Perceptual dynamics of ‘good’ and ‘poor’ service quality
Nicholas C. Georgantzas
Management Systems ¢ GBA * LL617-C
Fordham University @ Lincoln Center
New York, NY 10023-7471, USA.
ABSTRACT
Service researchers support the necessity of integrating policy and design dimensions with
service front-line variables in modeling service systems. Current research unveils multiple causes
of good and poor service quality as well as the goal that service designs for quality should attain.
The goal is neither to narrow nor to close, but to reverse the gaps among customer expectations
and perceptions of service quality. Grounded on the contributions of conceptual and empirical
Tesearch, a small three-sector system dynamics model describes the interactions of policy and
service front-line variables in'a typical quasi-manufacturing service. The firm treats customer
defections as measurable scrap and, in a company-wide effort to ferret out weaknesses against
potential loss, its top management is committed to soliciting feedback from defecting customers.
Computed decision scenarios trace the patterns experienced with performance to the inauspicious
effects of pulling on internal policy levers too hard. The resulting dysfunctional behavior shocks
the entire service system, including customers, defectors and profit per customer. A radical change
in the firm’s average customer life (avgLife) target triggers a cycle-doubling pattern in the calls
soliciting feedback from defecting customers. This chaotic pattern forces the entire system to
respond accordingly. System dynamics can provide the integrated-process view required for
understanding self-inflicted problems in services. Along with its policy analysis and service design
implications, the simulation output indicates the morphology of the topology possibly underlying
customer perceptions of service quality.
INTRODUCTION
In their service-system design matrix, Chase & Acquilano (1989) incorporate both policy
analysis and design variables, such as innovation, operational focus and worker requirements. The
strategic dimensions of this well-known matrix both propel and curb service front-line variables.
Filtered down into service front-line behaviors, policy dimensions determine the sales opportunity
of a service. The matrix allows grouping services into pure, mixed and quasi-manufacturing, along
the customer contact dimension. Pure services, such as legal and medical services require direct,
face-to-face contact, with loose specifications between the customer and the service provider.
Quasi-manufacturing involves little or no contact among customers and workers. The use of on-
site technology and the mail contact between a firm’s headquarters and the local government
agency are examples of quasi-manufacturing services. Mixed services have two components: the
first requires little labor, but the second is labor intensive. Examples are television broadcasting,
with its transmission and production components, and research and development (R&D), which
requires both expensive equipment and human thought. Like pure services, mixed services require
intense customer contact, but within tight specifications. Customer contact can be active or passive
(Mersha, 1990), with its intensity varying across industries. Generally, producers of tangible
goods, such as soap and beer, limit customer contact to the retail end. Customer preferences are
very important for manufacturing production, but actual customer presence is not (Buffa & Sarin,
1987: 38). Conversely, customer contact intensity is high both in pure and in mixed services
(Fitzsimmons & Sullivan, 1982). Quasi-manufacturing services offer a high innovation potential.
Their explosive growth matches that of associated technological advances, with service customers
reaping the benefits of great variety at a declining cost (Chase & Tansik, 1983).
SYSTEM DYNAMICS '93 151
Intense customer contact blurs functional boundaries and makes measuring service quality and
productivity difficult. Service employees who interact with customers perform production and
marketing functions as well. Bitran & Hoech (1990) urge firms to train front-line employees for
handling diverse customer requests and temperaments. Customer perceptions of service quality
depend on interpersonal skills, such as courtesy, friendliness, tolerance and pleasantness (Hobson,
Hobson, & Hobson, 1984). Service quality involves both the processes and the outcomes of
service production, delivery and consumption (Parasuraman, Berry, & Zeithaml, 1991). Also, it
depends on the differences among customer expectations and perceptions of quality (Lewis &
Booms, 1983). Service promotions raise customer expectations of quality; if not met, they cause a
gap between customer expectations and perceptions of quality (Hore, 1986). Psychological
theories, such as the adaptation level, assimilation contrast, prospect theory and the Weber-Fechner
law complement each other in explaining the formation of such gaps (Laitamiki, 1990).
