Predictive analysis can pay off big — if you look at it right
- By Rutrell Yasin
- Oct 26, 2011
Analytics software is a core technology for helping agencies understand and improve their operations, but it requires a whole new way of thinking about how to measure data to make the right decisions, members of a panel of federal and industry representatives told a Washington audience Oct. 25. Even with all of the predictive data in hand, managers still might not make the right decisions, they said.
Agencies can use insight generated from consumers, citizens and individual transactions to determine whether programs are effective, Gartner analyst Rishi Sood said during a keynote address at a government symposium sponsored by Information Builders.
Organizations investing in business intelligence are getting a better return on investment than they are from some horizontal applications such as customer relationship management and enterprise resource management, he said. Organizations using analytic software to ferret out fraud, waste and abuse are reaping benefits. For example, “the Health and Human Services Department and the Center for Medicare and Medicaid Services, in particular, have been turning up hundreds of millions of dollars for the organization over the long term of analysis,” Sood said.
To survive tight IT budgets, focus on results, analyst says
How data wizardry can revive America’s cities
Agencies are starting to use analytic software to discover improper payments, another area associated with waste, fraud and abuse — not just on the back end when the payment is already made and then tracking it down. The software is being applied upfront before a payment is made to determine whether a payment to a doctor, landowner or whomever is justified. Agencies such as the Education Department and Social Security Administration are starting to realize dramatic returns on investment with respect to improper payments, Sood said.
Agencies historically have relied on business intelligence software to create dashboards to help senior managers better understand how their programs are running. With the information at hand, the managers can make concrete decisions on how to change program operations either midstream or at the end-of-year evaluation.
But now it is time to move toward more predictive analysis, with which the real changes can be made. “We have to get away from the dashboard only, which is a snapshot of what has happened, to push in the direction of what will happen,” Sood said. How changes are made to improve operational results down the road is the distinction being made between traditional and predictive analytics.
“The hard part is getting people to act on the information they have,” said Gary Winkler, former Program Executive Officer, Enterprise Information Systems for the Army, during a panel discussion on how agencies are achieving success in the face of adversity, moderated by Tom Temin of Federal News Radio.
Take the American Recovery and Reinvestment Act as an example. “Those of us who work in the federal environment know how long it takes to get a contract in place,” Winkler said, adding that federal money can’t be spent unless there is a contract on payroll. The idea that close to $800 billion would get into the economy within a year is crazy, he said; it would take up to a minimum of three years before that money started circulating in the economy.
“Getting leaders to act on predictive information is challenging," said Winkler, now founder and CEO of Cyber Solutions & Services. "There is a lot of information out there and plenty to make good decisions. Unfortunately, good decisions are not made even when there is near-perfect and predictive data."
The U.S. Postal Service is using analyzer technology hosted by a software company to gain better visibility into its transactions with banks, said Elizabeth Schafer, USPS' treasurer of bank relations.
Before this arrangement, 85 different districts paid bills separately in a paper-intensive process. “There was no visibility across the Postal Service if we needed to figure out what all of our banking costs were,” Schafer said. Now, all the banks send electronic files to the analyzer, which analyzes the data and lets the Postal Service determine whether it is being overcharged. The information also helps the agency predict budgets and look at past trends to help it predict what needs to be done to cover banking costs for the next fiscal year or five years down the road.
“We also use the information to compare banks on what they are charging and see if we are losing money on the table,” Schafer said. This allows the agency to make a solicitation for better pricing or negotiate individually with banks.
Analytics tools will require a whole different thought process on what to measure, said Dan Mintz, former CIO of the Transportation Department and now chief operating officer of Powertek, a systems integrator. For example, in the case of waste, fraud and abuse, organizations need to examine the factors that caused the fraud to occur instead of just measuring the fraud.
“It is a change in thought processes, in terms of thinking through what it is you are trying to measure,” Mintz said. “I think predictive analysis requires a different kind of thought process.”
Rutrell Yasin is is a freelance technology writer for GCN.