Components A Partly Cloudy Future for ATMs?

 

The idea of managing ATMs through centralized, virtual systems is attracting adherents. But not all ATM makers are sold on it.

By Lauri Giesen

It seems the virtual world has touched nearly everyone these days, both in their professional and personal lives, with virtual games, social networks, and the like. Now the virtual world may be coming to the venerable automated teller machine.

 

Some ATM manufacturers are proposing a virtual approach to fleet management by moving much of the intelligence currently found in ATM software into “the cloud.” In this case, the cloud is a backroom server, which would contain all the intelligence needed to manage applications for a large stable of ATMs.

 

The idea of the virtual ATM first garnered a lot of attention last summer when Diebold Inc. debuted a model at a trade show. But North Canton, Ohio-based Diebold is not the only manufacturer looking at the concept. Executives with Germany-based Wincor Nixdorf, which has U.S. headquarters in Austin, Texas, say they have worked with European clients on virtual ATMs and are just starting to see interest in the U.S. as well.

 

Leading ATM manufacturer NCR Corp., however, has not come around to the idea of moving nearly all of the intelligence currently found in ATM software into the cloud. Still, executives there say it might make sense to move some of the brains to a backroom setting.

 

‘More Agile’

 

The obvious benefit of a virtual ATM is the lower cost of the machine itself, since terminal owners would not need the equivalent of a personal computer inside the unit. But concept proponents say the real cost savings is not so much in the hardware as in information-technology management.

 

When a repair is necessary or a change in features is required, the cloud approach lets ATM deployers make those changes remotely to all the ATMs in the fleet at one time, rather than reprogram each terminal individually or even have a technician sent out to repair or change the device.

 

That could result in substantial savings, and it might induce ATM owners to become more willing to make changes because it wouldn’t be as complicated and time-consuming as in the past.

 

That’s the theory, at least. Skeptics wonder if the idea might slow down the time it takes to complete a transaction if the ATM has to constantly communicate with a back office. This concern comes up most when discussing more complex ATM transactions, such as bill payments, rather than basic cash withdrawals.

 

The concept appears to appeal most to large ATM deployers—those with hundreds or even thousands of ATMs to manage. By contrast, some observers say virtualization may not be as economical for community banks or ATM independent sales organizations that only have dozens of ATMs or even less.

 

These observers also say the concept works best with deployers that have strong network expertise and high bandwidth—something more commonly found in larger institutions.

 

Still, there appears to be growing interest in the idea.

 

“This has a lot of validity. It is similar to the technology reconfiguration that has been going on with PC networks, where there has been a debate about thin-client vs. thick-client networks,” says David Lott, senior vice president with Alpharetta, Ga.-based Speer & Associates Inc.

 

And bankers are intrigued as well.

 

“We just learned about this a couple of months ago and there is definitely some interest in this idea for us,” says Bruce Livesay, chief information officer for Memphis-based First Horizon National Corp., parent company of First Tennessee Bank. “I think this would make us more agile and we could respond quicker to customer demands for new services or functionality.”

 

‘Home Pages on Yahoo’

 

Diebold first proposed a virtual ATM in the U.S. when it unveiled a unit at the VM World conference last August in Las Vegas. Diebold developed the prototype in collaboration with VMware, a Palo Alto, Calif.-based software company.

 

Although no U.S. financial institutions or ATM ISOs have announced they are using the technology, there has been a lot of interest in the system since that debut, according to Balaji Devarasetty, director of cloud services for Diebold.

 

“We’ve been watching the technology trend over the past couple of years toward greater virtualization and wondered what it would mean to ATMs. We decided to take this concept to the point where much of the ATM software would reside in a cloud. That would allow greater speed to market and faster adoption of new services on ATMs,” Devarasetty says.

 

The biggest advantage of the system is that ATM owners could update software and introduce new services quickly. By sending the new programming out in one stroke to all the ATMs, owners avoid having to reprogram each machine individually, Devarasetty says.

 

First Horizon’s Livesay says it currently takes several days to implement a change or add new functionality to all 393 ATMs in his bank’s fleet. He estimates this could be done in an hour or two in a virtual setting.

 

Plus, the virtual approach avoids the issue of whether or not the ATM has sufficient memory to accommodate new programs, Devarasetty adds.

 

Another benefit is troubleshooting, which supporters of virtualization say could be handled easier. If there is a problem at an ATM, the central computer could get the necessary data from the terminal, diagnose the problem, and fix it at the data center, according to Devarasetty.

 

While Diebold attracted a lot of attention to the idea of a virtual ATM in the U.S., Wincor Nixdorf has been working with European banks on developing virtual ATMs for the past three years. But Alan Walsh, vice president of the firm’s U.S. banking division, says the company has seen interest from American banks only in recent months.

 

Walsh believes virtual ATMs could go a long way in helping banks turn ATMs into more complex banking centers than they have been in the past. “Banks want their ATMs to be self-service touch points for the bank, not just cash-and-dash machines,” he says.

 

Banks could find it will be easier to add more features to ATM-based services because functionality can be added quickly through the cloud. “Being able to make a fast change will encourage banks to do more with the machines,” he says.

 

It also would be easier for banks to keep the messages and appearance of the ATM screen consistent with other alternative banking channels such as Internet banking. “Whatever touch points customers use, banks want to have consistency in look and message,” Walsh says.