In their ongoing survey, Parasuraman et al. (1991) focus on accessibility, reliability,
Tesponsiveness, competence, courtesy, communication, credibility, security and understanding and
knowing the customer. Also, they stress tangibles, such as physical facilities, appearance of
personnel and the sophistication of the tools used to provide a service. Miller-Duffy &
Fitzsimmons (1988) assess service quality along similar dimensions, but find complaint data
highly suspect for measuring service quality. Yet, the case by Bitran & Hoech (1990) and the data
of Mersha, Adlakha & O’Brien (1988) verify. the importance of these variables. Specifically,
knowledge of the ‘product,’ attention to customers and courtesy rank as the top causes of good
quality. Rudeness, indifference or “I don’t care” attitude and reluctance to correct errors rank as the
top causes of poor quality. Price correlates negatively with perceived quality. If customers feel that
the price of a service is unfairly high, then their perceptions of service quality will be low.
This brief survey includes both case and empirical studies converging on salient causes of
service quality. The empirical studies contribute to. theory building through hypothesis testing, but
the cases provide an equally important, indeed necessary contribution. Some of them suggest
linking service front-line variables, such as customer expectations and perceptions of service
quality, with promises that firms make to customers. Others call for integrated models to help
service managers and researchers consider quality at the service design phase (Bitran & Hoech,
1990; Lyth & Johnston, 1988). Similarly, Reichheld & Sasser (1990) suggest combining statistical
process control (SPC) with customer defection analysis to measure and to improve service quality.
System dynamics provides a structure for modeling real systems. This internally coherent
structure offers the integrated perspective required for understanding services. Grounded on the
contributions of current research, the model presented here describes the interactions among policy
and service front-line variables in a typical quasi-manufacturing service. Besides its direct
implications for practice, the simulation output reveals the morphology of the topology- possibly
underlying customer perceptions of good and poor service quality.
MODEL STRUCTURE
With customer defections identified as measurable scrap, the top management of a quasi-
manufacturing service is committed to soliciting feedback from defecting customers. In a company-
wide effort to improve quality against potential loss, defection analysis guides the firm’s
continuous improvement. The variables of the STELLA® (Richmond & Peterson, 1992) diagram
in Figure 1, both propel and curb customer buying decisions. These variables depict customer
expectations of attention, courtesy and knowledge as well as their perceptions of quality.
In Fig. 1(a), potential users consider the service at the prevailing industry rate. If they use the
service, according to the industry’s use norm, they become customers; some respectfully decline.
Customers stay with the firm until the industry’s defect norm or a gap in the perceived over
expected quality turns them into defectors. Defecting is not necessarily permanent, but a state of
152 ~ SYSTEM DYNAMICS '93,
5| Fig. 1
{@) Service customers.
complaints (b) Customer expectations & perceptions.
Piet (©) The “hut of quality’ in services. useNorm
03 gap
0 gap 2 f
Potential Users reuse
@ consider defect
’
re) O Customers Defectors
enabling use — avgtife defectNorm !eave
ol decline SO
0 Quality 7
Expected Quality Perceived Quality
2EQNOt aPq\at
2.30}
gap
feedback 0) =
gepf Defectors Calls
0.75) Quality
0 gap 2 gapDelay 4
Quality
O'g'Qvot
1
orf]
ie) gap 2 © = ~ = as g
good highQ- enabling low poorQ
Calls feedback gap
1s aciet
goodg| oo complaints
ol = C)
0 calls 50 Defectors decision limCall avgLife target
1 32| 1 3
highQ| limCalt fowQ poorg|
al dl ol ol
0.8 “Quality 1 0 avglife 2 0 Quality 0.2 0 Calls 50
' target
SYSTEM DYNAMICS '93 153,
flux. A few defectors stop using the service completely. Depending on the industry’s use norm,
most of them reuse the service unless, of course, a gap of perceived over expected quality
emerges. This gap, which determines sales opportunity through the use, reuse and defect rates, can
move defectors permanently into a competitor’s customer base. As long as customers stay with the
firm, they generate sales and profit. The longer the firm keeps its customers, the more profit it
makes. Defined as the net present value of profit per customer stream over the average customer
life, this pattern is common across different service industries (Reichheld & Sasser, 1990).
