How traditional enterprises tap the value of their own big data

Currently, traditional (non-Internet) enterprises have recognized the value of big data, but how to effectively apply big data in combination with the status quo of the enterprise is still generally confused. In response to this status quo, the following article puts forward some feasible ideas and suggestions for enterprise customers to understand and implement, based on the relevant service experience of enterprise big data application.

The content of this article is suitable for large and medium-sized enterprises with more customer resources (ToC and part of ToB) and internal data, and it is also useful for the application of big data in government organizations (e.g., tax) with a large amount of enterprise/personal management data.

I. Where is the most valuable enterprise data

The value of big data comes from the data, and for the most valuable enterprise data, we think there are two points:

1) Internal business big data (rather than external big data) has the highest application value

The big data of an enterprise can be classified into internal (all the data generated by the production and operation links of the enterprise) and internal (all the data generated by the production and operation links of the enterprise) and internal (all the data generated by the production and operation links of the enterprise) and internal (all the data produced by the production and operation links of the enterprise). (all the data generated by its own business production and operation) and external (from external sources, such as third parties/Internet). Currently, enterprises are keen to introduce big data from external sources (e.g. Internet/e-commerce/mobile Internet) and related service applications, while ignoring the fact that the existing internal business big data is the biggest value mining target.

Most large and medium-sized enterprises have already completed the first stage (the construction of information systems and the automation/normalization of business data collection) in the process of informationization and data application. Various business information systems established over the years have accumulated a large amount of business data. But after entering the second phase (mining data to improve enterprise business operation management), the progress is slow. Compared with external data, internal business data is large in volume, diverse in content and long in time span, and is the main body of enterprise big data. Because it is directly related to the characteristics of the enterprise and y covers all aspects of operation, its value to the enterprise is much greater than that of various external data. However, these data seldom play out their due value, and most of them lie dormant or even become burdensome.

2) In internal business big data, priority should be given to data related to service customers

Internal business big data, if divided by logical attributes, can be divided into two categories:

1) Product/service related: data related to products/services (R&D/design/raw materials/production/manufacturing/feedback) around the enterprise

2) Service Customer-related: around the target customer (can be B or C) related (pre-sales / sales / customer service / operation and maintenance / activities / CRM, etc.) data

The above two types of data, service customer-related business behavior has a huge impact on business operations. Its data is also the main body of the enterprise internal big data, should be prioritized as the target of internal big data mining applications.

Two, the process of implementation

Next, for the enterprise's most valuable internal business data set, combined with consumer research and labeling research methodology, we introduce the mechanism of how to effectively mine the value of its big data.

First we give a main flow, and each step will be explained in detail subsequently.

Step1 Overall System Design: Reconstructing and Designing the Existing Internal Data

The existing business data system, combined with the actual situation and future application goals, re-organize and plan the data. The process should focus on two points:

Point 1: the organization of data, to shift from function-centric to customer-centric (organized by lifecycle stage). Business data within the enterprise, most of the current business function (system)-centric organization, not fully connected to each other. Business data used for value mining should be centered on each customer, with the user's lifespan as the line, stringing together data from all of its business function stages.

Point 2: Establish a data description system for customers with the idea of class labeling as a framework for future panoramic data integration. The source data of the description system contains not only internal data, but also external data (auxiliary). The actual data integration processing will be based on this system: existing data can be directly introduced, and missing data content as the main target of subsequent collection/outsourcing.

Taking an automotive customer as an example, its related big data, corresponding to the 9 internal business systems, are generated independently. In the data system reconstruction and integration, the schematic diagram of reconstruction is as follows:

Step2 Data Integration and Concentration : Practical Integration of Existing Data and Establishment of a Unified Big Data Platform

Based on the planning scheme obtained from STEP1, the existing business data are integrated into the Unified Big Data Platform from various business systems through technical means. The platform serves as a data analysis platform, separate from the production business system, providing support for data warehouse/structured/unstructured data.

In the integration, it is important to note:

(1) The design of the data model as well as the data ETL (cleansing/transformation) need to be unified and planned in a customer-centered manner

(2) Give full consideration to the future incorporation and integration mechanisms for the missing/inadequate data content in the new data system.

Step3 Labeling Analysis : Conduct all-round labeling analysis on customers and generate labeling description results

On the multi-dimensional user-centric data space obtained from the integration of STEP2, a user labeling system is established based on consumer research and business characteristics, and an actual labeling analysis is conducted on customers. The definition of the labeling system should take into account the basic user information, business characteristics and the purpose of future applications, and constantly expand.

For example, the aforementioned car customers, the user labeling, has defined the following categories: basic attributes (gender, age, purchasing power, occupational class ...), family situation (family with children, the second car), models / driving preferences (such as preference for SUVs, pay attention to safety ..., the pursuit of speed ...). The pursuit of speed ...), accessories focus (like original, like functional accessories), interior preferences, maintenance habits, participation in activities preferences, media habits.

Step4 Business application/excavation: Through business activities, the actual excavation and application of customer big data value

The results of labeled descriptions obtained from the analysis of all customers can be provided to all departments within the enterprise for practical application through a unified customer analysis platform. Departments can flexibly and accurately screen target customers through labels according to actual business needs (e.g., the marketing department can find target customers with children and strong purchasing power after 80s to promote MPV family car models), or discover the deeper characteristics of the product customer base (the product design department can analyze whether the target customers of the model are the same as the actual purchasing customers).

The above is a small number of traditional enterprises how to tap the value of their own big data related content, more information can be concerned about the Global Green Ivy to share more dry goods