Data governance is an important part of an enterprise’s digital upgrading process. How can you make data governance work twice as well, work more smoothly, and ultimately serve business scenarios effectively? In the recent TechDay online technology live broadcast, push data in the platform solution consultant a Xuan around the theme of data governance, to share a push practical experience and methodology, and combined with practical cases of enterprise data governance in the process of avoiding pit for a detailed analysis.

As a data intelligence company, For many years, Gepu has continued to manage its own accumulated tens of billions of data, and has a deep understanding of the pain points and difficulties in the process of data governance. Combined with their own data governance experience, I push the implementation process of data governance into clear strategic needs, data planning, organizational construction, scheme design, implementation and other five steps.

Step1: identify strategic needs

According to the author, in the process of data governance, top-level design and organization construction are particularly critical. ** We need to view data governance as a strategic activity with clear goals and direction so that data governance ultimately meets business needs and creates value for the business.

So, how can organizations organize and identify the values and goals of data governance at a strategic level? The first suggestion is to adopt a systematic approach — “3C theory”, that is, to comprehensively consider the current digital status, demand and capability of enterprises from the aspects of company, customer and competition. “Enterprise” refers to the current financial situation, IT construction level, data assets and product design of the enterprise; “Customer” refers to the market share, sales, marketing and other aspects of the enterprise in the industry; “Competition” refers to the situation of competing products, their own market competitive advantages, external market environment, etc.

The case of fashion clothing brand is analyzed. The customers of the clothing brand have been focusing on the main business for many years. Facing the competitive pressure from the external market, they hope to find new growth points through digital construction. Customers hope that through unified management of data, big data can be used to form user portraits, and the user information and commodity information can be linked for analysis, so as to improve clothing design and increase sales.

Combining with the current customer data island status and technical ability are relatively weak, and the wisdom of marketing, digital data to guide product design, logistics and other intentions after data application benefit and risk assessment, we recommend that customers will eventually intelligent marketing as the highest priority target, to planning and follow-up data governance path.

Step2: plan data

The premise of data planning for enterprises is to fully investigate and sort out their own data status and environment. We suggest starting from the following four points: 1. Sort out your own business system. 2. Classify existing data assets across systems and businesses. For example, in the e-commerce industry, business data can be divided into buyer data, seller data and purchase behavior data. 3. Fully investigate the details of each type of data asset, including data volume, key fields, update status, saturation, reliability, call scenario, and frequency. 4. Fully communicated with current business parties and data operators about problems and difficulties in using data. For example, in e-commerce business, there may be problems such as order data not updated in time, data heterogeneity in different regions, incomplete information in some fields of commodities, and many unstructured data.

After a thorough investigation of the internal data environment, the enterprise can make a targeted governance plan for the internal multi-source heterogeneous data. One of the key points is to integrate the data assets and build a data asset model that conforms to the business logic. For example, for the data governance of e-commerce data, it is necessary to connect the order data, user data and commodity data of different systems, so as to realize that the data of the same business category have unified fields, which will create the foundation and conditions for accurate data calculation and business algorithm model construction in the future.

For data of different business categories, it is necessary to sort out the connections between the data. The relationship between data assets can be visualized by data ER diagram or blood graph to facilitate the correlation analysis between the data. For example, if order data involves buyer, seller and commodity data, then order data can be correlated with buyer data, seller data and commodity data to build a perfect data asset relationship graph.

The last step of data planning is data stratification, which refers to the clear classification and stratification of data according to the stage of data governance and the purpose of data, such as ODS access layer, DW middle layer, ADS application layer, etc. In this way, data operators can quickly refer to the specific data layer to check and retrieve the number when the business side puts forward the demand for data use, so as to improve the efficiency of data application.

Step3: data governance organization construction

In building a data governance organization, an enterprise needs to take into account factors such as data scale, data governance difficulty, and business complexity to build a data governance organization of appropriate scale and with a deep understanding of business requirements.

For most start-ups, at least the four basic roles of data governance leader, data analyst, data steward and system steward should be set up to ensure that the requirements and objectives in the whole data governance link can be implemented. At the same time, the four roles should have a clear division of labor and a mature coordination mechanism to ensure the effective operation of the organizational structure system.

For large enterprises, they should pay more attention to the systematic management of data and information, pay attention to the improvement of incentive system and the efficient and safe coordination among multiple internal departments. Among them, the problem rising channel and strategy implementation channel is very critical. The problem rising channel refers to the timely feedback of the data problems encountered by the business department in the process of using the data to the data steward and governance team, so as to drive the continuous iteration of the data governance solution, so as to continuously improve the data quality; The strategy implementation channel refers to the enterprise from the overall business and strategic level to promote the solution of data quality issues, reduce cross-department, cross-system collaboration resistance, so as to make the process of data governance more efficient.

Step4: Data governance scheme design

When designing a data governance scheme, an enterprise should focus on data management system and data value system.

The data management system is mainly designed to give business meaning to data, scientifically measure the quality of data assets, and realize the automatic management of the full life cycle of data, including automatic online, ETL, and offline, on the premise of ensuring data security. In the fast fashion industry, the update and iteration of clothing styles are fast, and correspondingly, the life cycle of clothing commodity data is shorter. This requires fast fashion enterprises to carry out effective management of commodity data, timely update and governance of commodity data.

The data value system quantifies the value of the enterprise’s data assets in multidimensional ways to provide basis and support for the operation and decision of the enterprise’s data assets. The data value system includes three modules: data flow, data service and data insight. It refers to the ability to transfer the data value out, respond to the complexity requirements of the business in an agile way, and provide effective reference for business analysis and insight. Enterprises need to continuously improve and iterate data governance processes in order to build the value creation loop of data assets to truly activate data assets.

Step5: data governance implementation

The last step is implementation. According to the data governance scheme, the enterprise completes the data governance links such as overall planning, standard landing, data registration, data integration, data exploration, monitoring and evaluation in turn. It should be emphasized that the process of data governance in an enterprise is very complex and that human resources alone are not enough. Automated implementation tools are equally important.

How to select a data governance tool? Suggestions from the following three aspects to consider: 1. Technical functions, that is, product functions to be comprehensive, to meet the implementation of the process of all the needs. 2. Reusability refers to the function of the product that can be used in a variety of business scenarios. 3. After-sales improvement, due to the complexity of data governance, simple products are not enough to meet the needs of enterprises, detailed use of training and professional data governance consulting services are equally important.

Here we will focus on the next push of the data platform products – daily conquer number platform.

With many years of experience in data governance, the daily governance platform has complete functions and provides data governance services for the whole link from the data access layer to the data application layer. The platform visualizes the entire data governance process and enables business people to easily develop and use data. Daily-conquer number platform has been serving many industries such as brand marketing, smart transportation, smart city, etc., and can meet the diversified and complex needs of different enterprises to carry out data governance. At the same time, it also provides professional data warehouse planning and data modeling services to help enterprises build data asset center and truly dig out the practical value of data to business.

conclusion

In general, data governance is a systematic, long-term project. Enterprises also need to pay close attention to business development and trends in specific data governance practices, and timely iteration and optimization of data governance strategies, in order to properly govern data and release the value of data.

In the future, Getui will continue to share the content of data center, data governance, data mining, algorithm modeling and other aspects, and Getui TechDay live streaming will continue, please continue to pay attention.