The ATM is a stalwart of the banking experience — always there, always on, always ready to dispense cash.
Like any electro-mechanical device, an ATM can break down, go dark, maybe even fail permanently — as the fleet of machines out in the field, so to speak, gets older, the upkeep becomes ever more pressing.
It might be said that when it comes to ATMs, variability is the one constant.
To that end, Rich Johnstun, head of service technology and innovation at Diebold Nixdorf, said in an interview with PYMNTS, financial institutions (FIs) need to manage the technical data about individual machines to see where attention must be paid.
After all, being proactive about maintenance and upkeep can save headaches and costs down the line — and advanced technologies such as machine learning and artificial intelligence (AI) can play a role in that proactivity, addressing pain points in new ways.
He said the stage is set for FIs to leverage advanced technologies to determine the “individual personalities” of ATMs, and the stressors they encounter — weather, for example, or where an urban location may see heavier demand and use than a rural area.
The data? It traces back to the engineering level, tied to sensors, motor function and other granular details that show how well a machine is running — and streamlining service efforts. Johnstun said developing the profiles at an ATM level is akin to creating a “curated” experience that can take into account different types of transactions in an age where mobile devices are increasingly gaining traction and FIs are actively redesigning their end-user experiences as they upgrade machines to accommodate those new features.
“We are looking for a specific type of interaction,” said Johnstun. “We want that technology to feel sleek and lean. As the branch footprint diminishes, as the reliance on tellers diminishes, then the reliance on that ATM goes up” and the FI is ever more reliant on the machine to keep banking relationships sticky and loyal.
The consistent advances in technology, in the ability to collect, store and analyze data means that, for example, FIs and their service providers can be alerted to “transport recovery” issues that may eventually impact a machine when it tries dispensing cash.
In that case, Johnstun said, the ATM goes through a clearing action and recovers.
“In theory, that machine is working,” said Johnstun, “but we can look a bit deeper at sensor data and find out why that issue occurred. We’re not simply looking at a single message that says there’s been transport recovery.” Sensor-level data, coupled with those traditional message-based systems, he said, can help detect patterns in a machine’s behavior or profile.
Being able to crunch that data and look for patterns using things like machine learning and AI means being able to predict some of those failures. The root cause data, known as “low level engineering data,” fed through AI and machine learning, can pinpoint and predict points of failure (and even a failure timeframe seen, in, say 72 to 96 hours, a window of time that can allow firms to schedule maintenance).
Johnstun said Diebold Nixdorf is in the early stages of examining how the sensor-level data can be leveraged to illuminate fraudulent activities such as skimming. The data collected and examined, he said, must be moved in and out of machines in accordance with regulations tied to PCI and GDPR.
Moving Toward Mobile
There’s been a continued blending of mobile technologies and mobile interactivity with what might be termed the physical cash world, said Johnstun — where the ATM is the point of interaction. He took note of pre-staging activities, where the need to swipe or insert a card is eliminated. Here, the transaction is done across a mobile device, perhaps with biometrics as an added layer of security, and there’s no chance for skimming.
“I create a relationship between my mobile device and that physical piece of hardware there that's giving me the cash,” said Johnstun. There’s at least some variability of adoption of such transactions, he said, in certain geographic regions — where, for example, the comfort level with biometrics is lower in the U.S. than might be seen in Europe.
“What we are really talking about here,” Johnstun said, “is driving availability, driving uptime, and driving an experience to make sure that whenever a consumer walks up to a machine that it’s up, live and running.”