Since the epidemic, the top-level thinking mode of enterprise management has begun to change, and more and more attention is paid to data-driven business and data-based management, hoping to take the first step of data-based construction by relying on BI.

However, throughout the BI industry, BI tools of various manufacturers are also varied, and product logic and functional architecture are very different. BI selection has become the key to the landing of business intelligence.

Business intelligence BI has been mentioned for a long time, but it seems to be not sophisticated. It is only used for statistical data and kanban. Such a seemingly simple product, but IT personnel in product selection is often still “confused”. There are a variety of traditional BI, OLAP-style BI, Agile BI, Collaborative BI and cloud BI on the market. Easy to use such as Power BI, comprehensive practical such as soft sail, rich chart such as Tableau. It seems that every company has its own characteristics, such as Dugu Jiujian, broken sword, broken sword, broken gun, broken whip, each has its own strengths.

A wide variety of products are derived from the various application layer requirements, BI is undoubtedly the most partial application layer. Let’s list the three and five requirements that project managers might be more likely to empathize with.

**1. Business requirements: ** Strategy Department needs to make complex statements; The financial Department needs manual data replenishment; The operation department needs real-time data display; The user center needs BI to support multidimensional self-help analysis; Decision makers need to manage the cockpit; Sales reps need mobile presence…

**2. Technical requirements: ** needs to support real-time data processing; Need to support ETL secondary development; Need to support Echarts external chart plug-in docking; Need to support the rapid and efficient development of large screen and mobile applications……

The emerging BI trend in the market also seems to cause more confusion: should the mathematical algorithm platform be integrated in the BI layer? What is the value of enhanced learning to BI? And so on.

The left hand is a dazzling array of BI products, the right hand is a cloud of different BI needs, BI selection is really difficult. However, there should be a way to solve the selection dilemma of “one chicken feather”. We will match and sort out the requirements and product functions, and summarize them in four aspects:

1. Basic data preparation

Many Agile BI products boast that “Agile BI requires no complex data modeling, no Cube, business requirements don’t have to be passed on to THE IT department, business people need IT on demand”. However, no BI product should expose business people directly to raw data that is not governed. Any BI system based on “dirty and messy” data lacks practical application value. Therefore, no matter how the product logic of BI system evolves or changes, “basic data preparation” is a necessary module. The contents are as follows:

**1. Data connection capability: ** Most BI systems support the mainstream relational database, distributed database and text data source at the same time. Based on specific business requirements, specific attention should be paid to whether the BI system of the solution provider can support real-time data docking and its ability to support API interfaces.

**2.ETL capability: **ETL engine is the basis of data governance and data consolidation. The ETL capabilities of BI systems on the market are divided into three types: first, three-party ETL tools such as Kettle and Kafka are adopted; Second, we independently develop simple and basic ETL tools to support a few common ETL logical calculation functions. Third, we independently develop a comprehensive and complete ETL tool, which encapsulates 50-100 kinds of logical computing functions. However, it is not necessary for us to purchase BI products with complete and comprehensive ETL. In the process of building big data system, most ETL work should be completed at the level of data platform, while BI system only needs to undertake some simple ETL work, such as table merging, association, grouping, and weight removal. When an enterprise lacks a complete data platform system and has many data quality problems, it is recommended to choose BI products with complete ETL functions.

**3. Data modeling: ** OLAP-based BI emphasizes the establishment of multidimensional data model to help analysts explore and analyze data in a multi-dimensional manner, and to ensure efficient analysis by creating cubes for data prediction, but this sacrifices real-time performance. Agile BI integrates the technical architecture of big data and focuses on real-time analysis of data through distributed computing, in-memory computing, column storage, in-library computing and other technologies, so as to achieve the second-level response of hundred-million-level data. Therefore, it eliminates Cube and only needs slight modeling to meet the requirements of exploratory analysis and self-service analysis.

**4. Data mart: ** For the consolidated and modeled data, it is recommended to present it clearly and systematically in the form of data mart, which is well reflected in Agile BI.

Second, fixed report development

Requirements such as BI kanban, management cockpit, and mobile visualization are essentially fixed report development and are required by most enterprises. The development of fixed reports is mainly oriented to technology, which pays more attention to the standardization, high efficiency, the richness of plug-ins and materials, and the support ability of secondary development.

**1. Chart development capability: ** The richness of chart types, support for 3D visualization charts, map integration and optimization capabilities, chart expansion capabilities (such as integration of Echarts charts) are important indicators to evaluate the chart development capability of BI products.

**2. Chinese-style complex statements: ** strategic department, financial department and other business departments may have more requirements for Chinese-style complex statements. If BI tools lack the ability to support complex statements, technical personnel are just “bricks without straw”. However, the performance of BI products in foreign countries is obviously weak in the development of complex statements. In China, Fansoft is prominent in this field. However, FineReport (statement) and FineBI (analytical BI) are currently two sets of products, but according to a party A who uses Fansoft, they can be technically integrated and data can be exchanged.

