|
 |
 |
 |
|
|
Big Stakes
Business intelligence solutions don't come cheap, and they aren't plug-and-play either. A lot of homework has to be done before they yield meaningful results.
By Ashwani Mishra
A few years back, the core banking solution (CBS) used by Centurion Bank of Punjab (CBoP), one of the leading private sector banks in India, didn’t have any provision for recording the date of birth of the bank’s retail customers. Then in 2006 CBoP switched to a more advanced core banking platform and also decided to go in for a business intelligence (BI) implementation to meet various business objectives.
But after putting in a data warehousing architecture with new parameters, and consolidating the customer databases residing in various silo applications as well as all transaction data in the data warehouse, the bank noticed a snag—the birth dates of at least half of its 5.7 million customers were missing.
The absence of this data—traced back to the legacy CBS—made reconciliation and elimination of duplicate entries a mammoth problem. Says Sanjay Narkar, CTO, CBoP, “A major issue with all banks is legacy data. The quality of this data is poor. At the same time, 100 percent data quality is an unrealistic target.”
The good news is that post-consolidation, CBoP managed to reconcile data issues to the extent of 3.6 million customers. The remaining 2.1 million customers were discovered to be different entries posted by the same customers, but as gathering customer information is still in the happening phase, it will be a while before BI begins to deliver complete value.
It’s not only the financial sector that is grappling with data quality. Bajaj Auto still faces some challenges while collecting disparate data from various sources even after three years of SAP Business Information Warehouse (BW) deployment. “We still face the challenge of data inconsistency but it has improved to a great extent. Our SAP is an integrated ERP (enterprise resource planning) system where the data is collected at the point of transaction itself. Also, the material and financial accounting system enable data accuracy,” notes Rajib Kumar Jena, senior manager for MIS at the company.
Drivers of adoption BI applications have been around for nearly two decades, but of late they have caught the fancy of organizations which want to derive value out of their accumulated data as a result of automating their business processes through applications like ERP and customer relationship management (CRM). According to a Gartner forecast, the BI platform software market in India is expected to reach $50.8 million by 2011.
There is no doubt that BI has become more pervasive. Competition and compliance have been the primary drivers for its wide-scale adoption. BI is transcending organizational hierarchies and penetrating various levels of operational and tactical decision-making in the organization. It is also morphing from query and reporting tools into areas that include customer management, risk, and product management.
Yet data consistency seems a distant dream. Issues like name and address mismatch, missing fields and inconsistent data formats are some of the common pains in customer data integration across verticals. “No matter what you do, it is tough to address the issue of data management. The reason is the dynamism in business,” comments Sanjay Deshmukh, country manager, India and Saarc, Business Objects.
Many businesses today are on the path of inorganic growth. Apart from mergers and acquisitions, the addition of new applications, databases, regulations and access points for data only add to the information and hence create new challenges with regard to data quality.
Organizations forget this when they make a beeline for BI. They wrongly assume that having enterprise systems like ERP and CRM automatically ensures data quality and consistency. However, automation of business processes is a pre-requisite for BI implementation; it is a critical component of the overall BI infrastructure.
“For banks, BI wouldn’t make sense unless they have in place a robust core banking system that records transactions. Operational systems that have gone through the right maturity cycle are ideal candidates for BI. What matters is not the amount of information a user receives with BI but how much of exception reporting he gets with it,” says Sudipta K Sen, CEO and managing director of SAS.
The transactions that enterprise applications (such as ERP) record need to be analyzed for things like trends and performance improvement aspects, and this is where BI comes into play. It can help enterprises understand the return on income (RoI) from these deployments.
For example, with ERP, the time required to shift products out of the factory could have reduced by two hours, and the time for printing the delivery slip and ship the goods could have decreased by an hour. But have these yielded any business benefits for the CEO and business users?
“Anyone who has implemented ERP is struggling with information. The reporting system or information delivery system of any ERP is not good, and that’s why it is a compelling reason to invest in BI. This is because the information that business users need for decision making has still not reached them and lies in the ERP,” explains Deshmukh.
Once these applications get stabilized users start looking for data that is cross-functional because they want to have a holistic view of the data that resides across all the applications in the enterprise. That’s where the other equally important component of BI infrastructure—data warehousing—comes into play.
“Data warehousing is a better option because if a user has SAP he will query the SAP database directly, or if he has an Oracle he will query the Oracle database. This creates problems because you are upgrading and querying at the same time. Also, since you need one version of the truth, you do not want to reconcile the data between all the different applications that you have,” says Bhavish Sood, principal analyst, software markets, technology and service provider research, Gartner.
According to a Gartner report, 60-70 percent of the BI challenge is about cleansing the data, getting it out, transforming it correctly, and storing it in a properly designed warehouse.
“We are seeing very strong data warehouse adoption among all top SAP customers. At least 70-80 percent of them are having BW projects; if an organization has 500 ERP users the BW users could be around 2,000,” estimates Atul Sareen, VP, overlay sales, SAP India.
According to Sareen, most of SAP’s customers are looking at analyzing operational data gleaned from their ERP systems to do things like making intelligent decisions, checking out which customer or region is more profitable, performing simulations, and finding out the impact of a variable on the company’s top and bottom lines.
The prime driver for BI users is the gain that they realize by having a single coherent view of data across the organization. Also, with the nature of tools available today, most BI implementations find that users can easily perform their own analysis and essentially break actions down to simple decisions. Also, as interactions between organizations and their customers evolve, having a common view of historical data has become a necessity.
“Solution providers can build the best systems but if an organization hasn’t got the skill sets to analyze the data and execute on that analysis, and hasn’t got the business processes to utilize that analysis, it is ineffective,” says Dennis Samuel, area VP, South East Asia and India, Teradata (a division of NCR).
According to industry opinion, it is essential that organizations have a mature and stable IT system. There must be a need for analysis and the users must be prepared to adopt the system. Also, a proper online transaction processing system should be in place so that the given data can be pooled in systematically.
“However, the ability to scale BI operations across more users, and the latency of information available for BI-based analytics, is still a hindrance for many users. A large number of BI implementations still load their data once a week or once a month, which has a negative effect on decision making,” says Arun Ramachandran, director, technical, Sybase.
Also, it is imperative to have a dedicated analytics server to handle data warehousing requirements with ease for faster and optimized query performance. Another common challenge for enterprises is the lack of clarity of specification requirements and lack of ISVs skilled in implementing BI in a cost-effective way.
“Often there are changes in user requirements during the deployment phase. Incorporating these changes on the go needs a dynamic implementation of BI to fulfill user requirements,” says Pallavi Kathuria, director, server business group, Microsoft India.
As organizations extend BI tools to more users within the organization, they will also have to prepare the end-users to use these tools. What if a successful BI project delivered a churn rating for a customer but the input was not used by the concerned department within the organization? The value made by the correct delivery of information will never be realized, and hence deliver a negative return.
Jena recollects that getting users to use the information was another challenge which Bajaj Auto had to face. “Now we are in the process of creating personalized Web-based reporting irrespective of the levels in the company. As a result, instead of asking for reports, users can themselves access reports, select key performance indicators, define thresholds values, and drill down on issues.”
|
|
|
|
|
|
|
|
|
|
|
|
|
|