In the late 1990s, the Bank of the Southwest (BSW), a regional financial institution located in the southwestern United States, decided it was not getting its fair share of the equity loan business. Residential housing prices were climbing at a steady rate in virtually all of its markets, and credit card debt also was rising. Many customers were converting credit card debt into lower-interest-rate home equity loans, but research showed that fewer than 25 percent of BSW customers who converted were doing so through the bank’s equity loan facility. BSW was getting only 2 percent of the conversions to equity loans in the market, even though it had a banking market share in its largest markets of between 11 and 23 percent.
The CEO set the goal of doubling both the rate of credit-card-to-equity-loan conversions for current customers (from 25 to 50 percent) and the percent of overall equity loans it was capturing (from 2 to 4 percent). He wanted that accomplished in 12 months. When these goals were announced, they were considered aggressive, in fact, the bank personnel in charge of promotion considered them unattainable.
Enter Quality Function Deployment
A consultant suggested the bank use quality function deployment (QFD), a powerful method for translating customer needs into design requirements. Though historically applied to product design work, QFD can work just as well for designing services and processes.
BSW decided to try this route, and set an objective that the bank’s promotions group use QFD to design a home equity package that would capture a greater share of the market. The promotions group started its work by gathering reliable data on customer needs. These include stating clear objectives, determining what elements of the market are in scope for the design effort, segmenting the customers based on functional needs, then creating a plan to gather voice of the customer data from the segments of potential customers most likely to affect the objectives.
Acting on the Research Results
An analysis of the current process for producing equity loans showed that the average loan was taking more than 24 days, compared to the expectation of five days or less. In addition, there was tremendous variability from loan to loan. The quickest were done in 15 days, but many took more than twice as long. What was the reaction to these facts? Management felt that it was unrealistic to expect incremental improvement efforts to reduce the processing time from 24 days to five. So they turned once again to their consultant, who suggested the bank change the focus of its QFD effort. Rather than aim at getting the right loan package, he said, they should use QFD to design a new, much faster process.
That plan was endorsed, and the promotions group was instructed to redesign the process, including a redesign of the team itself. The membership changed to include people from all parts of the current process so that all of the bank internal customers as well as regulatory and legal customers were represented.
This change in focus turned out to be the key to getting an effective process designed, piloted and implemented in less than 22 weeks.
Some necessary changes to the process became obvious early on. In working with several representative equity loan customers (from two different customer segments), the team found that customers did not want to have to make several trips to the bank as the existing process required. Actually, the physical presence of the customer at the bank was really needed only once per loan – when it was time to sign the legal documents. All other exchanges could be done by phone and/or email. Under some circumstances, the signing also could be arranged to be done at a place of the customer’s choosing rather than at the bank. Other changes to the process came about through the detailed QFD analysis.
Executing the Steps of QFD
Quality function deployment asks project teams to work through a series of steps to translate what customers need into specific design requirements.
It starts by having the team determine priority themes in the voice of the customer statements. Usually this is done by sorting customer statements using an affinity diagram, then summarizing the discovered needs in a tree diagram. The BSW team had to balance external customer requirements against those of, for example, the legal department and regulatory agencies. Internal banking policies also had to be satisfied.
The rest of the QFD analysis is based around a series of specialized matrices (each a house of quality) that help the team link the needs first to specific performance requirements and then to operational factors in the process that will guarantee those requirements are met.
House of Quality 1
Customers ask for things to be easy, fast, intuitive, etc. Unfortunately those words are not definitive enough for a designer to act upon them. The first house of quality translates the wants and needs of the various customers into technical critical customer requirements (CCRs). In addition, the first house of quality:
- Captures how well competitors are able to accomplish these same wants
- Does a pair-wise comparison of CCRs to see if they are in conflict or not
- Provides a technical comparison of various ways of putting the CCRs into practice
Most important customer needs – The research allowed the team to identify the most important customer needs and assign each a priority number (1 to 10, with 10 being the most important). The top four needs and their priority scores were:
- Fast (priority = 9)
- Require very little of their time (priority = 8)
- Give them a maximum loan amount (priority = 5)
- At a competitive interest rate (priority = 7)
The team established the following CCRs to answer the wants:
- “Fast” was defined as cycle time of five business days +/- one day.
- “Require little time” was defined as initial contact time of 60 minutes and total follow-up contact time of 90 minutes.
- “Maximum loan amount” was determined by an algorithm based on appraised value, value trends in last five years and type of building (risk related to collateral).
- “Competitive interest rate” depended on the region and the person’s payment history (risk related to person).
In many design projects, it is advantageous to do further analysis on any new, unique or especially difficult CCRs to determine what factors in the process can influence those CCRs. This analysis happens through the second house of quality.
In this case, the bank had never monitored or tried to control contact with the external customer (a key input), so the two “contact time” requirements were transitioned to the second house of quality. The bank decided to include the “fast” requirement in this analysis as well. Though the bank had historically monitored cycle time, there had been no attempt at control.
House of Quality 2
While there is not enough room here to go through every analysis that went into the House of Quality 2, here is one example: To make sure the company could meet the CCR for initial contact time (less than 60 minutes), the team worked on designing a structured-yet-flexible interface process that provided for all modes of initial contact – phone, in-person and email. A phone answering script rapidly took the person’s query through initial qualifying filters and provided for getting application forms to the customer via mail, email or fax. All of the forms were in the same format, no matter what their point of origination, so all bank clerks had the skills to process them. That provided tremendous flexibility in allocation of data entry activities. And, in the case of internet-based contacts, the customer did the data entry. However, real-time filters were designed to flag the most frequent mistakes and thus reduce the need for corrections later.
House of Quality 3
There was one further translation step in process design. A third house of quality was created to address how the important process factors would be monitored and controlled. The team identified relevant data and set up mechanisms so that it was collected and charted daily – often in real-time but in no case longer than 24 hours after the event. Anomalies were recorded so that decision rules could be updated as more data/information became available. In other words, a “learning” system was implemented as an integral part of the measure/control system.
The beauty of these interlinking houses of quality is that the project team can explain exactly why each process element and control mechanism is in place – because the team can link it directly back to customer statements or regulatory requirements.
Results Beyond Expectations
Bank of the Southwest decided to pilot the newly designed process in a major metropolitan market for six months, and then evaluate it for an enterprise-wide deployment. But the results after three months were substantially better than anticipated (an increase of 30 percent for BSW customer-initiated equity loans), so the CEO directed the promotions group to suspend the pilot and begin implementation.
Just six months later, the response from current BSW customers averaged an increase of 80 percent (the smallest gain was 62 percent and the largest was 131 percent) in the bank’s seven largest markets. However, this response was not duplicated for non-BSW customers. Market share in the non-customer segment only grew from 2 to 2.6 percent.
Statistical analysis of the ongoing equity loan operations revealed an unexpected result – the shorter the time to complete the loan transaction, the better the loan payment record. Further study showed that applications from persons with very good credit went through the new process faster. So, the new process became a win-win, with the bank increasing its success in the home equity loan market and increasing the quality of the loan payment records.
Conclusion: QFD Not Just for Manufacturing
Tools like QFD have such strong roots in manufacturing that many people in services either think these tools do not apply to their situation or think they are simply too complicated. Obviously, the kind of detailed analysis possible with QFD is just as effective in helping companies understand complex service processes.