In Fig. 2(b), quality and the calls soliciting feedback from defectors determine each defector’s
sampling opportunity. Quality is enabling with diminishing retums (Samuelson, 1980). The higher
quality is, the more permeable each call contact, the higher eaca defector’s sampling opportunity
and the effective feedback that the firm turns into material improvements. Conversely, the lower
quality is, the less transparent the contact, the less effective the sample, the less effective the
feedback. Naturally, services do not allow inspecting quality before delivery and consumption.
During calls, defectors share the cost of appraisal, verifying and inspecting quality in ad-hoc client-
worker teams. Errors need not ‘get out’ to affect perceived quality. Defectors witness costly
investigations and adjustments, penalties and lost accounts right there on the ‘shop floor’ during a
call. Many defectors are disappointed customers. Even if they defect temporarily, assuredly they
spread negative ‘word of mouth’ with a delay (Reichheld & Sasser, 1990).
Modified through calls, the average defector’s gap affects reuse directly, but the use and defect
fates also embody word-of-mouth weights though conditional statements (Appendix). Although
reciprocal, the defect and reuse rates carry weights of equal magnitude because retained customers,
who have experienced the defectors’ call treatment already, are sensitized and thereby more
perceptive to word of mouth than potential users. The parameters in these rates and in some of the
model’s auxiliary graphical functions stem from the considered judgment of service manages and
workers, manifested in interviews and discussions with them. Objective service data are
increasingly available, but the data of Reichheld & Sasser (1990) helped to calibrate the model’s
baseline parameters. Most helpful was Homer’s “impressionistic” contribution (1985: 46),
illustrating a fundamental tenet of system dynamics. Carefully testing the behavior produced by a
potent model’s endogenous structure yields accurate insight and applicable results, even without
numerically precise data.
The service firm monitors and calls defecting customers from its ‘hut of quality’ in Fig. 1(c).1
Along with the effective feedback, which the firm needs to improve quality, the gap causes
unsolicited complaints. Typically, complaint data do not constitute effective feedback because they
are highly suspect (Miller-Duffy & Fitzsimmons, 1988). Yet, they result from low perceptions or
high expectations (upper left corner of Fig. 1), which can stimulate rudeness, indifference and
reluctance to-correct errors in a caller’s response (Bitran & Hoech, 1990). Such incidents are
manifestations of poor quality (d‘p’Qldt), and thereby affect it independently from the firm’s
continuous effort to improve it (6‘g’QIdt).
Call decisions depend on quality’s enabling effect, on feedback effectiveness and on the call
limit (limCall). The firm’s managers control this limit through the avgLife target. The legitimacy of
increasing this internal policy lever is well established, not only in this firm, but in the entire
service industry (Reichheld & Sasser, 1990). The target lever is also the means to curbing the
enthusiasm of enabled callers, currently receiving effective feedback from defecting customers.
After the 3rd or 5th call, however, legitimately, some defectors may reperceive attention as
tudeness or, worse yet, invasion of privacy. This concern, which the management shared, led to
several madel tests, assessing how the firm’s defectors, customers and profit per customer might
Teact to incremental changes in the avgLife target.
1 As opposed to the ‘house of quality’ used in manufacturing (Hauser & Clausing, 1988).
154 SYSTEM DYNAMICS '93
et) 7
| os ee ees Ce
C) w(t) 156
—— Customers
32| '
by ef
22 .
0 w(t) 156
—>— Defectors
40 T
so(t)-
10) 4
g wit) 156
—— Calls
1 T 1 T 1.2| y
L > pat) | gae(t) 1
L ol L ol 1
10 se(t) 40 to) a(t) 1 te) a(t) 1
— cals vs ——— Quality vs Perceived —— Perceived vs gap
Fig. 2 T y
Performance resulting 3 LS
from a 20% increase in
the average customer life Q
(avgLife) target, 30 weeks the 4 tr 7
from the initial time t=0. so) oe)
‘The calls cause defectors -
to reuse the service and
thereby increase the 2al 1 ol -
‘customer base, but only 60 c(t) 80. 0 n(t) 1.2
temporarily. —— Customers vs Defectors ——— sample vs gap
SYSTEM DYNAMICS '93
155
MODEL BEHAVIOR
Initializing the model in equilibrium prevented latent artifacts of relationships operating within
individual sectors from contaminating the computed behavior patterns of defectors, customers and
profit per customer. Setting the avgLife target equal to 3 years matched yet another norm prevailing
at the quasi-manufacturing service. To conduct the tests, the model was disturbed from equilibrium
by 1% step increments in the avgLife target, 30 weeks from the initial time. This simple procedure
examined how the firm’s defectors, customer base and profits respond to incremental changes in
the avgLife target, and made the model’s behavior patterns easy to interpret.
