by Rene T. Domingo (email comments to rtd@aim.edu)

The key to efficiency and fast customer service in the service industry is the correct and dynamic matching of demand and capacity. Given the seasonality and unpredictability of the various types of banking services and transactions, this process may be difficult but not insurmountable. The important step is to separate the bank's factory-like transactions which are more or less standardized, like check encashment, withdrawals, and check deposits, from the personalized ones like opening accounts, loan application, or marketing a new service. It should be recognized that the former constitute the bulk of banking transactions. Factory-like transactions are fairly predictable in terms of duration or cycle time, and occurrence. Their seasonality or peak-and-low periods during the day, week, and month can also easily be discerned from a careful study of past data. By matching the demand and capacity of its factory transactions, a bank can decongest its lobby or ATM booths, improve over-all customer satisfaction, and provide its staff with ample time and better composure to attend to the more personalized transactions.

Long-term capacity planning is a critical task of bank management. No plan or a wrong plan is planning for failure or bad service which leads to customer attrition. After the right the capacity is set and installed, whether tellers, verifiers, or ATMs, there is a need to dynamically match capacity to a changing demand pattern. In manufacturing, this short-term process is called production control. Both long and short term capacity matching has to be done carefully and adapted to the bank's particular environment. In many banks, capacity management is characterized by fire-fighting and gut-feel decision making. Tellers in a branch are added or subtracted from the front-line, according to across-the-board head office guidelines which are not consistent with local realities or demand pattern. Moreover, the branch manager may request head office for more personnel after overtime has become unmanageable or customer complaints due to long queues have mounted.

Perhaps the most blatant flaw committed by many banks is basing its capacity plans on its total dollar value of business, e.g. deposit base, loans outstanding, or total assets. To compound the error, a productivity index is derived by dividing the dollar value by the headcount. The head office may stipulate a standard index like so much dollars in deposits per employee. Branch managers are enjoined to observe this index by downsizing or risk getting adverse performance appraisal. Often this index is used by head office to reject the branch's "unreasonable" request for additional manpower. Basing capacity on value will either lead to overcapacity and idle resources, or under-capacity and long customer queues.

In banking, as in many other service establishments, what use capacity, consume resources, and drive costs are not the financial value of the transaction, but rather the volume of transactions. To process a $100 check deposit takes exactly the same teller or ATM time to process as a $200 one. A $100 passbook withdrawal or check encashment would take about the same time to transact as a $200. Even in the non-factory like transactions, there is no direct relationship between value of the service and the amount of bank resources utilized to deliver it. For instance, in general, a $2,000,000 loan will definitely take less than twice the time to process a $1,000,000 one. To correctly match capacity to demand, it is important therefore to derive the total volume of factory-like transactions and translate these to teller time or man hours, or machine hours for automated services. Thereafter, these hours can be translated to the number of tellers or ATMS required to meet the expected demand.

A branch may get the average number of checks encashed, check deposits, cash withdrawals, cash deposits, utility payments on a daily, weekly, or monthly basis, whichever is appropriate for its capacity planning. It is tempting to use the number of customers accounts, particularly the number of active accounts, to compute for capacity. While this is more accurate than dollar value, and easier to use than transactions, it is not as accurate than the latter, since clients will have different volumes and types of transactions. Tellering, whether manual or automated, is transaction driven, not client or value driven. The best way to determine teller deployment is to use the number of transactions - regardless of number of clients or value of transactions - and translate these to transaction hours, and then to headcount. With the increasing power of computer technology at the disposal of banks, getting an accurate count of transactions should not be difficult. But should it be tedious or impractical to do so, you may impute transaction hours based on weight, like 5 man hours to process 10 kilos of checks.

If a service transaction is done by one station or one-stop-shop, say by one empowered, multitasking, multi-skilled teller, capacity planning is simple and straightforward. But if the transaction should pass through several processes or stations, it is necessary to match and synchronized all their capacities to the demand volume being received by the first station. Usually, in a series of processes or hand-offs, there would exist the slowest process, or bottleneck, that would dictate the over-all capacity of all the processes or system capacity. It is a waste of resources and money if a front-line capacity like tellering is increased to match demand when the bottleneck capacity is determined by a slower backroom operation, like signature verification or computer processing. In this case, system capacity will remain lower than demand in spite of the investment. Both front-line and backroom capacities have to be adjusted to meet the demand.

The last step in capacity planning is fine tuning for unplanned activities that decrease capacity. These are machine downtime, and errors and rework. Machine or computer breakdown can slow down other albeit efficient processes. The normal response is to provide for an allowance in capacity planning to account for this, say 5%. The better response of course is to do preventive maintenance to eliminate downtime. Clerical errors, especially teller mistakes, and their correction consume a lot of man-hours and can significantly cut capacity just like downtime. Providing an allowance for these abnormalities is tantamount to tolerating them. It is much better to enjoin everybody to do his job right the first time.