An enterprise-level BI project seems simple, but the actual construction difficulty is far beyond imagination. How to build a BI project from 0 to 1, whether there are detailed implementation steps, and whether there are suitable BI tool recommendations, these are the questions in the minds of many enterprises before adopting BI systems. Therefore, this article will discuss in detail what capabilities are needed for the construction of enterprise BI capabilities.
01
Business intelligence is not equal to reporting
Business intelligence BI generally refers to providing data query and analysis report functions for business personnel after building a data warehouse, visualizing data and obtaining effective information from data, thereby supporting business analysis, managing risks, and supporting high-level decision-making analyze.
Many partners think that business intelligence is just making watches. In fact, it is not just making watches. It integrates a series of technologies, including: data warehouse technology, online analytical processing, data visualization and other technologies to realize the commercial value of enterprises. , to help enterprises carry out knowledge transformation, so as to make more scientific and accurate decision-making.
Of course, we didn’t do it directly when building a BI system. We set a goal, which is to mine the value of data and promote the improvement of decision-making management capabilities through data collection, centralized management and control, and data application construction:
● In data collection, our goal is to uniformly store the data of enterprise groups
● In centralized management and control, our goal is to build a unified data view and realize unified data sharing
● In data applications, our goal is to flexibly analyze applications and dig out the value of data
Business intelligence is neither a function nor a product, but a complete set of solutions, the core of which is the collection of data sources, data integration and processing, and the application of data analysis.
After years of project experience, Yixin Huachen has integrated it into this picture:
We can put useful data from different business systems of the enterprise into the warehouse through operations such as extraction, conversion, and loading, so as to perform various analysis and applications on these data, and finally release them to users to provide decision support for the enterprise.
02
The "Tao, Law, and Instrument" of Business Intelligence
Yixin Huachen summarized it from three aspects, one is the perfect process, the other is how to implement it, and the other is a tool to help the implementation quickly and efficiently. We call this the construction of business intelligence: Tao, law, and device.
*
The Way of Business Intelligence
On the way of business intelligence construction, Yixin Huachen summarized and divided different tasks and links, including project construction, data warehouse construction, data analysis, online maintenance and so on. Let's focus on project construction, business intelligence, and data warehouse construction.
1. The way of project construction
Our business intelligence is managed as a project and also as a project life cycle, which is mainly divided into five activities:
The first step is project planning, including: component team (find the key person of the project), project initiation, clear goals, project plan, etc.
The second step is design, including demand research, establishment and review of requirements specification, prototype design and review, and detailed functional design.
The third step is project construction: system environment construction, data warehouse construction according to requirements specification and prototype, development and testing of data analysis applications (process testing, functional testing, and stress testing according to established procedures).
The fourth step is to go online: a trial operation must be carried out before going online, organize beta users to conduct trials and make suggestions, optimize and confirm, and then carry out training delivery.
The last is iteration and optimization. In the process of use, users will raise requirements and enhancements. As enterprises develop and progress, there will also be optimization of architecture and analysis models.
2. The way of data analysis
The most important thing in project construction is data analysis. We have refined data analysis into data analysis lanes. As a process of researching data, data analysis will go through several steps: clarifying goals and controlling analysis requirements, sorting out analysis index system, integrating data sources, designing data warehouse architecture, and designing analysis reports.
The first is to clarify the goals and control the needs. First, consider the business goals and business strategies, determine the business goals and frame the scope, and it is not appropriate to make the range too large. Demand control is a very important link. Demand control does not mean to control which needs are not implemented, but to restore the user's scene in an optimal way, and sort out the indicator system according to the demand, that is, what is needed to realize this analysis? Indicators and data, what is the caliber of these indicators, which systems do they come from, how can their granularity meet my needs, and whether we need to make supplementary records if they cannot meet the requirements, etc. Based on these data, we need to build a data warehouse and put it into the warehouse , and finally the analysis link of the visual report.
3. The way to build warehouses
There are also divisions in the construction of data warehouses:
Based on the previous business requirements, we need to design, design and implement data warehouse layers: conceptual model, logical model, physical model:
Conceptual model: a model framework for tailoring, merging, and expansion based on comparative analysis of actual business
Logical model: that is, according to the analysis requirements, assign attributes to the conceptual model
Physical model: it is the table object in the database, and then perform data mapping and ETL development and testing
*
The Law of Business Intelligence
We generally use a top-down approach, or a bottom-up approach:
Top-down, as the name implies, is the KPI index refined based on the management and control of the enterprise. It is determined according to different industries, different goals, and development strategies and stages. cost and so on. Design indicators according to the needs of the enterprise. The top-down step is to drive according to the indicators, refine the customer's operation KPI, and drive relevant personnel to participate in the construction according to the KPI.
Bottom-up, it is provided according to the user's role and scenario, and is divided into different roles. If the objects used are mainly production personnel and human resources, research around the scenarios of production and human roles, find indicators, and meet the daily needs of these roles Work.
*
Business Intelligence Tools
We know that if a worker wants to be good at his work, he must first sharpen his tools. This is what it means. The previous Taoism needs the blessing of tools.
