Business Intelligence has become one of the crucial competitive factors for many organisations. Hsinchun Chen, Roger H. L. Chiang and Veda C. Storey (2012) state in their article that a survey of the state of business analytics by Bloomberg Businessweek (2011) reported that 97% of companies with revenues exceeding $100 million used some form of business analytics.
1. What is Business Intelligence?
Matteo Golfarelli suggests Business Intelligence (BI) is the process of turning data into information and then into knowledge.
According to Microsoft, Business Intelligence (BI) uncovers insights to make strategic decisions. BI tools analyse historical and current data and present findings in visual formats.
There are four key steps BI follows to transform raw data into easy-to-digest insights:
Step 1: Collect and transform data from multiple sources
Step 2: Uncover trends and inconsistencies
Step 3: Use data visualisations to present things
Step 4: Take action on insights in real-time
According to Tableau, BI abides by business analytics, data mining, data visualisations, data tools and infrastructure, and best practices to help organisations to make more data-driven decisions.
There are several BI methods:
Data mining
Reporting
Performance metrics and benchmarking
Descriptive analytics
Querying
Statistical analysis
Data visualisations
Visual analysis
Data preparation
Developing a BI strategy requires us to:
Know your business strategy and goals
Identify key stakeholders
Choose a sponsor from key stakeholders
Choose your BI platform and tools
Create a BI team
Define your scope
Prepare your data infra
Define your goals and roadmap
According to IBM, BI is software ingesting business data and presenting it in user-friendly views, including reports, dashboards, charts and graphs. BI tools allow business users to access different types of data: historical and current, 3rd party and in-house, as well as semi-structured data and unstructured data like social media. BI allows institutions to ask questions in plain language and get answers they can understand. They can make decisions based on what their business data is telling them. BI may help organisations become data-driven enterprises, improve performance and gain competitive advantage. It can:
Improve ROI
Unravel customer behaviour, preferences and trends
Monitor business operations and fix or make improvements
Improve supply chain management
The best BI software supports this decision-making process by:
Connecting to various data systems and data sets
Providing deep analysis
Presenting answers in informative and compelling data visualisations
Enabling side-by-side comparisons
Providing drill-down, drill-up and drill-through features
Hugh J. Watson (2009) explains that, in the past, we developed the first computer applications for scientific purposes and transaction processing. Michael S. Scott Morton’s (1967) doctoral dissertation research paved the way for BI by building, implementing and testing a system to support planning for laundry equipment. His influential Sloan Management Review article (Gorry and Scott Morton 1971) and book (1971) further spread decision support concepts and coined the term – management decision systems. In the late 1960s and early-to-mid 1970s, we began to use Decision support systems (DSS) to describe decision support applications which became the name of this emerging field. The DSS field underwent tremendous changes over the years following technological advances, including the Internet and web-based applications, the collection, analysis and use of real-time data, dashboards/scorecards, data visualisation software etc. Howard Dresner coined the BI term in 1989. It is what we now know as our present BI and has become especially prevalent in industry and academia to describe all decision-support applications.
2. Present BI
Hugh J. Watson (2009) asserts not all current BI initiatives are the same. Some may focus on a single application or several applications, while others may aim at providing enterprise-wide BI. There are three particular BI targets (Watson 2006). The first target is the development of a single application or a few related BI applications. It is similar to the initial DSS that organisations created. It is often a point solution for a departmental need. We usually build a data mart to provide the necessary data. The second target is the creation of infrastructure supporting current and future BI needs. A critical component is an enterprise data warehouse. Senior management often provides sponsorship, approval and funding. The third target is organisational information to change how a company competes in the market. BI supports a new business model and enables the business strategy. Sponsorship, approval and funding start at the highest organisational levels.
Different BI targets will require different BI environments. We need data integration technology and processes to prepare the data for decision support use. There are a few components to this:
- Source systems: there is usually some “organisational pain” motivating BI. This pain can lead to information requirements, BI applications, and source system data requirements. We can use various source systems, including operational systems, ERPs, web data, 3rd party data, and more. The trend is to include more data types. These source systems often use different platforms and store data in various formats. We should use data profiling software at the beginning of a data warehousing project to understand the data better. We should also address other source systems challenges. There are multiple systems containing some of the same data. We should select the best system as the source. We should decide how granular the data needs to be. It is the universal wisdom that we best store data at a highly granular level because people will request it at some point.
- Data integration: we need to extract data from source systems, transform it and load it into a data mart in a warehouse. We call it ETL, but people increasingly use generic data integration because we can handle the source system data in an increasing number of ways. Hand-written code and commercial data integration software can perform data extraction. Many organisations opt for commercial use since it is easier.
- Storing the data: we can use many architecture and data models. One architecture originated from DSS, where we arrange the data around the app rather than treat it as an organisation-wide resource. While independent data marts might meet localised needs, they do not offer “a single version of truth” for the entire organisation. An enterprise data warehouse development begins with an enterprise analysis of data requirements. We pay special attention to building a scalable infrastructure for enterprise data warehouses. Using this enterprise view of data, we develop the architecture in an iterative manner, subject area by subject area. We store data in a warehouse in the third normal form and create dependent data marts to source data from the warehouse. Since the dependent data marts get their data from the warehouse, we maintain a single version of truth.
