BI City

BI City >>  Expert  >>  

Abhishek Jee

"Business Intelligence Practices-Path to BI Success "

Abhishek Jee
Director e-Governance & Green IT
Abhishek Jee, Information technology professional with a passion for transforming data into knowledge through the use of Business Intelligence & Data Management Techniques. Strategic Information Architect, a thought leader with 20+ years experience with IT Consultancy Organization and Corporate Technology. He is currently working as Director e-Governance & Green IT with NASSCOM.

Business Intelligence as we all understands is the platform that allows the ability to deliver consistent, accurate and timely information to support corporate performance management and decision making process within an organization.
Over the past two decades, Business Intelligence (BI) & Data Warehousing (DWH) technology has evolved a lot. We, BI/DWH practitioners, are constantly being asked to deliver BI projects amidst numerous challenges of being time consuming, high on development costs, failing to address critical needs of business and assuring any form of guarantee on providing the desired outcome.  Additionally, with evolution of new trends in Business Intelligence, the desire and expectation from BI has grown manifolds. There is no mantra for ensuring your success with BI deployment, this article attempts in describing some of the practices that has worked and helped many organizations in realizing their benefit from BI Implementation and path to BI success.

Creating BI/DWH Blueprint

First and foremost is the availability of the BI/DWH Architecture Blueprint which encompasses business architecture, data architecture, technical architecture and implementation view.  A need for strong architecture base at enterprise level is must in order to deliver changes and enhancements in shortest possible time, an ability to build upon layers rather than starting from scratch. In next paragraph, I will touch upon various tools that integrate all your components.
Every business in this world has two key elements Business Entity & Business Process, all that we have to ensure is that we have good Metadata Management Tool which stores complete detail of all entities, their attributes & details of business processes. For storage and management of master data, data sources and mapping of business data with technology data, we need to ensure a good Reference Data Management Tool or Master Data Management Tool. The data modeling is most important component for success of Business Intelligence; it entails data modeling of Operational Data Store, Date Warehouse, Data Marts, etc. Specifically, in reference to Business Intelligence requirement the Dimensional Data Modeling offers flexible and scalable data models allowing you to add dimensions, attributes, facts and metric seamlessly. Now a days for every Industry & common business service area like Finance, HR, Sales & Order, etc. we have readymade packaged data models configurable to your business needs. Earlier, we had these applications running on Mainframes and now they have migrated to ERP systems.
Most leading BI vendors such as Cognos, SPSS and SAS and more horizontal solution providers such as Teradata, SAP and Oracle have developed integration solutions into ERP systems as well as other line of business applications such as call center, marketing and revenue assurance, thereby enabling BI to reap more benefits from ERP Systems, its data and workflows. In case your organization is running an ERP application then it makes sense to integrate with BI solution offered by the same vendor, it will foster tighter integration between your technology department and rest of the world. Irrespective of packaged solution from BI vendor or in-house developed data model, an Extract, Transform & Load (ETL) Tool plays an extremely important role. While selecting tool for your organization, please make sure you look into other requirements such as automation of jobs, automation of raising severity tickets in case of failure and notification by email/alerts, automation of workflow to kick start next job upon successful completion of last job, data load logs, parallel data loads, resume from failure points, incremental & full load, configurable mapping functionalities of source to target, etc. These are important factors while evaluating the ETL tool in order to keep peace post production rollout else managing voluminous data load will become nightmare. Next I will jump into Reporting & Analytical toolset as a separate practice area.

