In what may be a first, Bank of America Corp. is using optical character recognition technology to verify that others have encoded check payment amounts correctly.
Carreker Corp. of Dallas. developed the system for the Charlotte, N.C., company, which started testing it in December at a check-capture facility that processes about 400,000 checks a day. About one check in 5,000 turned out to be miscoded, said John E. Quinn, the banking companys transaction services executive.
It is a historic problem for banks, he said, and Bank of America wanted to catch errors on that check before they become errors in our process.
Last week it said it would install the system, called AEV for amount encoding verification, at its 21 other check-capture facilities in the next several months.
AEV scans check images to compare the courtesy amount written in a checks number box with the magnetic ink character recognition code that banks and some of their corporate customers use to speed processing.
That coding is done by clerks and sometimes they make mistakes. When the Carreker system spots what looks like one, it flags the check image so a bank employee can review it and correct it.
Bank of America had already adopted check imaging technology for other reasons, Mr. Quinn said, but, once you have the image of a check, there are a lot of things you can do with it.
Laura Fahey, who led the programmers who wrote the software, said Carreker knows of no one else using character-recognition technology to back-check MICR-line data.
One big challenge, she said, was to avoid flagging actually correct coding as incorrect and thereby generate a huge and flawed suspect list.
Bank of America does not plan to use the new system for coding checks, Mr. Quinn said. The company seems to be satisfied with its quality control in that area.
Error rates vary by bank and customer, Mr. Quinn said, and Bank of America will be looking especially at checks from places we think have the biggest error opportunity.
For a company that handles tens of millions of checks daily, even a tiny error rate compounds quickly, said Joe Rowell, the president of Carrekers Global Technology Solutions division. (Carreker bought Check Solutions Co. of Memphis, which he founded, last June.)
Most checks nowadays dont get looked at by humans, Mr. Rowell said. For banks, a typo that turns a $100 check into a $1,000 processing item means a customer-satisfaction problem and a lot of egg on their face even though it wasnt their problem and they processed it correctly.
We have a number of customers who are very interested in adopting AEV, Mr. Rowell said. For banks asking How can I justify the expense of getting on an image platform? he said, heres another example of how it can help you.
Mr. Quinn said the idea came up through brainstorming with Carreker. The innovation came from us being very passionate about minimizing errors, he said. The technology will help in reaching Six Sigma quality-service goals and should reduce customer complaints about errors in checking statements, Mr. Quinn he said.
Mr. Rowell said Carreker had already thought of using character recognition to confirm MICR coding This was something we in our own shop had analyzed and said, We can do this kind of thing, he said. But we had nobody who was willing to step up and say, We have this problem.
Last month Carreker announc-ed another check-image product using character recognition. It said Mellon Financial Corp. was using its Payee Name Verification software to detect check fraud. The software matches the payee line on corporate checks against the positive pay file that commercial customers provide the bank.
And on Monday, Carreker said it had sold nearly 1.3 million shares of stock in a private placement, netting $9.3 million in new equity capital after fees and expenses. The per-share price was $7.83; Carreker shares were trading on the Nasdaq market at $9.10 early Tuesday afternoon.
From Our Archive
