Regarding the construction of an enterprise's big data system, it can be divided into 4 levels, and there can be a progressive relationship between each level. Although the business dominance is different, the construction ideas are the same.
The following picture is a summary of the essence of this article, and I will discuss it with you one by one.
1. Data base platform
The basic data platform construction work includes the construction of basic data platform, data specification, establishment of data warehouse, data quality, unified business caliber, etc.
The data of many companies cannot be effectively used. First, the data is scattered on the servers of products of various departments, and the data of each business system is not connected. Second, there is a lack of unified data specifications. The business system data is reported according to their own caliber and understanding habits. Standardized SDKs and reporting protocols make it difficult to build high-quality data warehouses.
The construction of the big data platform architecture is not a high-level technical activity. The realization of the value of the entire platform actually requires the cooperation of various departments of the company, which is an interdependent relationship. For example, the establishment of the key data indicator system needs to be refined from the business indicators of each department and approved by the business department. Common key indicators, such as new users of marketing business, effective new users, active conversion rate, cumulative retention, channel effect, etc. For example, the sales department, daily sales, monthly sales, payment ratio and so on.
2. Data reporting and visualization
In the first level, by standardizing the data indicator system, unified definition, and unified dimension distinction, standardized and configurable data report design and intuitive visual output design can be easily carried out, including financial, sales, supply chain and other data categories. . Common data reporting tools include FanRuan FineReport, birt, and Crystal Reports. Small scale can also be replaced by Excel, but it requires a certain amount of development and usage. Enterprise reports can usually be divided into basic query reports, management analysis reports and subject analysis reports.
-
Basic query report: from basic business and daily work, the function acts on a specific work, such as sales performance query, commodity inventory query, in-transit inventory query, purchase order query, etc.
-
Management analysis: It is not limited to a specific job, but covers a certain job module of the relevant personnel. For example, store manager performance management kanban, inventory management, abnormal store management, etc. This type of report is based on daily management work. By viewing this type of data report, we can monitor the current state of the responsible business and find problems, which are mainly used for decision-making assistance.
-
Theme analysis: Different from daily management reports, this type of report is more targeted and proactive. It needs to analyze a certain module and theme, and find and think about problems by analyzing report data.
Each category serves different purposes at different levels. Basic reports are used for business personnel inquiries, and management reports are used for management analysis to make decisions and report to superiors. Thematic analysis is used to analyze problems and develop business.
3. Refined business analysis
Certain businesses require refined management, such as the operation of Internet e-commerce, for which the concept of "growth hacking" has also been proposed. On the basis of establishing a data platform and visualization, various analyses are carried out on the existing sales user behavior, income data, etc., and daily, weekly, monthly, and various special analysis reports are output. Taking the Internet as an example, common data analysis tasks are as follows:
1. Product analysis and optimization through A/B testing;
2. Use the funnel model for user touch analysis, such as the conversion of advertisements from exposure to activity;
3. Real-time feedback of marketing promotion activities;
4. Long-term business health analysis, such as analyzing product growth and health from the user flow model and product life cycle;
Common data analysis tools
Free tools: excel, SPSS, R, Python
Paid apps: SAS, Tableau, Big Data BI tool FineBI
The following figure is a common data analysis idea on the Internet:
4. Strategic analysis and decision-making
Strategic analysis and decision-making are more based on business-level analysis and analysis of major decision-making changes. These decisions often require the support of a large amount of data and indicators, but in the past they relied on reports and experience.
If an enterprise wants to implement the big data system, it is recommended to use machines to monitor business operations, and on this basis, let people do empirical analysis and strategic judgments that humans are better at.
In essence, data can play a better role in the operation and refined management of enterprises. It is a daunting task for an enterprise to build a big data system. No matter who is in charge, it is necessary to persuade the top management and provide strong top-down execution.