Figure 2 shows how the customer level rises and the defector level drops by 0.50% (1/2 of a
1%), respectively, in response to a 20% step increase in the avgLife target. These changes show
up 21 weeks from the step increase in the target, at t=30 weeks. The resulting discrepancy between
the avgLife and its target causes calls to increase, while the defector expectations exceed their
perceptions of quality. The rising calls convey increased attention, so the defector sample and
perceptions rise only to raise their expectations. With the gap dropping quickly after its initial
increase, complaints rise, pushing quality down, independently from the firm’s effort to improve it
through calls soliciting feedback from defectors. Quality’s low enabling effect reduces the calls and
associated complaints, so quality start rising again, increasing the defectors’ perceptions of it.
This 13.5-week cycle repeats itself, while quality and the gap reach higher levels each time
they peak. Once quality exceeds its equilibrium level, the reversed gap of perceived over expected
quality causes a drop in defectors. The defectors reuse the service, increase the customer base, and
their avgLife goes up, to reduce the calls again. The phase plots clarify the characteristic features of
the service quality cycle. The call fluctuations affect the each defector’s sample, which in turn
affects the defector perceptions and expectations of service quality. It is only at the peak of the gap
between perceived and expected quality that defectors respond, setting the simultaneous plotted
values of customers and defectors into a small cycle during the evolution of the quasi-
manufacturing service system.
Under pressure from an unyielding avgLife target the calls enter into a cycle-doubling pattern,
unable to return to their equilibrium position. This pattern is more apparent in the time development
plots of Figure 3. The 10-year avgLife target increases the amplitude of the call limit cycle as well
as the amplitude and frequency of the customer and defector cycles. The figure’s phase plots verify
the dramatic increase in amplitude, particularly in the customer and defector cycles. By itself, the
policy of “zero defections” — keeping every the firm can profitably serve, will not produce the
desired performance the quasi-manufacturing service managers anticipate. Figure 4 shows how
futile the adoption of such a policy might be, if not combined with more substantive process re-
engineering efforts. The average profit per customer pattern, computed from the data of Reichheld
& Sasser (1990: 109), mirrors the vicious-circle of customer defections.
Also, Figure 4 presents the morphology of the topology possibly underlying the defector gap
of perceptions over expectations of service quality. Pulling the internal policy lever, avgLife target,
on too hard can shift attention from good to poor quality perceptions. This perceptual cross-over is
evident in the phase plot of the simultaneous defector sample and gap values. Temporarily,
perceptions of good and poor quality may be independent dimensions, but their medium of
existence customers and defectors carry permanently between their ears.
CONCLUSION
System dynamics is the study of how structure produces behavior in real systems. System
dynamics modeling and simulation should be used to describe service systems, before more
service managers adopt more company-wide ‘zero solutions’ to self-inflicted problems that treat
customers, defecting or not, as measurable scrap. Even a small system dynamics model can
accurately describe the interactions of policy and service front-line variables in a quasi-
156 SYSTEM DYNAMICS '93
80)
w(t)
156
w(t)
156
°
—— Calls
156
ol i
ol 1
ott)
10 se(t)
Fig 3
Increasing the average
customer life (avgLife)
target from 3 to 10 years,
30 weeks from the initial
time, totally shocks the
service system. The calls
enter a cycle-doubling
pattern, leaving customers
& defectors no choice but
to respond accordingly.