**3. Large screen design ability: ** Management cockpit is one of the main requirements of Large enterprises and institutions in China, and the cool large screen design often becomes the “finishing touch” of BI projects. At present, there are three main ways to design large screens on the market:

One is based on Echarts, Highcharts and other front-end visualization development tools. It is the slowest to develop and requires UI support.

The second is to use BI/ report system fixed sample table function module to achieve. Its development is fast and requires UI support.

Third, BI products directly provide cool screen designer, built-in a variety of COOL Html components, to provide a variety of UI skin. Its development efficiency is the fastest, the dependence on UI personnel is very low.

**4. Mobile application integration: ** can seamlessly connect with PC, mobile phone, tablet and other devices, can connect with Android, IOS and other operating systems, can integrate with wechat, Dingding, Feishu, small programs and so on. BI abroad often lacks the integration ability with domestic mainstream applications. After all, Domestic Internet products are ecological mature.

Sail soft mobile application

Self-service BI

In the era of big data, business departments’ demand for BI has shifted from traditional kanban and indicators to analyzing and mining operational problems from big data. As the need for instant, diverse data analysis becomes more widespread, self-service BI for business people becomes the trend. However, the business personnel lack the development ability, so how to make the business personnel can use, use well, and use self-service BI has become the focus of the enterprise. Self-help can be achieved in a number of ways:

**1. No code, improve usability: ** Typical representatives are Tableau, PowerBI and FineBI. These products are committed to “everyone is a data analyst”, and on the basis of no code, pull visualization, the “usability” to the extreme.

**2. Reduce modeling and improve agility: ** eliminates data modeling, CUBE and other redundant links that only technicians can understand. In the face of massive data, pull and pull at any time, while ensuring real-time response of big data. This is the self-help analysis approach advocated by BI.

**3. Enhance learning and improve intelligence: ** integrates the recommendation algorithm of artificial intelligence on the basis of BI to accurately recommend the chart style desired by users; Based on NLP and speech recognition technology, it can quickly understand the user’s intention through the way of search assistant, search and present the result that the user wants. Using AI to make BI silly has become a new trend in BI product development. However, relevant technologies are still in the early stage. Foreign head BI products have accumulated excellent AI capabilities, but they can only understand English, not Chinese, and can only be “irrelevant” in domestic application and promotion. Although a few BI manufacturers in China are also carrying out AI exploration, they are still in the research and development stage. It is still acceptable to be a “stunt”, but difficult to be elegant.

In short, all roads lead to the same result. Codelessness, agile BI, and enhanced learning are all paths to the Roman Empire of self-service analysis. BI products that can achieve the ultimate in a single point are worthy of attention.

Other functions

**1. Data filling: ** This is also one of the most common scenarios of the report system, mainly to solve the problem of how to fill in and supplement the manual data and ledger data of front-line business personnel, such as financial personnel, warehouse personnel, production management personnel and so on.

**2. Authority management: ** Data is the lifeblood of an enterprise, so how to do different fine-grained control of data is very important, which tests the management ability of BI decision-making platform. Fansoft BI is good at this, while FOREIGN BI is basically weak in this area, Tableau and Power BI are positioned as instrumental products.

• Fansoft BI can set permissions through roles, and permissions recipients include department, role, position and user

• You can set permissions for personnel management, directory permissions, management system, data connection, data permissions (data tables), sharing permissions, scheduling management permissions, and other permissions with rich granularity

• The permission setting object is deep, down to the component or data row level

**3. Mathematical algorithm: ** In enterprises, the demand for data science and data prediction is gradually increasing. Whether BI products should provide data algorithm platform capability has become a perplexing point in the selection of some enterprises. Personally, BI is closer to the application layer and business end, and it should have the ability of simple and basic mathematical algorithm to ensure that business personnel can make simple data prediction, especially the FMCG retail industry has a stronger demand for mathematical algorithm. But not everyone is a data scientist. Algorithm scenes with large scale and complex models are often completed by special algorithm engineers, while BI products are prone to collapse when dealing with tens of millions or billions of data. It is more appropriate for complete data science capabilities to be built in an independent algorithm center or integrated in the data center.

All in all, the ever-changing Dugu Jiu Jian seems to be able to achieve the unification of ten thousand swords, so we can form a relatively complete BI selection evaluation system. Selection is only the beginning of BI, and the promotion of BUSINESS intelligence relies more on the establishment of the operation index system. The goal of “everyone is a scientist” needs the cultivation and edification of “self-service analysis culture”. Massive data mining and analysis still follow the principle of “simplicity”. The ultimate goal of BI system construction is to “be able to use, use and use well”.

Therefore, in the process of BI selection, in addition to the tool product itself, we should also consider the manufacturer’s business plan ability and learning promotion service system. The more mature the top manufacturer is in this area, the richer the experience of the peer is for Party A. In short, we should take what we need and combine hard and soft!