Sie sind auf Seite 1von 4

A Data Mining Approach to Performance Measurement

in the Banking Industry


Kin-Nam Lau ,Chongyan Gao *
1
Marketing Department, Chinese University of Hong Kong, Hongkong
Business School, Southwest Jiaotong University, Chengdu, 610031 China
1
knlau@baf.msmail.cuhk.edu.hk, gchongyan@gmail.com
Abstract- Performance Management (PM) becomes one of the for the supporting staff. The major problem of these methods
most important initiatives in bank management today. One is that millions of bank customers have no say at all in the
urgent problem for banks is how to link up employee evaluation process and this is against the trend to build a
performance to bank profitability so that resources can be CustomerFocused Organization.An employee evaluated as
optimally allocated and fully utilized to meet competition and very good by the peer may not be regarded as very good
support increasing demand of quality products / services from
customers. Mining customer data to measure Productivity by customers. Moreover, sales quota determination is always
and enhance Teronnance Management is not only feasible la controversial. Employees with poor ranks may not accept the
this information era but also in line with the transformation of a evaluation results and always regard these methods as
bank into a Customer driven organization. In this article, we subjective, biased and unfair.
propme a data mining approach to performance management in Every bank wants to transform into a customer driven
the banking industry. organization and every employee understands hisher salary
Keywords: Performaace Managen ent (PM); bank;
profitability
ultimately comes from customers. There is no debate that
customer voices should be collected and used in PM.
I. INTRODUCTION However, a bank can have millions of customers. Collecting
millions of customer opinions daily is infeasibIe, but a survey
The globalization, intensified competition and rapid on several thousands of customer opinions in a year i s not
.hanges in the socio-economic environments have created reliable to differentiate good performance from bad
serious challenges for banks to survive in the 21st century. performance. Fortunately, most banks have invested billions
One urgent problem for banks is how to link up employee of dollars in information technology to meet their Customer
performance to bank profitability so that resources can be Relationship Management (CRM) initiatives in the past
optimally allocated and fully utilized to meet competition and decade and produced remarkable results to enhance
support increasing demand of quality products / services from effectiveness of their marketing activities. Customer voices
customers. Consequently, Performance Management (PM) are actually computerized in every second of everyday into
becomes one of the most important initiatives in bank bank customer databases. These voices, which are loud and
management today. clear about their needs and satisfaction, have not yet been
Performance management is the process to translate plans incorporated in existing performance evaluation mechanism
. into results (Cokins, 2004, and it is about whos going to do even though the marginal cost of using such information is
what, by when and how will we know the job has been well almost zero.
done? (Parsons, 2001). The focus of PM is to measure
employee performance, department productivity, and II. THEOIUTICAL FOUNDATION
management effectiveness. Traditionally, the PM function is
managed by the Human Resource (HR) department. Mining customer data to measure Productivity and
Among academics, Data Envelopment Analysis @EA) enhance PerformanceManagement is not only feasible in
(Charnes, 1994) is the most popular operational research this information era but also in line with the transformation of
technique to evaluate branch performance in banking industry a bank into a Customer Focused Performance Management
(Thanassoulis, 1999). However, as far as we know, none of System (CFPMS). This system relies on massive customer
the banks in Hong Kong use it for PM because DEA results information supplemented (see Figure 1) with workflow data,
can be very ambivalent (see Metiers, King-Metters and survey data, and peer ratings as the integrated basis to
Pullman, 2003 for managerial concerns about DEA). In evaluate the performance of bank departments and
addition, PM requires more than measurements and DEA employees.
cannot provide specific plan for each unit to improve
performance.
The PM practices did not change too much in the past
decade, Employee evaluation in banks today still mainly
relies on the result-based method (e.g., sales quota) for the F I G m 1 FOUNDAT~ONS
OF CFPMS- TRANsACnONS AND A m
sales staff, and the peer / superior l 3 6 0 degree rating method

- 1009 -
02005 IEEE
0-7803-897 1-9/05/$20.00
where (aj,
dj)are requirements of banks A and B to produce
one unit of attraction. We further assume the market share of
bank A (Ma) is proportional to the attraction of A:
Ma = exP(Va )4exP(Va)+exp(Vb))
Mb = exp(vb)/(exp(va)+exp(Vb)) '(4)
From (3) and (4), we illustrate how the market share of A is
related to its efficiency production capability (ah,),
resource allocation and competition from Bank B (vb).
This simple model, which can later be extended to allow n
product departments and m supporting departments, can be
te&-adlmMre
analyzed in input output framework (Leontief,l986) to derive
performance measures using various sources of customer data,
departmental workflow data and survey data.
We propose to use the input-output analysis (Leontief,
1986) as the basis to quantify the interrelationship among all
m. IMPLEMENTATION FRAMEWORK
departments within a bank and deduce productivity score for The tasks of developing the CFPMS include:
each department. We present here a simplified version of A. Data Engineering.
Input-Output model to illustrate the sophisticated relationship
Voices of all customers embedded in the customer
among efficiency, resource allocation and competition as the
database are directly or indirectly used in the performance
first step to develop a "Customer Focused Performance
evaluation of all departments and employees within a bank.
Management System (CFPMS)" for banks. To illustrate the
model, we assume two banks A and B selling only one Data engineering is the process to convert raw customer data
into actionable customer information about their needs,
product in the market and each bank has three departments,
satisfaction, etc through data cleansing, data fusion and data
where department 1 and 2 are supporting departments, and
mining so that we are able to evaluate the department outputs
department 3 is the product department (i.e., sales
from customer perspectives. For example, we will calculate
department). Let Xu be the number of man hours provided by
the waiting and service time as one of the proxy variables for
department i to support deparhnent j. Then the relationship of
customer satisfaction measure. Mining on massive customer
departments is formulated as an input output table below.
data allows us to produce specific recommendation plan to
each department and each employee to improve their future
Dept I 2 3 performance.
B. Workj7ow Analysis.
2
3
XZI
x3I x32 2 1 We study inputs and outputs of departments, and determine
their evaluation criteria (e.g., speed, impact).
C. Extended Performance Management Model.
The performance management model will be extended,
estimated and applied to compute the department productivity
score which will subsequently be used to allocate resources
and determine their budgets.
D. Objective scoring on productiviq and pe@ormance.
Productivity score of a department will be evaluated by
comparing the resources used and its contributions to
customers. Similarly, performance score of an employee is
determined by comparing hisher salary with hidher
contribution to the department.
Employee bonuses and salary adjustments are determined
by their performance scores. The balanced score card for each
employee within a department will be derived from customer
data, workflow data and peer ratings Io determine the
contribution of a particular employee to bank profit relative to
hisher salary.
The building blocks of the proposed sysb-m are illustrated
in the flowchart below:

- 1010 -
f rank I Sales I Effort I Service I
mPEI

Combine 1 Rank I
Score Score Score Score
1
...
n

IV. PERFORMANCE MEASUREMENTS B. Campaign P e l f o m n c e


Here are some examples to illustrate how the performance As an example, consider the evduation of a customer
of frontline employee, branch and marketing activities (i.e., acquisition campaign by direct mailing which is planned and
marketing campaigns ) can be evaluated by mining massive executed by the insurance marketing department in bank
customer data : headquarter. In most banks today, the campaign is regarded as
successful if the campaign response rate can exceed certain
A. Frontline employee and branch target level (say, 5%), regardless of how much revenue a 5%
A time stamp and a branch code are associated with each acquisition rate can contribute to the bank in short / long term
customer transaction in a branch visit. Therefore, banks know and how much other internal resources ( e.g., service provided
the exact number of customers that they serve by hour, date by retail credit risk department and branch) have been used to
and branch. By counting the number of complaints, and support the campaign. In fact, since alI tansactions are
estimating the waiting time/ service time of each branch visit, computerized, banks can easily trace the revenue generated
banks know how comfortable each customer is with their from each new insurance customer over certain time period
services. By examining the change in product utilization, and also identify all internal resources provided by other
banks know how satisfied each customer is with the bank departments. In other words, banks do have the capability to
product and the price. Prom the productivity view, banks can include all cost and revenue factors from the database to
deduce the total number of service transactions provided by derive a more accurate estimate on the return of each activity
any particular frontline employee and compare their relative conducted by different supporting departments I staff.
efficiency.From the sales perspective, detailed knowledge on Therefore, their performance in terms of contributions to
individual customers can now be recovered through data bank profit can be deduced.
mining so that banks are able to design a more realistic sales
quota for each sales person according to the sales potential of V. CONCLWSION
hisher clients. Moreover, given such valuable customer
On one hand, the integration of massive customer
knowledge, the sales persons have better capability to target
information with workflow data and peer ratings to measure
the right products to right customers at the right time so as to
and improve employee performance in the service industry is
maximize their sales performance. After we derive the
totally new to practitioners, yet this is an urgent matter to
individual performance score for all employees in each
Banks in China to transform themselves into a "Customer
branch, we can aggregate them to produce performance score
driven organization" in order to meet the WTO challenges.
for each branch. The process is illustrated in figure 4 below:
On the other hand, the integration of the input output analysis
(i.e., Stage 1 model) with the "Customer Driven Balanced
Score Determination (i.e., Stage 2 model)" in a bank setting
has not yet been attempted by any academic researchers. If
the ideal is successful, the, results will not only provide
concrete guidelines for oth&+Chinesebanks to follow in their
road to modernization, but also add a new chapter in the
performance management area and stimulate more creative
FIGURE 4 THE CONSTRUCITON OF BRANCH PERFORMANCE SCORE
research in the future. Moreover, tb 3 models and experience
in the banking industry can be tralsplanted tci cther service
industries to help them strengthen their aanagement
effectiveness in the international arena.

- 1011 -
REFERENCES [4] [4] Meaen, R., K. King-Mettem, M., Pullman. Service Operations
Management,Thomson. 2003.
Ill [I1 A., w.Cooper, A. L e w h L.Seiford.. Data Envelopment
151 151 parsoPS,job.,*;&mnt approaches to measmment within fie
Analysis, Kluwer, 1994. service sector & service sector/ white collar institutions. Asian
[2] [2] Cokins, Gq.,Performance Management:finding the missing Productivity Organization. 2001, pp. 11-40.
pieces, John Wiley & Sons, Inc., 2003. [6] [6] Thanassoulis, E., Data Envelopment Analysis and Its Use in
[3] [3] Leontief, Wessily. Input Output Economics, 2nd edition,Oxford 3anking, Interfaces Vol 29: 3 M a y - June 1999, pp. 1-13.
University Res. 1986.

- 1012 -

Das könnte Ihnen auch gefallen