40
— alls vs Quality
22
q(t) 1
—— Quality vs Perceived
oo
ont’)
ol
\
60 c(t) 80
—— Customers vs Defectors
L
0 A(t)
—— sample vs gap
1.2
SYSTEM DYNAMICS '93
157
manufacturing service. Computed decision scenarios trace the behavior patterns experienced with
performance to the inauspicious effects of pulling on internal policy levers too hard. A radical
change in the firm’s average customer life norm triggers a cycle-doubling pattern, forcing the entire
system to respond accordingly. System dynamics can provide the integrated-process view required
for understanding self-inflicted problems in services. Along with its policy analysis and service
design implications, the simulation output points to the morphology of the topology underlying
customer perceptions of service quality.
APPENDIX
Equations of the quasi-manufacturing service defection analysis model
Levels Auxiliaries
Calls(t)=Calls(t-dt)+(aClat)*dt avgLife=100/defect
INIT Calls=13.89 decision=min ( Defectors*limCall,
Customers(t)=Customers(t-dt)}+(reuse+use-defect)*dt Calls*feedback*enabling )
INIT Customers=61.73 gap=Perceived_Quality/Expected_Quality
Defectors(t}=Defectors(t-dt}+ gapDelay=DELA Y (gapf+sample,8*DT)
(defect-leave-reuse)*dt sample=Calls*Quality/Defectors
INIT Defectors=27.78 complaints=GRAPH(gap)
Expected_Quality(t)=Expected_Quality(t-dt)+ (0.00,5.00),(0.25,4.00),(0.5,3.00),
(@EQIt*dt (0.75,2.20),(1.00,1.60),(1.25,1.12),
INIT Expected_Quality=1.57 (1.50,0.8),(1.75,0.55),(2.00,0.3)
Perceived_Quality()= enabling=GRAPH (Quality)
Perceived_Quality(t-dt)+(aPQIat)*dt (0.00,0.01),(0.2,0.4),(0.4,0.7),
INIT Perceived_Quality=sample (0.6,0.9),(0.8,1.00),(1,1.00)
Potential_Users(t}=Potential_Users(t-dt)}+ feedback=Gl »)
(Consider-use-decline)*dt (0.00,2.30),(0.25,1.90),(0.5,1.60),
INIT Potential_Users=100 (0.75,1.35),(1.00,1.15),(1.25,1.00),
8Cldt=decision-Calls (1.50,0.9),(1.75,0.8),(2.00,0.75)
Quaicy() Qualia yi e'e Qlat-d*p'Qiat)*at gapf=GRAPH(gap)
INIT Quality=0.91 (0.00,-1.00),(0.25,-0.5),(0.5,-0.2),
(0.75,-0.07),(1.00,0.00),(1.25,0.07),
Rates (1.50,0.2),(1.75,0.5),(2.00,1.00)
consider=100 goodQ=GRAPH (Calls)
defect= if gap<1 then (0.00,0.01),(10.0,1.00),(20.0,1.25),
Customers*defectNorm*1.25 30.0,1.35),(40.0,1.45),(50.0,1.50)
elseif gap=1 then Customers*defectNorm highQ=GRAPH(Quality)
else Customers*defectNorm/ 1.25 (0.8,1.00),(0.85,0.9),(0. y 0.7),
__ use=if gap<I then (0.95,0.4),(1.00,0.01)
Potential_Users*useNorm/ 1.125 limCall=GRAPH(avgL ife/target)
else if gap=1 then Potential_Users*useNorm (0.00,32.0),(0.2,16.0),(0.4,8.00),(0.6,4.00),
else Potential_Users*useNorm*1.125 (0.8,2.00),(1,1.00),(1.20,0.5),(1.40,0.25),
reuse= if gap<1 then (1.60,0.125),(1.80,0.0625),(2.00,0.00)
Defectors*useNorm/ 1.25 lowQ=GRAPH(Quality)
else if gap=1 then Defectors*useNorm (0.00,0.01),(0.05,0.4),(0.1,0.7),
else Defectors*useNorm* 1.25 (0.15,0.9),(0.2,1.00)
decline=Potential_Users-use poorQ=GRAPH(Calls)
leave=Defectors-reuse (0.00,0.01),(10.0,0.12),(20.0,0.3),
dEQ\6t=Expected_Quality*gapDelay (30.0,0.6),(40.0,1.20),(50.0,3.00)
Constants
defectNorm=0,36 {grand USA mean} ?