There are many potential BI users, including IT developers, Front line workers, analysts, information workers, managers and executives, suppliers, customers and regulators. Some users are information producers since they primarily create information for others. On the other hand, some users are information consumers, including managers and executives.
IT professionals and users must have metadata about the data in the marts/warehouses. We should create metadata during the development of the marts/warehouse rather than as an afterthought. We best maintain metadata in a centralised repository accessible by IT professionals and users. The poorest option is Excel. Another option is relying on the metadata storage capabilities of the ETL and BI tools.
BI tends to expose companies’ perpetuating data quality problems. The profiling of source systems typically uncovers a plethora of issues consisting of missing data, dummy values, multipurpose fields, and reused primary keys. We can address some of these problems by using data-cleaning software. However, the better long-term solution is correcting data quality problems at the source. It is difficult because business managers who are data owners must sometimes understand the value of spending resources to rectify problems. We must solve data quality problems. Otherwise, they will negatively affect the usefulness and credibility of BI.
Several understandings, principles and approaches are crucial to data quality. First, data in most organisations are strategic resources, and we need to treat it accordingly. We need to expose inferior practices and change human processes. We require plenty of training and education. We must put people and processes in place to deal with data quality. However, high data quality does not require zero defects or complete accuracy. It is relative to business needs and expectations. It needs to be accurate per its use. All of the necessary data should be available. It should be timely enough. It should be easily accessible, understandable and usable.
The scope of BI governance is broad. At the strategic level, it helps to ensure that we align our BI efforts with a company’s objectives and strategy. At the tactical level, it guarantees BI projects are on time and schedule. At the operational level, it answers the metadata and data quality issues (Watson, Fuller, and Ariyachandra 2004). We perform governance by multi-level, cross-functional committees and teams.
BI may provide us with several benefits: cost savings from data mart consolidation, time savings for data suppliers, time savings for users, more and better information, better decisions, improvement of business processes, and support for accomplishing strategic business objectives.
BI is constantly changing. Technology enhancements drive much of it. It could be because of business needs or better ways of doing things. Hardware and database software advancements have impacted BI scalability – handling more sizable amounts of data. Organisations can bring BI to more users and allow them to evaluate data in a new and more powerful way. A goal in many of them is making BI more pervasive to “democratise BI for the masses”. There are issues with pervasive BI, including time, cost, and complexity in considering, acquiring and implementing BI tools, training and supporting users. An associated problem is that many users may find the tools difficult to use on the underlying data and how we can use them in their jobs. Various solutions to these particular problems can be: using open-source BI software to reduce the software cost and decrease the time required to select and implement software. We can develop customised and web-based applications to solve the ease-of-use problem. Interactive dashboards and scorecards can help users and require little or no training. We can also design business processes with BI embedded to make BI part of the work system.
3.Beyond BI
Mihaela Muntean and Traian Surcel (2013) believe that agile BI can be the future of BI. Forrester Research defines agile BI as "an approach that combines processes, methodologies, organizational structure, tools and technologies that enable strategic, tactical and operational decision-makers to be more flexible and more responsive to the fast pace of changes to business and regulatory requirements". Agile BI "addresses a broad need to enable flexibility by accelerating the time it takes to deliver value with BI projects". It can include technology deployment options such as self-service BI, cloud-based BI, and data discovery dashboards that allow users to begin working with data more rapidly and adjust to changing needs. According to Data Warehousing Institute, an agile BI solution should enable access to accurate information in the correct format to the right person at the right time.
Hussain Al-Aqrabi, Lu Liu, RichardHill and Nick Antonopoulos (2019) speculate that cloud computing has instilled future horizons for BI. It is because there is a concern that BI will face a resource constraint circumstance due to the never-ending expansion of data warehouses and the online analytical processing (OLAP) demands on the underlying networking. Cloud BI solutions have recently gained prevalence among businesses since many are grasping data analytics advantages. Cloud BI is the idea of delivering BI capabilities as a service with cost efficiency, flexibility and scalability, reliability, enhanced data sharing capabilities and no capital expenditure. Cloud computing has the potential to offer a new lease of life to BI and OLAP frameworks.
Real-time BI is also worth mentioning. According to Daniela Ioana Sandu (2008), Operational BI has evolved into Real-time BI. The system analyses data as soon as it enters organisations. Real-time means delivering information from milliseconds to a few seconds after the business event as the latency has reduced to zero. A real-time BI system is dependent on a real-time data warehouse. It has increased refresh cycles to update the data more frequently, thus, achieving near real-time data updates. Operational and real-time BI optimises the decision-making process by eliminating latency. However, implementing real-time BI is exceedingly costly and not always necessary. The crucial matter is defining an optimal time frame, the right time for any decision process, and an interval which should reflect the business needs and offer the best risk-costs ratio.
Social BI has been very popular. We can collect social media content to analyse and process to obtain valuable knowledge. We can refer to social BI as the process of collecting social data and analysing it to make better decisions. Social data is not enough to make sound decisions, but it does provide grounded insights.
As BI has and will become more data-centric, addressing data challenges to ensure that the insights that we get are the most accurate, most up-to-date and meaningful is one of the most crucial problems. It starts with reviewing the quality of our data at the source. New future developments of BI may come from academia, where research of management science and operations or information systems may add original dimensions.
References:
Comentarios