If we have an advanced one-stop data processing and analysis platform, which integrates data warehouse technology, data mining display, and online analysis and processing technology, such a platform will definitely allow us to get twice the result with half the effort.
Just like Yixin ABI, it can effectively integrate data, and then conveniently carry out data construction, and can carry out index construction and maintenance, and display and apply various data based on unified construction indexes, including report analysis and agile analysis , report application, large-screen display, etc., and finally can be displayed to customers through PC, mobile terminal, and large screen. At the same time, it can also be easily integrated into third-party applications, including various APPs and micro-sets, as well as customer business systems.
03
Five core steps of business intelligence implementation
Based on the experience of business intelligence projects in the past two decades, we have summarized five general steps for the implementation of business intelligence, namely: clear goals and analysis requirements control, analysis index system combing, data source integration, data warehouse architecture design, and visual analysis Report design.
Next, let's break it down one by one:
The first step is to clarify the goal and analyze the demand control
In the process of sorting out business requirements, we often encounter the following problems:
1. Business departments often fail to put forward specific analysis requirements. It seems that they want this too, and they want that too.
2. The IT department does not understand the business requirements and cannot propose effective business requirements
3. BI project requirements analysis involves many departments, which causes cross-departmental communication problems
Based on this situation, we generally adopt a prototype to provide solutions, which can stimulate the desire of business personnel to express their needs, simplify communication, and reduce deviations.
In addition, the business needs should take into account the development stage and business goals of the enterprise to determine the key indicators of BI construction. The key indicators should not be too many, too many will cause energy dispersion and lack of focus.
Clear goals and analysis requirements can be divided into four steps: clear user scenarios, determine core processes and information architecture diagrams, determine business processes and page flow charts, and finally draw prototypes
The second step is to sort out the data analysis index system
The combing of the indicator system will determine the KPI indicators. It must be carried out by the business personnel of the enterprise and the BI vendor. It must not be done alone. The business personnel provide business, and the technical personnel provide ideas, methods, and technologies. Together, business indicators and data are matched to determine The extracted data determines the calculation caliber and method.
It is mainly to match KPI indicators with specific data, and determine the data that needs to be extracted to calculate KPI indicators. Some indicators are calculated from multiple data, and the calculation method needs to be clarified to provide a basis for data preparation.
The steps are as follows: First, define the indicators, define the analysis model, specify the storage of indicators, specify the number and quality of indicators, build an indicator system platform, and implement implementation.
The third step is the integration of data sources
The data source may come from external sources such as the public data of the Statistics Bureau, or from multiple internal business systems, or it may need to be supplemented by users, or it may be an excel form. These are source data. We need to extract these data through ETL Form an application-oriented data mart.
The fourth step is the architecture design of the data warehouse
In business intelligence, the construction of data warehouse is also one of the cores, which plays a role of linking the past and the future. It undertakes data from various data sources downward and supports various visual analysis reports upward. The construction level of data warehouse directly affects business intelligence. The overall quality of BI.
The final step is the design of the visual analysis report
Including: the structural design, logic design, UI effect, user experience, etc. of the visual table are also a subject of knowledge.
The visual analysis report is determined according to the index system built earlier, and the layout is determined according to the key indicators. Generally, the main indicators are placed in the C position, and the analysis form is designed according to the characteristics of the indicators. For example, the periodic analysis indicators use line graphs, and the structural analysis of indicators Use stacked or pie charts. Commonly used visual analysis is divided into five categories: data templates, fixed reports, multi-dimensional query analysis of indicators, detailed data, and decision-making dashboards.
● Data template: Its advantage is that it responds quickly to needs, does not require technical personnel to develop, and is not limited by the online time. Users can customize the data template, and share and run the data template with relevant users to meet business needs.
● Fixed report: It is characterized by fixed format and fixed requirements, which can be run after selecting different parameters and can be used for a long time.
● Index multi-dimensional query: Unified definition of caliber, unified data processing, business personnel can use data flexibly and independently, can drill down to view detailed data, and support various statistical methods, can quickly respond to needs, and improve the efficiency of data reporting.
● Detailed data: supports detailed data query and secondary analysis, that is, further statistical analysis based on the query results, such as grouping, filtering, sorting, etc. The detailed data display methods are also various, including charts, cross tables, etc.
● Dashboard: It is the leadership cockpit or decision-making cockpit, displaying KPI indicators, such as operation management, risk dynamics, etc., to support management decision-making.
04
summary
In this article, it mainly explains the definition of BI, and the method of business intelligence, and also briefly talks about the five core steps of business intelligence.
As one of the earliest BI manufacturers in China, Yixin Huachen has been deeply involved in the field of business intelligence for 17 years. Yixin ABI integrates data source adaptation, data warehouse, data modeling, data analysis, data reporting, and workflow A one-stop data processing and analysis platform built with core functions such as , portals, and mobile applications. It can not only provide data warehouse-oriented data analysis display for implementation personnel, but also provide self-service data analysis capabilities for business personnel, fully satisfying users' data application scenarios.
We have always maintained our original intention, aiming to provide truly valuable data services for enterprise users, and steadily polish our products. Therefore, if you have any questions about BI, please feel free to private message or leave a message in the comment area.