length: useNorm=0.25
& integration method: Runge-Kutta 4. target=3+STEP(7,30)
2 Reichheld & Sasser (1990).
158 SYSTEM DYNAMICS '93
200) t
p10(t)
ae
p3(t)
0 w(t) 156
a
Profit per Customer: avgLife target = 10 years
Profit per Customer: avgLife target = 3 years
good quality
poor quality
1.2| i
Fig. 4
It may well be that
perceptions of good
& poor quality resemble gap( t)- aI
non-homotopic paths on —
tori surfaces. On each torus,
there are infinite closed paths
which do not touch at any point. ft) 1
Also, there are closed paths which 0 n(t) 12
cannot deform one into another continuously. —— sample vs gap
SYSTEM DYNAMICS '93 159
REFERENCES
Bitran, G.R. & Hoech, J. 1990. The Humanization of Service: Respect at the Moment of Truth.
Sloan Management Review, 31 (2): 89-96.
Buffa, E.S. & Sarin, R.K. 1987. Modern Production/Operations Management (8th ed.). New
York, NY: Wiley.
Chase, R.B. & Acquilano, N.J. 1989. Production’ and Operations Management (Sth ed.).
Homewood, IL: Irwin.
Chase, R.B. & Tansik, D.A. 1983. The Customer Contact Model for Firm Design. Management
Science, 29 (9): 1037-1050.
Fitzsimmons, J.A. & Sullivan, R.S. 1982. Service Operations Management. New York, NY:
McGraw Hill.
Hauser, W. & Clausing, D. 1988. The House of Quality. Harvard Business Review, 66 (3): 63-
2B.
Hobson, C.J., Hobson, R.B., & Hobson, J.J. 1984. People Skills: A Key to Success in the
Service Sector. Supervisory Management, 29 (10): 2-9.
Homer, J.B. 1985. Worker Burnout: A Dynamic Model. with Implications for Prevention and
Control. System Dynamics Review, I (1): 42-62.
Hore, F, 1986. The New Wave of Management Thinking for Service Industries. International
Management, 37 (4): 82...
Laitamaki, J:M.-1990. The Formation and Role of Reference Prices in the Choice of Consumer
Services. Ph.D. Dissertation, Ithaca, NY: Cornell University & Ann Arbor, MI: UMI.
Lewis, R.C. & Booms, B.H. 1983. The Marketing Aspects of Service Quality. In L.L. Berry, G.
Shostack, & U. Upach (Ed.), Emerging Perspectives on Service Marketing , (pp. 99-107).
Chicago, IL: American Marketing Association.
Lyth,-D.M. & Johnston, R. 1988, A Framework for Designing Quality into Service Operations. In
Proceedings of the Annual Operations Management Association Meeting, Coventry, UK.
Mersha, T. 1990. Enhancing the Customer Contact Model. Journal of: Operations Management, 9
(3): 391-405.
Mersha, T., Adlakha, V., & O’Brien, W. 1988. Consumer Assessment of Service Quality. In
Proceedings of the Annual Decision Sciences Institute Meeting, Las Vegas, NA, Nov. 21-23:
1242-1244, :
Miller-Duffy, J.A. & Fitzsimmons, J.A. 1988. Measuring Quality of Care in Nursing Homes. In
Proceedings of the Annual Decision Sciences Institute Meeting, Las Vegas, NA, November
21-23: 1248-1250.
Parasuraman, A., Berry, L.L., & Zeithaml, V.A. 1991. Understanding Customer Expectations of
Service. Sloan Management Review, 32 (3): 39-48.
Reichheld, F.F. & Sasser, W.E., Jr. 1990. Zero Defections: Quality Comes to Services. Harvard
Business Review, 68 (5): 105-111.
Richmond, B. & Peterson, S. 1992. STELLA® II: An Introduction to Systems Thinking (2.2.1
ed.). Hanover, NH: High Performance Systems.
Samuelson, P.A. 1980. Economics (11th ed.). New York, NY: McGraw-Hill.
1600 SYSTEM DYNAMICS '93
CC BY-NC-SA 4.0