CDA holder Wang Mingyue

2 years of experience in data products, doctoral student in management, second-level license holder of CDA data analyst.

I. What are data products?

I checked the details of the data product manager position recently released by Boss Direct Recruitment on Today's Headlines, Meituan, Xiaomi and other companies, from which I can have a preliminary understanding of the data product manager.

In the job description of Today's Headlines, the data product manager is known as the soul and leader of big data empowerment business. This position is mainly responsible for the planning and design of data products for business lines, and supports TikTok, short videos, live broadcast, education, games and other key business lines. At the same time, it is necessary to go deep into the front line of business, combine business scenarios, link data center resources, output data product design schemes for business lines, and promote product implementation. In addition, it is also necessary to have a keen insight into the pain points of the industry and business, and use data capabilities to provide big data solutions with industry characteristics for business lines to empower the rapid development of business.

The job details of Understanding Car Emperor are similar to today's headlines, both of which focus on the design, implementation and data empowerment of data products corresponding to the business. In addition, it is necessary to build data tools to improve the efficiency of problem discovery, troubleshooting and solution from a data-driven perspective, and have independent product planning and resource integration capabilities.

The position of Meituan is closer to the work of data analyst, responsible for the construction of data analysis system in the direction of intelligent customer service, including the construction of core business index system, report analysis application construction, etc. It is necessary to deeply understand the logic of business operation, combine data extraction and analysis, and put forward suggestions for data services and system optimization.

In addition to the above two points, Xiaomi's position focuses more on business growth services. In addition, you also need to be responsible for the iteration and user operation of data products, manage the progress of the project, coordinate product design, research and development, testing and other links to ensure the high-quality output of data products.

In summary, a data product is a service or application that helps enterprises improve decision-making. It may be an APP or an external application. The main goal is to use data processing to help managers make decisions. The screenshot above is a typical data product, which can monitor the new users of the APP, the number of startups, etc., and conduct real-time statistics.

II. Work content of data products

When I first interned, I did a product manager-related job. At that time, I had not been exposed to data products. At that time, I thought that data products might be more difficult, and I was more inclined to technology. But after I did these two jobs at the same time, I summarized some differences.

The product manager finally delivers the requirement document, that is, the PRD document, as well as the tracking iteration and maintenance of the version. What the data product manager delivers is the latest version of the indicator dictionary and the definition management optimization of data indicators, especially the statistical caliber, because there may be subtle differences in the understanding of the same indicator by different business parties or data products, resulting in large data differences.

Data analysts are indispensable in the work of data product managers. The focus of data analysts is to analyze the business operation of the organization through data. For example, in the field of express delivery, due to the dispersion of outlets, we cannot observe the whole process of products from raw materials to finished products like the workshop assembly line.

At this time, we need to use data to analyze the business operation, or simulate the business operation through data twin technology, find problems and put forward suggestions for improvement. This is also the reason why I took the CDA data analyst exam. The preparation for the whole CDA data analysis has greatly improved my personal ability. It is recommended that all friends improve their data analysis skills.

The work of a data product manager combines the responsibilities of a product manager and a data analyst. We use the conclusions of data analysts to design and plan data products and decide in what visual form to display the data products to the demand side. At the same time, we also need to understand the work content of product managers such as market research, product design and project management.

III. 5 data analysis skills that data products should have

In my work, data product managers need to have 5 data analysis skills, mainly including data platform construction, number collection, automatic mail, reports, strategies, etc., among which strategy is the closest to the work of data analysts.

1. Build a data platform

This is a relatively basic and technical job. After entering an enterprise, we will integrate data, build a data warehouse and database, and build a data access layer based on the data generated by users when using the product. In the data access layer, we will build reports, data acquisition settings, automatic mail and other functions.

Data governance and security are also important work contents. Although I don't have much contact with it, there is a special position to be responsible for. The construction of data warehouses is equally important. I have participated in some when I was in the enterprise, but not much. When preparing for CDA Level 1, I learned a lot about the stored data models of data warehouses, such as star models and snowflake data structures, as well as relational databases, which are especially useful in work.

Two. Take the number

After building the data platform, the data product manager of the new company usually needs to take on the number requirements. At this time, we only need to obtain data on the platform built by our predecessors. In this process, we can understand the business process more easily and understand the storage logic and type of the database according to the business process.

Take the express delivery industry as an example, we will first receive the customer's order, and then the courier will collect it to form a package form. The package goes through the collection, on the car, takes the trunk line and branch line, passes through multiple outlets, and goes through many unpacking and collections, and finally arrives at the delivery outlet and is delivered by the courier. We need to understand these processes in order to know which table to take the number from and how to track the whole process of the package, including delivery time and time limit.

3. Automatic mail

Automatic mail is generally a timed automatic mail. We must pay more attention to two aspects: one is its sending frequency, and the other is the size of the file, because the size of the file actually affects the success of the automatic mail. In addition, the overall content of this email will be considered. What is the caliber of which indicators it needs, and then whether there are some existing calibers, but the existing caliber may be problematic, so under different business scenarios, this caliber will also have certain changes. In addition, whether this automatic mail needs to be made into a report, there is another point to pay attention to: it is possible that the business side will call at night to say that the mail reception has failed, so whether our recipient's situation should be set to the receiving group situation. These are all details to pay attention to.

Four. Report development

The use of BI tools, such as Tableau and Power BI, is also necessary for data product managers to master. These tools can help us develop reports, improve the work efficiency of the demand side, and let them see their workload more clearly. In addition, enterprises will also build their own dashboards, which are usually more flexible, not limited by third-party templates, and do not require additional costs. They are usually built by technicians.

5. Tactics

I think strategy-type work is very important in enterprises, and this part of the work is very close to the responsibilities of data analysts. When our business encounters urgent and important problems, such as the problem of bursting warehouses in the field of express delivery, especially during the Double 11 promotion, there is too much backlog of parcels in the outlet, the inability to deliver, or the loss of profits due to cost control problems, we need to analyze the reasons for these problems.

At this time, we will use data analysis tools, methods and models, including tree structure analysis, 28 analysis, four-quadrant analysis and peer group analysis, to identify the essential cause of the problem. We need to combine the knowledge base and understanding of the field to output strategies to solve these problems.

For example, if an enterprise encounters a decline in total sales, we need to determine whether it is caused by members or non-members. Assuming that it is found that the proportion caused by members is relatively large, we need to further analyze it. It may be due to the decline in the number of visits per capita, resulting in a significant decrease in total sales. Through these analyses, we can formulate corresponding strategies to solve problems.