Enabling Role of Power Users

Empowering the business users to access information they need, on-demand, without impact to IT is the key to success in today’s Business Intelligence Implementation. We, BI/DWH practitioner, had all good intention to consolidate all data into single place as single source of truth but by restricting access to end users resulted in pain for business users and for them finding information in the Data Warehouse was like finding your favorite pair of socks in a huge bin of laundry. The business community felt that waiting for IT was a long and tedious process and at the end they were not getting what they wanted hence, they started building their own small functional data marts in silo. These functional silos has created new challenge for the current BI/DWH practitioner as how to consolidate these multiple data marts into single data warehouse which would mean similar problem will arise again after sometime. Therefore, the need of the hour is to provide BI platform as Business Intelligence Self Service as service that enables business users to access the information they need by themselves, using an easy to understand User Interface that is defined in business terms and not IT jargon. Perhaps, the most critical feature of such a service it to eliminate the “IT Touch” for the users to create new reports. The platform should support ability to create custom reports, ad-hoc reports, drill down capabilities, interactive & exploratory analysis, integration with metadata to enable online help on data fields, etc. These business users will form the community of “Power Users” Once power users get familiar with the platform and realize how easy it is to create and execute reports, they will start training other users and making them power users too who would then be using facts to make decision rather than their gut feeling; thus enabling the next level of enterprise maturity in business decision making process.
These Power Users may be drawn from traditional business team who are IT savvy and have experience with transactional system, excel formatting, formulas & macros, etc. Similarly the BI resource on technology side should be business savvy and understands how the business works. IT resources and business users must work closely together to ensure this divide is bridged.

Agile BI Implementation

One thing for sure is that "Big Bang" approach does not work for DWH/BI implementation. The rapidly changing business environment, an endless demand for changes & enhancement for any BI solution mandates for iterative development. At the beginning of any DWH/BI project, I recommend you to do some initial architecture modeling to identify a potential vision for how your team will build the data warehouse. The initial architecture views should entail some of the technologies you intend to use, as described above under DWH/BI Blueprint & Business Intelligence Self Service Platform, create a high-level domain model over-viewing the business entities and the relationships between them.
It is extremely important to note that requirements will change throughout the lifecycle of your project for a variety of reasons.  If you want to develop a solution which meets the needs of your stakeholders then you will need to take an evolutionary (iterative and incremental) approach to development. As you move into iterative mode which implies a robust change management system for evaluating and prioritizing changes and development team to adopt evolutionary database development techniques such as database refactoring, evolutionary data modeling and database regression testing. As your requirement and development continues, you need to ensure that an independent testing team exists to perform user acceptance testing; functional testing, regression testing and performance testing are performed independently in parallel and feeding into construction. Lastly, you need to realize that operations and support staff play an equal important role in the whole implementation of DWH/BI therefore engage them sooner on the project. Several Business Intelligence Solution lacks close-the-loop capability, the close-the-loop refers to complete cycle in problem solving; detect the business problem – put together an action plan to solve it – implement the actions using DWH/BI solution – then measure the results, you need to ensure that you have put the close-the-loop for each iteration to achieve the maximum value from your agile BI implementation.

BI Governance:

The biggest problem with all Information Technology solutions, including Business Intelligence, has been how to manage the ever increasing complexity and difficulty in delivering real business values. It has been observed that a culture of distrust starts to exist between the business and technology sides of the organization resulting in business not allocating fund for technological growth and on the other hand, technology building solution in silos resulting in widening off the gap between business and technology further. Therefore, in the era of diminishing budgets without demand on services rendered by IT being reduced proportionately, the role of Governance is becoming critical.
The Business Intelligence Governance program has to look into nuts and bolts of how to establish selection and prioritization of BI project; it needs to be evaluated on criteria such as alignment to overall business strategy, business prioritization, ROI, budget, resource availability, resource skill-set and infrastructure facilities. A solid BI Governance process will establish proper change management, training policies for promotion & adoption of overall BI use, minimize fear & resistance of new technologies and drive right decision for infrastructure & technology.
The key step in establishing BI Governance process will be identifying structure of BI Governance team. This should include team from Business, Technology and CFO Organization. The business should be represented by one person from each department and they will be responsible for sponsoring particular projects for their department, explaining the benefits of these initiatives with respect to organization. The business impact assessment needs to be done on three parameters namely regulatory, brand loss and/or revenue loss. The Technology team should be represented by Project Manager, Architect, Data Modeler Lead, Data Integration Lead and BI-Front Lead as they can jointly assess the technical feasibility of each request. A matrix analysis of Business Impact versus Technical Feasibility should be conducted for selection of projects. Lastly, the BI Governance team should have representation from CFO Office. As BI project are considered strategic and implementation cost can easily escalate to high value therefore, executive representation from CFO office will play crucial role in facilitating the prioritization of projects.