Influence of Transaction Mix on the “Demand for Teller Time” If all transactions were simple deposits, banks could save significant time and money training tellers and could guarantee fast service times to their customers.  But they are not all simple deposits.  Instead, what looks like a stream of “average” customers walking into a branch for service turns out to be a stream of (predictable) inconsistency.  Because the transaction mix changes by time, by day, by day of week or day of month, measuring and predicting only the number of customers or the number of transactions can lead to service breakdown. 
If we are able to understand the impact that transaction mix has on service time and hence the labor required to provide that amount of service at that time, we can take specific actions to improve our odds of service success:  We can better predict what the demand is, staff more accurately and deliver wait times that comply with the target (90% in 5 minutes, or 80% in 2 minutes).  Or, we can identify specifically where service is breaking down and fix it.  For now, we will look at the impact of transactions mix on the “Demand for Teller Minutes.”

Through their Electronic Journal systems, banks already know what the transaction mix is and are equipped to do a good job of predicting transaction mix.  When the mix predictions are combined with accurate service times by transaction type, a clear vision of the “demand for teller minutes” is painted allowing for staffing decisions that put the right number of people behind the counter at the right time.  Data from one bank branch above shows that by looking at 6 frequent transaction types, we can see that there is a great deal of variability in the time it takes to complete each of them, with more than 1 ½ minutes difference in the transactions that are most similar.  Brickstream’s technology collects the data necessary for such analysis every minute that every installed branch is open.  This provides distinct advantages over periodic sampling and time studies.  They include: enormous samples available for all situations – busy, not busy, Mondays or Fridays, lunch hour or 3pm, pay day vs. other days.  Also, because data is always being collected, there is no Hawthorne Effect temporarily driving up the performance of the bank staff.  What is measured is true performance.

It is generally understood that longer service times lead to longer wait times.  In this article, we will focus on the ‘demand for teller minutes.’  If our goal is to improve, or shorten, customer wait time, a good place to focus is teller service time.  In order to make improvements in teller service time, it is essential to understand what forces drive teller service time. 

The first place to look is transaction mix.  By understanding which transaction types occur most often and which ones take the longest, we can start to design training and operating processes to address those transactions with the highest impact on average service time/demand for teller minutes.  There will probably be more to gain by shaving 5 seconds off of a routine deposit than 5 minutes off a rare, but lengthy transaction type.

Banks have long had the ability to understand what transactions are taking place in the branch by using the Electronic Journal system.  But the service time by transaction type from such systems can be misleading.  The transaction time recorded only represents the time the teller is working in the EJ system and not the total time that a customer is standing in front of the teller.

First recognize the impact on service time that transaction mix has.  Then, learn if there is a pattern to transaction mix - Do the 1^st and 15^th of the month have longer, corporate transactions as well as more consumer transactions related to payroll?  If there is a pattern, then determine the impact of different transaction service times (not EJ transaction times) on wait time, queue length, customer satisfaction and retention.

There exists an opportunity to forecast transaction mix as well as transaction count in order to determine the proper staffing levels for branches.  With this view on what transactions are occurring, we will have a more accurate forecast of the demand for Teller Service Minutes, giving the staffing manager a clearer view of what staffing levels are needed to meet service and wait time goals.

Also, armed with this information, the teller process manager is better able to set service time targets for particular transaction types and to target specific types for process improvement initiatives.  Process improvement initiatives will yield increases in teller productivity.  An increase of 10% translates into $21K a year for a 7 teller branch and $30K per year for a 10 teller branch.

Using the combined EJ and Brickstream data, it is easy to identify where large improvements can be made.  If we look at the pre-transaction time, transaction time and post-transaction time within and across branches, we can identify where individuals or groups of individuals can benefit from training programs.
Measuring the transaction length from the Electronic Journal (EJ) system is not enough as it does not account for all of the time that customers spend at the teller window.  Each customer service event is comprised of the portion of time from when a customer arrives at the window until the start of the EJ transaction record, the duration of the EJ transaction, and then the time after the EJ transaction ends until the time that a customers walks away from the window.

The first segment of the service event, or “hello time,” is made up of customer preparation, the teller “catching up” with a familiar customer, or the transfer of paperwork between customer and teller.  The period after the transaction, or “goodbye time,” consists of selling by the teller, additional inquiries by the customer and possibly, relationship building for both.  Of all of these, the transaction type will greatly influence the preparation time, the paperwork transfer, and the selling by the teller.  An EJ system alone cannot shed any light on these influencers of service time.
By indexing the components of service time across branches, we can compare how they are performing against each other.  In the chart below, you can see service components that are good candidates for improvement (highlighted in yellow).  In each case, the average time for that task at that branch is more than 10% greater than the same task across a representative network of branches.  These are ideal areas of process for the branch manager to focus his or her efforts.

In addition to the efficiency gains, banks can expect to reap the same measurable gains in employee satisfaction experienced by other Brickstream users.  These include reduced turnover related costs, reduced training costs, and fewer stress induced mistakes.

In summary, paying close attention to exactly what work is being performed, when it is being performed and with what efficiency can lead to higher utilization, higher customer satisfaction and higher employee satisfaction.