30 Minute Interview

BI for everyone

Soumendra Mohanty, Head of Business Intelligence / Data Warehousing Practice at Accenture India talked to Varun Aggarwal about the benefits of BI adoption and some industry best practices

Soumendra Mohanty

How does Business Intelligence help a company get closer to its customers?

Business Intelligence is all about measuring and monitoring the state of business. Customers are core to any business. BI helps to understand what customers want, how they are behaving, what their preferences are, why they are choosing to become customers or why they are churning. BI is the more essential currently, since today’s business is all about diversification and globalization.

Can you throw some light on the economic downturn and the need for a customer-centric approach on part of the companies?

At a broad level, the economic downturn is largely attributed to ‘not understanding your customer well’ and also not having robust data management and analytics capabilities.

While looking for expansive business opportunities is important for any business to grow, at the same time it is all the more important to have data governance and analytics capabilities institutionalized across lines of business within an enterprise. A siloed business view is dangerous as it gives a false impression about the state of the business, across the board. Companies should focus on KYC (know your customers) and KYT (know their transactions).

Another aspect, which has always been an afterthought, is Data Quality, any strategy or business insight is as good as the quality of the data the insight is drawn upon. In addition, I think, the intelligence that BI provides also needs to be pervasive, it should be available to the masses so that they can detect anomalies, and take timely actions to do the course corrections.

How are CIOs currently viewing Information Management and Business Intelligence?

CIO surveys done by Accenture have found that BI is the topmost priority for CIOs across the globe and quite rightly so. I would like to share a different perspective here, with the technology advantages, the increased efficiency in data distribution—both in terms of speed and bandwidth—inevitably has led to more data being sent and received, which in turn makes us act on more data and thereby more analysis and more often than we are used to do. The time between the moment when we receive the data and the moment when we decide to act or not to act on it has definitely also decreased. But has the quality of our decisions gotten better?

Perhaps the pure speed by which we make decisions nowadays actually increased the number of bad decisions that we make! If that is true, then it is likely to assume that what we gained at one end by having become more efficient in distributing and consuming data—more or less—got lost in the other end due to an increasing number of bad decisions.

This is extremely important. We cannot have analytics and BI kept exclusive for a few people, we need to make it pervasive across the enterprise. CIOs are baffled by the information overload or infoglut, and hence they are in search of newer ways to tame the issue. Information Management vendor consolidation has aggressively happened through last year, which is also driving the CIOs to choose between a few options.

Tell us about the best practices for a successful BI implementation?

There cannot be a predefined set of steps for a successful BI implementation. It is a journey and it requires commitment from the top to bottom. Today, one can find tons of best practices on BI implementation. Instead of repeating the same age-old rhetoric, I would like to share some of my conversations with clients, their expectations from BI and how some of these are challenged with changing times:

• We need one version of the truth: This was born out of fear for multiple versions of data across the enterprise. Consistent data across the enterprise is a must. However, the reason for data inconsistency is also largely attributed to the lack of proper metadata management. Understanding enterprise data and preserving the business context surrounding the data ensures that as long as there is one definition of data, one can have different version of the data per business context and still achieve a higher level of consistency.
• BI projects require a business driven approach: Any BI project or program should have business value to be delivered. Creating a business intelligence environment without a goal is a mission impossible. Businesses need solutions that can create more sales or reduce churn. A data warehouse or business intelligence project can support this goal. However, the notion that BI projects should always initiated by the business user is debatable. IT can play an important role in taking the lead by showing what is possible (prototypes, analysis models, sample dashboards, list of KPIs, etc) and helping the business user to finalizing the business requirements for the BI project.
• BI development should be done incrementally: What would be the opposite of this? Probably some kind of big bang approach where all the functionality is delivered in one single release. That sounds nice but often this is not possible for BI.
  To start with, when users start to interact with the new BI system they come up with new insights followed by change requests. A successful incremental approach focuses on the identification and prioritization of the most beneficial increments or slices. This should be based on the priority of the objective (is this a key process, is it in line with the business strategy?). Other qualifying criteria can be availability and quality of the data. Each slice should be a complete solution. BI projects—just as in real life—should have a first things first approach. Delivering incrementally, while keeping the long term view in mind, allows for faster speed to value.
• BI projects needs high-level sponsorship: This stems from the notion that BI is only meant for strategy formulation. However with changing business conditions and competing market dynamics means everybody across the enterprise is a recipient of BI’s delivered values in the form of actionable information. Therefore, it should be an organizational sponsorship and involvement all the way from the work floor up to the CXO levels.
• BI = data + reporting: BI enables better decision making. BI acts on data, turns the data into actionable insights and then provides a mechanism (publish, subscribe, automated alerts, etc) for the information to be consumed by the enterprise. It is up to the recipient of the information to make use of it. Therefore, BI is not the same as data + reporting only, it is actually data + reporting + enabler to achieve intelligent business.
• BI should be real time: It depends. BI enables businesses to make better decisions by making a strategy, process or initiative - accountable or agile. For accountability, BI need not be real time, because all you are doing is staying on top of business performance by looking at how it has performed through historical data, which is actually after the fact. For agility, BI needs to be real time, because businesses need to not only keep an eye on how the business is performing but also should be able to provide means to steer the course in the right direction, if anomalies are detected during day to day business operations.
• Converging Technologies and Information democracy: We are getting into a converging technology world. Not because it is driven by vendors to capture a larger pie of the market but because the need for information at all levels has increased exponentially. A few years ago, BI was all about numbers (facts), with the integration of unstructured data, concepts like text analytics and opinion mining emerged. The need for information at all levels of the workforce and business processes meant that information cannot be ring fenced to be used by a selected few, it has to be permeated across the enterprise and systems thereby information democracy and hence distributed through portals, content management solutions, etc.
• BI is the answer to everything: For every answer or insight provided by BI, there are a thousand more questions triggered by users. Data is being generated around the clock all around the globe, collection of this data and assured through a robust data quality assurance program can help users get answers to their questions.

How does predictive analysis help in improving business in this uncertain economic environment?

Becoming predictive is everyone’s wish, but it requires good data and as well as capabilities (skilled people, process, technology). With uncertainties abounding, every initiative and funding needs to be carefully evaluated in terms of predictability to achieve the desired results.

Earlier data intensive analytics was probably more or less confined to customer facing areas like retail banking, telecom, sales, marketing and customer relationships. However, the analysis was on the past performance meaning analyzing what has already happened, learning from there and then applying it to newer initiatives. What predictive analytics does is operates on the same set of data, but provides intelligence of what is going to happen next and to what extent the outcome is predictable. For example, a customer complaint can actually be turned around to cross selling or upselling by looking at the customer demographics, past behavior, affinity to some other offer, etc and all of these done at a real time while the customer rep is talking to the customer and a decision tree providing predictive analysis at every stage of conversation guiding the customer rep towards the favorable outcome.

What are you doing in order to target SMB customers who require easy-to-deploy and low-maintenance solutions?

IM/BI solutions have evolved over the last decade. There has been a tremendous focus on enriching BI solutions with a wide variety of functionalities like unstructured data integration, real time data integration and analysis, reporting solutions with a click of a button etc. While all these are good, there is also a lack of BI Jumpstarters meaning, SMBs who cannot afford to wait any longer and spend a lot of money to implement a full-fledged enterprise-wide BI solution are in a need of a solution that would provide them a platform to start quickly, cost effectively and then evolve over time.

We are focused on concepts like rapid data mart implementation, industry specific Data Warehousing data models, enterprise metrics management framework and data analytics offerings to help our clients gain lead-time in their BI journey and at the same time work iteratively to develop a full-fledged enterprise-wide implementation. Examples like BI in a Box for several industry segments like Insurance, Retail, Banking, Healthcare etc are focused on point solutions for specific business functions, and these are widely appreciated by our clients. We also feel it is important to have robust data governance and data quality offerings for our clients. We have developed several accelerators in these areas that helps implement robust solutions for our clients.