Embedding business intelligence (BI) into operational processes can help workers make better decisions, but such IT initiatives must be focused on an area that will show a significant return on investment and be designed with business users.
That was the message in New York Wednesday from several enterprise users at IDC's Business Intelligence and Business Process Forum who have launched projects to infuse business processes with analytics to allow front-line workers to use exception handling to make decisions.
At Capital One Financial, for example, these intelligent process or operational BI automation projects have been focused on high-level corporate strategy areas such as lean operating initiatives and outsourcing, said David Hummelberg, managing vice president of IT at the McLean, Va.-based company.
"With BI, the net result of the value proposition is pretty foggy going into it," he said. "That is why ours have been drafted behind a specific business issue."
While the company has outsourced the majority of its call centers, it uses intelligent process automation to route calls to between 10,000 and 15,000 call center representatives based on capacity and volume, Hummelberg said. As a result, it has trimmed the operational costs of its centers by 12 percent year over year for the past several years while boosting customer satisfaction, he said.
The company has also placed a call center application on worker desktops that uses analytics to lead them through cross-selling opportunities based on the demographics and previous buying patterns of callers, he said.
Wayne Sipperly, manager of energy markets analysis at the New York Power Authority, said enterprises must focus these projects in areas where they can show results to win executive support. The authority has melded analytics and a rules engine with multiple transactional data sources to provide dashboards for its traders, asset managers and executives. Since going into production in February 2005, the project has saved traders an hour a day each, yielding US$500,000 to US$1 million in annual savings, he said.
"You can't try to sell it for human resources [projects]," he said. "You have to sell it where the dollars are."
The portal gives traders a dashboard with the real-time information they need to prepare 1,100 daily bids to buy electricity, which the company makes using 26 different sets of data from 11 sources. Before using the dashboards, "traders would end up with two screens and 20 windows open," Sipperly said.
"There were no alerts, no trending capabilities," he said. "There was a heavy reliance on manual spreadsheets."
The dashboards provide analytics so traders know the optimal trading strategies and can perform "what if" analyses. Dashboards can also alert users by e-mail in real time if the company's market position falls below a certain threshold for financial performance or plant commitments, according to Sipperly.
The project was not without its challenges, however, including the daunting task of convincing users to give up their beloved spreadsheets for automated systems.
In addition to homing in on a visible ROI, efforts to meld BI with business processes must be developed in close partnership with business users, said Rick Broughton, director of IT strategy at Dunkin' Brands.
Dunkin' Brands, which operates 12,000 Dunkin' Donuts and Baskin-Robbins stores, is replacing its manual system of collecting retail data with one that ties BI data to business processes. So far, 300 users have been trained on the new system, with another 200 to be trained, Broughton said. Business users must be tapped as data stewards for the projects because "they are the ones who know what the best data is and the source for it," he said. In addition, other business users must function as "BI evangelists" to convince users of the merits of the system, Broughton said.
One of the biggest challenges of these projects can be keeping the users focused on "practical business processes," said Steve Phillips, CIO of Avnet. For example, while access to real-time information is often associated with these projects, if data "isn't needed in real time, it can cause unnecessary work," he said.
The Phoenix-based manufacturer of semiconductors and computer equipment is doing gross-margin checks when its sales workers give price quotes to block orders that don't meet a certain threshold, Phillips said. Those orders are kicked up to executives to be reviewed before being processed.