FleetBoston Adds Recognition to Image Processing

When BankBoston merged with Fleet two years ago, ATM deposit volume at its Massachusetts processing site grew by 275 percent overnight. The volume explosion launched the search for a replacement to the institution's outgrown image-based system.

Today, FleetBoston Financial (Boston), with assets of $200 billion, processes a daily average of 300,000 ATM deposits with a new system developed by Dayton, Ohio-based NCR. Using recognition technology, it streamlines the keying and matching of deposit items. Not only has it reduced labor costs by 50 percent, but it also will position the bank for the post-Check 21 era.

"We only have to key about 20 percent of the items now, instead of 100 percent with the old system," said Ralph Catalano, deposit processing executive at FleetBoston. "We've picked up hours per night," he said. "On our heaviest night, we are easily finishing five hours earlier. It's making a difference in the number of dollars we can collect a day earlier versus the old process."

When the bank began its search for a replacement, there weren't many solutions in the marketplace. NCR was able to modify its Imagemark Capture platform to add ATM functionality. "We actually worked on that with them," said Catalano. "I suppose we were the alpha site, not even the beta site."

According to Joe Kniceley, vice president of NCR's Payment and Imaging Solutions group for the Americas region, 85 percent of the solution already existed. "One of the things Fleet wanted to do was keep the transaction integrity together," he said. "So we do image capture of the envelope, which helps us tie the data that's entered and captured at the ATM. "

FleetBoston began piloting the system last December. By the end of June, ramp-up was complete at its three processing sites in Massachusetts, Connecticut and New Jersey. Catalano said the bank was able to utilize much of its existing hardware, adding just a few more of NCR's model 7780 transports, which the bank was already using in conjunction with its old software.

Today, much of the processes look very much the same. ATM check deposit envelopes are delivered to the processing sites, where they are opened and put into batches. The envelopes and checks are run through the transports on a capture pass. Under the new process, in addition to capturing all the microline information, the system also attempts to read the amounts on the checks.

The checks are then compared to the ATM data feed. "That automated recognition process is the biggest single change in the system," said Catalano. "If we get a match on the deposit amount and the total of the checks, that recognition process allows us to avoid keying those items and also avoid a balancing step later on."

In the old process, after doing the capture pass, operators would look at the check images and key the amount for every single check, in addition to repairing microline data. Then, they would have to go through a balancing step for all of them. "In the NCR process, we avoid keying 80 percent of the items and also doing a bunch of the balancing because they're all proofed automatically," said Catalano. The remaining 20 percent that can't be read due to poor handwriting, smudges or busy check backgrounds are hand keyed.

Then, the processes become similar again. The items are loaded back into the transports and automatically coded with the amounts that have been keyed and captured. "In the NCR process, you get to that encoding pass much faster than you did under the old process," said Catalano. "There's a significant cycle time reduction."

Catalano estimates the system will pay for itself easily within two years. Though the project was primarily justified on reduced labor costs, he also expects to realize float savings, which wasn't used in the initial return-on-investment calculation.

"We saw through the financial analysis that it could stand on its own just as an ATM deposit-processing solution," said Catalano. "But it also provides us a foundation we can build on for other image processing applications going forward."

Fast ROI

Those future projects potentially include processing the bank's regular proof-of-deposit work that comes in from its 1,500 branches, truncation in the branches at the teller windows, and using the system for back-end processing and balancing of its high-speed work. "All those are yet to come, but are possibilities with this product," said Catalano.

In addition, because of the modular nature of the system, the bank is looking at remote capture. "We have the option...of putting some smaller devices into our busier or remote branches or even to deploy them to some of our higher-volume customers," said Catalano. "That would enable us to get a jump on processing before the checks arrive. Once you get to the Check 21 world, we wouldn't have to move those checks at all."

---

Fast Facts

INSTITUTION: FleetBoston Financial (Boston)

ASSETS: $200 billion

BUSINESS CHALLENGE: Develop a new deposit processing system to efficiently handle 300,000 daily ATM deposits.

SOLUTION: A modified NCR (Dayton, OH) Imagemark Capture platform.

KEY QUOTE: "We only have to key about 20 percent of the items now, instead of 100 percent with the old system." - Ralph Catalano, Deposit Processing Executive, FleetBoston