A new study from Arthur Andersen and Andersen Consulting concludes that the highest performing financial institutions rely heavily upon automated retail banking delivery channels.

The findings, from a best practices survey titled Strategies for High Performance, support bankers' efforts to move retail transactions away from branches to cheaper delivery channels such as automated teller machines.

However, the survey's directors maintain that more needs to be done if banks hope to weather the storm of nonbank competition that has arisen in the retail market over the last decade.

"Bankers still tend to look at the branch as a minibank, with its own [profit and loss statement] and sales staff," said Waino H. Pihl, a partner with Andersen Consulting in Chicago. "Maybe we need to unbundle branches and say some are sales units and others are service units."

This suggestion is not a new one. But the Andersen report, which compares various operations at high performing and low performing banks, serves as concrete evidence in favor of approaching branch banking from a new angle.

By patterning its branch operations after those of the top performers in the survey, a poorly performing bank with $5 billion in assets could expect to save $10 million annually, the survey reports.

Teller operations are the most in need of reform at the low performing institutions. The survey indicates that these banks could expect $7 million in annual sayings from reducing teller staffing.

These reductions to a large extent hinge on a bank's ability to move basic transactions from the teller line to automated teller machines.

In the early days of the ATM, many bankers found that the availability of self-service terminals did not significantly reduce the number of transactions handled by human tellers. However, as time progresses, the ATM is emerging as the main point of contact with banks for a rising number of consumers.

It stands to reason, then, that an institution deploying large numbers of ATMs can expect the number of transactions handled by tellers to drop. The Andersen study findings bear this out. High-performing banks on average invest 62% more on ATMs than low performing institutions.

This investment is at least partially responsible for the fact that with equal teller staffing, the branches of high performers are able to support 65% more transaction accounts than those of low performers, on average.

But ATM deployment is not the only contributor to bottom line improvements from the teller line. The study notes that the high performers tend to have larger and more numerous branches than low performers, which suggests that efficiencies of scale come into play.

In addition, high performers' accounts per household are 19% higher, which indicates that cross selling tends to be more effective at,those institutions.

The ATM contributes to a teller's ability to sell new products to existing customers, observers said. As routine transactions move to the self-service terminals, tellers presumably have more time to look at a customer's relationship with the bank and suggest new products, where appropriate.

Though tellers do not tend to be the most sophisticated of bank employees, experts said many are capable of identifying sales opportunities. However, they must first be trained to recognize such opportunities.

Most of the financial institutions participating in the survey recognize this fact. Eighty-four percent of respondents said they were standardizing procedures and improving training of branch employees.

Even when tellers are trained in sales, most institutions rely on their platform employees to do most of the work associated with selling new products.

The Andersen study indicates that high-performing banks support 71% more accounts per platform worker than low-performing banks, on average.

As such, a $5 billion asset bank in the low performer category could save $3 million annually by moving its platform operation into the high performer category.

However, because few institutions have off-loaded significant numbers of account-opening transactions to automated terminals -- only 6% of new accounts are opened via alternative delivery channels -- even the high performers can improve their platform operations.

Most of the improvements would aim to free the platform worker from jobs that can be automated so that the worker can do value-added work that might convince customers to partake of more products or services.

"Less than half of platform time is spent on sales and new account setup, reflecting significant opportunity to remove nonvalue-added activities from the duties of platform personnel," the Andersen study states.

The survey's respondents are moving to fill this void, the study findings indicate. Improved sales training is the top strategy for 78% of survey respondents.

Overall, efficiency gains on the branch level are highly dependent upon an institution's ability to move customers to alternative delivery channels.

But if banks are to pare down their branch networks effectively, they need to revamp the way they measure branch productivity first, according to Mr. Pihl.

The Andersen study suggests that branches be valued in terms of productivity -- that is, the number of transactions handled each day -- rather than by the profitability of accounts that originated at the office.

Such measures will give institutions the best picture of which offices are superfluous.

With about 100,000 bank, thrift, and credit union branches in the United States, the banking industry's investment in brick and mortar is clearly overblown, the survey indicates.

The situation needs to be rectified soon, said Mr. Pihl.

"Nonbanks are not weighed down with such extensive investments in brick and mortar, and that allows them to be more nimble," said Mr. Pihl. "I think it's time to wake up and face the competitor, and it's not the bank across the street."

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.