BI — What is it?
BI is business intelligence. It is the art and science of getting computers to know and think in ways humans do not do well, but need to know in order to better run their businesses. It involves the capture, management and presentation of structured data for use by decision makers. These data are usually organized into multidimensional hierarchies of key performance indicators. The aims of Business Intelligence systems are decision support and performance management.
Business Intelligence is my preferred term for this activity because it is the broadest, but various aspects of BI have gone as
Data Warehousing
Decision Support
Analytics
Big Data
Data Science
The origins of these practices are found in the lessons of Operations Management in the growth of multinational corporations during the 1960s, 70s and 80s. Some thought leaders that led to the creation of this multidisciplinary field include the following:
PECCC
The benchmarks for decision support and performance management are as follows:
Productivity: Increase output.
Efficiency: Reduce costs.
Competence: Add capability.
Confidence: Improve satisfaction.
Communication: Deliver consistency.
Business Intelligence systems are all about improving one’s ability to know, and helping one assess the impact of the changes made because of that knowledge. It is cyclical, so that new discoveries are added and the consequences of those results are measured as well. Business Intelligence systems enable a framework of iteration to quality and maintenance of compliance.
Application Scope
BI systems include a selection of IT tools, technologies and practices arranged for the delivery of applications. Each application should be implemented as an independent project which is determined by the aegis and impact of the key decision makers for the company’s operational goal. One should ask, who is responsible for the execution of X, what do they need to know, how often do they need to know it? A proper BI application delivers actionable data. You cannot manage what you cannot measure. A proper BI application delivers the knowledge necessary to improve a specific area of the business.
The Four Pillars
There are four areas of expertise for a BI practitioner.
Front End
Back End
Governance
Inference
The Back End — Architecture & Process
Back end work is primarily about data management, and data modeling. There are two primary sort of tools one uses in the back end. The first is an ETL tool. The second is a database. ETL stands for extract, transform and load. It is a tool or a suite of technologies used to identify data sources, gather them up, translate them into a standard format and load them into a database. A database is a tool that has a store of data and gives it a particular structure so that it can be queried in part or in its entirety. The database makes the data ‘live’. The way the data in the database is stored is called a model. The most important task a BI practitioner can have is designing a proper data model, because this determines the ease and speed with which the data becomes accessible for many years to come. When data stored in a database is not easily or readily available we call it a ‘data jail’. All data that serves no purpose to humans is, relatively speaking in a data jail. The job of a BI practitioner is to free data from that jail so it can be part of productive society.
The Front End — Visualization & User Experience
Front end work is all about presentation. It is the arrangement and display of data to the end user in a compelling, accessible manner so that the meaning of the data can be readily communicated. A front end designer is responsible for the user’s experience in navigating and interpreting the data. There are a wide number of graphical tools that are used in front ends. They are generally, tables and charts. But they are also alerts. They can be text, graphics, sounds and multimedia. They can be presented online via desktops and mobile devices or offline to paper reports. They can be delivered interactively or according to schedules. The front end designer must know his audience and insure that users do not become confused or frustrated when accessing data. Data presentation must be precise and crisp and encourage habitual consumption.
Governance — Security & Lifecycle
Governance work is about the meta process of business intelligence delivery. It starts with metadata management by considering the provenance and security of the data over the lifespan of the application. Who is authorized to use this data? Who is authorized to change this data? What are alternative sources of the data and how are they reconciled? How long will the data be live, when and where will it be archived? When and how must it be destroyed? What happens when something changes? Governance also can include resolving questions about the cost of maintaining the application and the scope of technologies used to deliver it with regards to upgrades and deprecations.
Inference — Statistics & Practices
Inference is the term I use to describe ‘data science’. This is a new aspect of Business Intelligence that manages the interpretive aspects of data itself. It is necessary because the scale of BI application datasets have grown exponentially. So particular care needs be taken when imputations and complex composite metrics are made over these large sets. For certain metrics, direct interpretation is straightforward. How many cars? “159”. For others more care must be taken. How much do people like those cars? “3.5”. A BI practitioner in this specialization assures that the data is taken only as seriously as it can be, and helps balance decision makers’ instincts and understanding against what the computer system is saying. They must understand the business goals and the application of statistical analysis to management. They must always draw the clear distinction between the map and the territory.
All of these areas of expertise must be engaged for a holistic integration of a BI system into a business process of continuous quality improvement.
Originally published at cobb.typepad.com on February 2, 2015.