Dispensing With FailureBy Samuel Greengard | Posted 2009-09-09 Email Print
The desire to put data to work in new, innovative ways has led NCR down a path of predictive analytics.
Dispensing With Failure
Sifting through mountains of data to find patterns and trends is no simple task. In fact, only a few years ago, the process was almost unimaginable. But advances in networking technology, remote diagnostics, shared databases and the algorithms used to run analytics software have taken a giant leap forward. Taking advantage of these advances, NCR began using next-generation monitoring and analytics systems a decade ago.
Over the past several years, the 125-year-old company has focused on moving from a reactive environment—one where it dispatches technicians to sites only after receiving repair calls—to one in which NCR increasingly makes repairs, updates and changes to systems before a problem occurs. It monitors data streaming in from units in the field, as well as from Websites, call centers, help desks and the BlackBerrys technicians carry with them.
The company uses this data to determine when hardware or software requires a fix. It also uses remote diagnostics software to facilitate repairs on systems that are located across town or on the other side of the world.
In addition, NCR Interactive Insight software—developed internally—utilizes advanced data modeling and leverages the company’s vast data store to mine variables such as mean time be-fore failure, installation date, tally counts and repair history so that it can predict service actions.
“If we’re aware that a mechanism typically fails after 10,000 times, we can schedule a repair at 9,900 transactions,” Wallace points out. “That means a customer doesn’t wind up frustrated by a machine that’s not working, and we are able to manage staffing better.”
NCR also gauges customer be-havior—using commercially available business intelligence products from Microsoft and Business Objects along with its proprietary software and algorithms—to better understand how individuals use ATMs, kiosks and self-service checkouts. This data streams into a 24TB Teradata data warehouse via an Oracle E-Business Suite and NCR’s APTRA financial systems management platform.
The company also put translation layers in place to tackle Extract, Transform, Load (ETL) functions. The data conversions occur anywhere from three times a day to near real time.
The result? Substantial gains for NCR customers. For example, a U.K.-based bank with more than 40 million customers worldwide used recommendations developed through the use of NCR’s Interactive Insight application to improve the availability of more than 6,000 ATMs. Through the proactive analysis of service data, the bank trimmed service incidents by 17 percent year over year. The bottom line? A 50 percent reduction in hardware downtime (from 0.8 percent to 0.4 percent) over the same period.
And a retail customer witnessed a40 percent decline in hardware downtime at its self-checkout lanes through the use of analytics. NCR introduced annual preventive maintenance activities that targeted the actual causes of downtime, and delivered an engineering redesign on a component that had a higher-than-expected failure rate.
Finally, NCR reviewed data for specific stores in order to identify sites with higher-than-normal equipment failures. This resulted in additional training so that employees could oversee equipment and avert wear and damage.