Any successful and growing organization would have an ecosystem of OLTP applications. At some point in time, organizations feel the need to be equipped with business intelligence (BI) software.
However, it’s necessary to be clear about one’s business needs before selecting BI or any analytical software. Purchase made purely on vendor recommendation without knowing exactly what the tool delivers, can be detrimental to the project.
This article offers a lowdown on various BI and allied software technologies that fulfill specific organizational requirements. Know your business needs and choose the software intelligently.
1) Data intelligence
In OLTP applications or in legacy applications data is captured from users and stored in databases. Owing to lack of standardization in capturing user-inputs or / and allowing free-form textual information, data quality issues occur leading to discrepancies in reports.
Since the real problem is data quality, BI software cannot fix it. The solution is to exercise a data quality assessment and cleansing drive.
2) Reporting intelligence
BI software needs to report data to analyze it quickly for decision making. Operational reports containing data with a lot of details require manual analysis to help decision making. OLTP or OLAP-based operational reports would be the same from a user’s perspective.
Proper use of BI software, such as info-graphics and visualizations, can represent the voluminous information contained in operational reports and still provide insights. Thus, intelligence is required in the way data is reported, not in the way data is gathered or organized.
3) Integration intelligence
An enterprise having discrete business units would need integrated reporting. Some of the exercises needed for this are given below.
Master data management (MDM) - Typically in a systems integration program, MDM is one of the first steps towards solution design and development. Whenever any transaction spanning data from multiple related data-stores is carried out, some form of MDM is required. MDM provides intelligence to identify and organize master level business entities that form the root of the entire business model.
Data warehousing - A common data repository is required to warehouse all the transactions within the enterprise. For data warehousing, data is collected from a variety of internal and external source systems using ETL (extract - transform - load) solutions.
Having organization-wide data stationed in a strategically designed data model within a central repository places the organization in a position to use BI software and report data from any corner of the enterprise. Data warehouse is generally expected to provide a single version of data after standardizing and integrating data from the entire enterprise.
4) Business intelligence:
Having its data positioned in a centralized data center, the organization is now in a position to meet its reporting and intelligence extraction needs. Mentioned below are some of the broad categories of business needs leading to the deployment of BI software.
Management information system (MIS) – ‘BI dashboards’ is a simplified term for MIS. Tactical and strategic dashboards are generally used by senior management to measure and monitor business health through a set of parameters, known as key performance indicators (KPIs).
Data Marts are compounds created from data warehouses; data required for MIS reporting is modulated and stored in data marts. Using MIS, organizations can keep continuous check on the health of the business, analyze trends and root-causes, extract intelligence and make correct decisions.
Complex event processing - Many a time organizations need to keep a tab on the pulse of the business when the nature of operations is volatile. For this, the software that the BI system equips them with can be utilized. Organizations need to identify business trends in real time, and provide decision-support accordingly.
Data discovery - Measuring and monitoring of business parameters is carried out on parameters identified as significant by business stakeholders. Based on scientific and analytical visualizations created by the BI software, it’s possible to extract intelligence.
These may need more than a regular set of business analysis methodology. For what-if analysis, volumes of data need to be manipulated on-the-fly to simulate situations which business stakeholders foresee and would like to assess. Data discovery empowers intelligence to be discovered rather than be extracted.
Predictive analysis - In a predictive analysis model, the BI software would be trained with past data of a few years based on the nature of prediction required and the scientific algorithm used for the projected time period. An outgrowth of data mining, it predicts results helping businesses to take corrective actions proactively to mitigate risks and to improvise policies whenever required.
About the author: Siddharth Mehta works as an associate manager and a technical architect for BI software projects at Accenture Services. He is known for his writing in the field of Microsoft BI software. He has won Microsoft Most Valuable Professional award. Prior to Accenture, Mehta has worked at Capgemini with clients including Walt Disney Pictures and Television, CitiBank, Transport for London, and ABN AMRO. He can be reached at email@example.com