Operational analytics can help determine if a person using a bank's cash machine is the rightful holder of the bank card, by looking at other data, such as where they made recent purchases, Balog said.
That's distinct from more traditional analytics, Bozman said. "If you're looking at transactions after the fact, that's analytics. If you're looking at transactions while you're actually doing business with the customer, that's operational analytics," she said.
It can also help to identify up-sell opportunities when a customer makes a purchase from a travel website, by looking at past behavior, Balog said.
"Bringing analytics into the transaction flow, I think, is the next big thing. It's almost as big as moving from batch processing to transaction processing," he said.
Customers would need to invest in one of IBM's Netezza data warehouse systems, which it has been integrating more tightly with its mainframes. Balog contends that's a better solution for operational analytics than the Hadoop big data platform, which is more suited to historical data.
Last quarter was a big one for IBM's mainframe business, with the launch of its EC12 system pushing sales 56 percent higher. It's a cyclical business, though, and the previous six quarters all saw declines in mainframe sales, after big gains in the first half of 2011.
Averaged out, Balog said IBM's mainframe business has had "three years of compound growth."
It has added about 180 mainframe customers in the past two years, he said, with China leading the way. The booming middle class there means more people are using banking and telecommunications services, requiring more computing power.