moat of the data department.

03 With business colleagues

There is a concept in data science called: explainability, which means that if you cannot explain your analysis method including algorithm logic clearly, not only your boss but also your colleagues will not accept it.

But often, especially when it comes to algorithms, it is a black box. Maybe you know it very well, but your business colleagues don’t understand it, or even don’t have data thinking, so there is no solution.

What will be the consequences if the colleague refuses to admit his mistake?

Either the data analyst compromises and simplifies the analysis logic, resulting in imprecise results and large errors, but the business can accept this .

Or your business colleagues don’t help you implement the analysis conclusions and suggestions .

04 With HR

Even HR will find fault with you and ask whether your work is valuable.

Where does value come from? There is only one place to test it: business growth .

At this point you understand the previous point. If your business colleagues don’t help you implement it, your analysis will be floating in the air and worthless.

Of course, your work is not completely worthless:

• Indirect value: the analytical insights provided can bring new insights to the business;
• Automate processes and develop robots using Python or RPA to help businesses save time;
• BI reports are from the perspective of data products. Once the product is made and used by the business, it can be considered as value. This is also the reason why BI has developed rapidly.

OK, let’s finish talking about external issues. In fact, many new data analysts also have “contradictions” in their own knowledge structure.

05 Focus on technology rather than business

Many newcomers think that knowing Python and SQL makes them data analysts. In fact, these technologies have no threshold and will not be your core competitiveness in the next 5-10 years. Insight into the business is .

Compared with data analysis, data operation is easier to implement. Why? Because the latter is in the business team, while the former is often an independent department or a middle office department, far away from the business.

What will this lead to? The analysis logic is divorced from business reality, and the suggestions given are far from business goals.

Let me give you a very simple truth: always do things that are closest to money, and money comes from business.

To put it more exaggeratedly, if you don’t understand the business, you can’t even build an indicator system .

What should I do?

As mentioned at the beginning, we expose problems in order to solve them, and then become data analysts who can generate value.

06 Exercise ability

1. Understand the business and grasp the focus

Go to the front line, or even rotate to the front line to sell goods, and get your hands dirty, so that you can understand how the business is done. Next, combine the idea of ​​points, lines, surfaces and solids, from sorting out business processes to building business models, refer to the article: How Data Analysts Build "Business Models" to Deeply Understand the Business

2. Communication skills and building connections

I usually maintain a good relationship with my business colleagues. If they smoke, I will pass them a cigarette and they will become my good friends. But I don’t smoke, and we can learn a lot from each other by having lunch together at noon.

3. Manage upward and think what the leader thinks

First of all, you have to understand your superior: if it is a department leader, you must know that his situation is to reflect the value of the department and prevent other departments from backstabbing him. If it is a boss, you must understand what he wants?

4. Analyze logic and always focus on "implementation"

What do we do with the information obtained from the first three points? We need to put it into the logic of data analysis:

1) The analysis logic should be integrated with business knowledge. For example, there are many ways to calculate the repurchase rate. Which one should we choose? It is based on business attributes.

2) Treat the analysis as a project, and let the business participate in the process. Synchronize with the business in a timely manner to analyze the problems you find during the process, discuss with the business about the possible reasons, and even create some suggestions together. Let the business feel involved so that they can understand the logic of the analysis.

3) Conclusion The reason for building personal connections by tying them to business KPIs is not only to understand the business, but also to obtain the latest business trends, such as problems exposed in monthly reports, quarterly goals, and business line plans. If your analysis conclusions can solve these problems, do you still worry about the business not being implemented?

4) Use upward management tools in a timely manner to help leaders share the burden

• Report to your leader regularly to make him feel safe.
• Consider the problem from the perspective of the superior, from the theory of point, line, surface and solid, that is, think on the "surface", think one step further
• If there are problems in the process of communicating with the business, pull your superiors out to put pressure on them. Make sure the suggestions are implemented, so that the analysis of the project is valuable and the leader has something to say in the meeting.

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