Upgrades aren't easy -- just ask the folks at ACS and MCI. Thousands of ATMs were affected after the companies tweaked their technology.by Ann All, editor
February 27, 2002
According to Affiliated Computer Services, a recent glitch in which thousands of ATMs experienced problems ranging from slowed transactions to incorrect account balances was an unfortunate byproduct of the conversion to a new operating platform.
In fact, said Joni Floyd, executive vice president of the Dallas-based company's Electronic Commerce Group, the problems were "typical of any kind of migration management or conversion effort."
While Floyd acknowledged that one of the reasons for ACS' switch to a new platform was Y2K compliance, she said, "It was one of about 10 reasons, and quite candidly not even at the top of the list."
Jim Welsh, an Oregon grocer, sees the issue a little differently.
Noting that the problems began at about the same time processors were obliged by the federal government to prove their systems were Y2K compliant, Welsh said, "They can deny it from now until doomsday, but this was a Y2K issue -- whether the particular part of the software that failed was date related or not."
At least two other major transaction processors, EDS and First Data Corp., also experienced problems while converting some of their customers to new, Y2K compliant platforms.
Floyd said ACS' new platform, which offers improved security, enhanced fraud detection and increased self-service banking opportunities, is one way of ensuring that the company remains competitive in the ever-changing world of electronic commerce. "You've got to upgrade and offer new products based on the market demand."
Testing, testing...
ACS began the conversion last January. The company had conducted integration testing on several different terminals, but Floyd said differences in machines, networks and telecommunications carriers made it difficult to identify every possible problem.
She estimates that the process was 75 percent complete prior to late June, when problems first surfaced. Certain terminals were affected, "based on device handler type and communications connectivity on the backside," she said, although service interruptions were sporadic.
Welsh, the owner of Manzanita Fresh Foods, leases an ATM from Evergreen Teller Services, a Grass Valley, Calif.-based ISO. Before he began noticing discrepancies between his ATM's electronic journals and his bank statements, Welsh never realized that Evergreen outsourced its transaction processing to ACS.
Welsh said he was met with "condescension" when he contacted ACS during the first week of July. Noting that he would be compelled to inform his customers of any potential problems with products at his store, he said that ACS largely ignored its responsibility to its customers.
"A number of people out there suffered some financial distress, but somebody at that company made the decision that it wasn't their problem because their corporate reputation was more important than doing the right thing," he said.
Floyd explained that ACS made every effort to contact distributors and ISOs, with the expectation that they would then contact retailers and other customers. While she thinks ACS did an "adequate job" of notification, she said the company intends to improve distributor communication efforts in the future.
"I've come to understand that you've got to cover all those bases, which means I need to learn how, from a business point of view, to communicate with all the endpoints," she said.
An Evergreen representative said, "It's in everybody's best interests, including the merchants, not to focus on what happened but how to restore service to customers." The representative said Evergreen had employees "working around the clock" to address any problems.
Another large ISO, Card Capture Services of Portland, Ore., apparently plans to reimburse some of its clients whose terminals went down, based on the average daily transactions per machine.
In a written statement, CCS President and CEO Dave Grano said, "Some of our customers were affected by ACS' installation of new software over the July fourth weekend. As a goodwill gesture, we are reimbursing those clients that were down due to the ACS problem."
ACS is gathering audit tapes and other materials from deployers to determine where adjustments are necessary. The company estimates that approximately 3 percent of its 20 million monthly transactions were affected, and that only about half of those will require adjustments. Floyd said that 80 to 85 percent of those adjustments have already been made.
"We're looking at a 30 to 45 day window to pull everything together and get it processed," she added.
Frame delay
Some ATMs also took a hit earlier this month when MCI WorldCom's frame relay network experienced 10 days of intermittent outages after a software upgrade. Hardest-hit areas included Chicago and New York. The data network, one of four operated by MCI, serves 3,000 customers.
MCI spokesperson Linda Laughlin said the company isn't sure how many ATMs were affected. "When we sell the service, we don't know which applications the customer may be using it for," she explained, noting that financial institutions use the network for payroll, email and file transfer, in addition to ATMs.
Kate Coleman, vice president of planning, marketing and sales for Cash Station, estimated that only about 1 to 2 percent of the Chicago-based EFT network's 7,000 ATMs experienced "sporadic, short-lived service interruptions."
"We never had any ATMs that were completely inoperable," Coleman added.
MCI was plagued by customer complaints that it reacted too slowly to the problem. One vocal critic was the Chicago Board of Trade, whose electronic trading system was stalled.
Service interruptions began Aug. 5. MCI took the network down for 24 hours on Aug. 14 to remove the problematic Lucent Technologies software. Laughlin said the company's sales representatives notified their customers on Aug. 13 after the decision was made. "As soon as that information was available, it went to the sales reps."
According to an article published in Inter@ctive Week, the company was also dogged by rumors that the glitch was due to a Y2K-related fix or integration operations between the MCI and WorldCom frame relay networks.
Laughlin denied the rumors. "Our network was already compliant. We had to have all of our mission critical systems in compliance by June 30," she said. "Our technicians and Lucent technicians are looking at the software, the installation and the equipment to find the root cause of the problem."
MCI said service is now back to normal. The company will compensate all 3,000 customers with two days of free service for every day the platform was affected.
"We're not asking customers to prove whether they were affected to get the credit," Laughlin said, adding that MCI's senior management is meeting with customers to "earn and regain their trust and confidence."