 

At the same time, Tony Hayes, a partner with Boston-based management-consulting firm Oliver Wyman, points out that this flexibility could enable banks to do more customization of messages to individual customers. “This would make ATMs look more like the home pages on Yahoo. Similar information and features are available to everyone, but each customer could have a unique and customized user experience,” says Hayes.

 

Frustrated Customers

 

But while Diebold and Wincor Nixdorf are supporting the virtual-ATM concept, Duluth, Ga.-based NCR has not jumped on the bandwagon. “Full virtualization is not viable,” says Robert Johnston, director of marketing for NCR.

 

His concern is that constantly sending so much data between the central location and ATMs would slow down the system. “Think about someone trying to pay four or five bills or deposit a lot of checks on an ATM. They might get frustrated waiting for the transactions to be completed, not to mention the guy waiting behind in line who just wants to make a withdrawal,” he says.

 

Rather than move all of the intelligence out of the ATM in what has been described as a “thin-client” model, similar to what happened in the PC world, Johnston supports a “smart-client” approach where some of the intelligence could be moved to a centralized location, but some of the brains would remain in the ATM.

 

For example, a bank could move applications governing simple withdrawal transactions out of the machine, while keeping the computing power necessary for more complex transactions, such as check imaging or bill payment, in the terminal itself.

 

While Diebold and Wincor Nixdorf executives say that today’s higher network bandwidths can support quick transfer of data, Johnston disagrees.

 

“Even with high bandwidths, if you have to transfer a lot of images related to deposits, you are sending a lot of data over the network and it is going to be slow,” Johnston says.

 

With a smart-client approach, he argues an institution could still save some operating costs without having to completely revamp its network. “With a smart client, you don’t have to replace or change the ATMs significantly. A bank could blend this in and not have to make overnight changes,” he says.

 

Johnston says the smart-client approach is in place with some NCR customers in Europe. And several U.S. banks are in the early stages of implementation. “I think we’ll see more of it in the U.S. in the next year to 18 months,” he says.

 

‘The Biggest Risk’

 

While some have questioned where the encryption and related security features would reside in a virtual model, Diebold’s Devarasetty explains that encryption features are embedded in the ATM hardware and PIN pad, not the software. As a result, those components would not be affected by a move to a virtual ATM.

 

And First Horizon’s Livesay believes virtual ATMs actually would be more secure because criminals could not hack into them on the street to get data. Instead, computer hackers would have to attempt to break into a highly secured site at the bank. “Today, there are more places for people to hack into than if you had dumb terminals,” he says.

 

While ATM deployers would most likely see slightly lower costs on new dumb ATMs that were part of a virtual model, Devarasetty explains that deployers would not have to purchase new machines. Any of Diebold’s Optima models, which were introduced in 2003, can be retrofitted into a virtual environment, he says.

 

Walsh of Wincor Nixdorf also says banks would not have to replace his company’s existing ATMs to move toward a virtual environment. They only would need to load a new version of software.

 

Walsh and other ATM experts say the virtual model would appeal mostly to ATM deployers that have many machines. “This has the greatest benefit for a larger deployer with a network of 500 units or more. Even with 50 or 75 units, the cost benefit is questionable,” Walsh says.

 

With a large stable, he estimates an ATM deployer could see a return on investment in 12 to 18 months.

 

Speer’s Lott agrees that the virtual concept works best for large deployers. He adds that geographic distribution also is a factor as the system makes the most sense for deployers whose terminals are concentrated.

 

“You have to have connections to an endpoint. Think of an octopus. You have the body, which is the central location, and the end of the tentacles are the ATMs. The farther apart the ATMs are, the higher your telecommunications costs would be,” Lott says.

 

Devarasetty, however, doesn’t believe virtualization is only for large institutions. He says Diebold is seeing interest from community banks and credit unions as well as big banks. “Anyone who has adopted the virtual model elsewhere in their banking operations would benefit from virtualized ATMs,” he says.

 

Additionally, Hayes of Oliver Wyman notes that most small institutions rely on third-party processors to drive their ATMs. In these instances, the processors could adopt a virtual approach and hold the intelligence for all their clients’ ATMs in a central computer. Then, the combined fleets of the institutions would be large enough to justify the conversion, Hayes says.

 

Still, Livesay notes that it might take plenty of expertise to make the concept work. “We recently revamped our data center so that we have a lot of network expertise internally. But not all banks have that networking expertise and I think virtual ATMs would be harder for them to manage,” Livesay says.

 

Also, banks that have problems with network reliability might have more problems with ATMs going down, he says.

 

And while Hayes supports the concept, his biggest concern is network and connection reliability. “The biggest risk is if the connection is lost. Then you’ve lost the brains to the operation,” he says.

 

The iPad Generation

 

Even if banks choose to move toward a virtual environment, there may be some components that they will want to keep at the ATM.

 

For example, banks that are doing video banking on ATMs or running advertisements may want to keep the graphics elements on the machine. Although it is possible to send such messages from a central site each time an ad runs or a customer requests a video conference, such a setup could be time-consuming.

 

Wincor Nixdorf’s Walsh recommends banks keep graphics that are in use on the ATM. Then, when a bank wants to change them or change its messages, it can send the new elements from the centralized location during off hours.

 

Regardless of how a bank chooses to implement a virtual model, many like the fact that managers and technology developers are trying to keep the 30-plus-year-old ATM up with the times.

 

“ATM vendors need to be innovative so that ATMs can compete and fit more into the modern world,” Livesay says. “ATMs are not going away, but they need to converge into a world where they can coexist with smart phones, iPads, and iPods.”