Overview of the Constructs Influencing Customer
Loyalty
There are some important differences
between ABC and traditional costing. The latter, which is based on volume,
distorts the allocation made to marketing activity since it often understates
the cost of slow moving products (Cassar, 1994) or the cost of credit
card holders with low credit turnovers/volume (PRISM Consulting, 1997).
This can lead to product line extensions appearing profitable, when in
fact they are not if new product development and batch set-up costs are
properly allocated. ABC can be time-consuming to implement, but its benefits
are considerable.
A study on what constructs influence
customer loyalty in the Maltese financial services sector (Caruana and
Chircop, 1998) revealed that frequent visitors to the same branch were
positively influenced by the tellers' perceived capability to understand
customer needs and wants. The following four constructs have been repeatedly
cited in the literature as influential factors. The relevance of the propositions
that satisfy these needs of distinct customer segments, the rewards customers
receive for their loyalty, the relationships that organisations truly
build with their customers on a one to one basis, and the retention of
that loyalty over time.
The constructs approach was
used by Runge (1985) and Alter (1975) in performing research similar in
nature to this research when they sought a taxonomy of decision systems.
Reich and Benbasat (1990 also used this approach to identify factors influencing
the decision to adopt. However, Grover (1993) found it necessary to construct
an instrument based on significant studies in innovation to identify constructs
facilitating consumers' decisions. Furthermore, Runge (1985) notes that
the 'selection of constructs for consideration is left largely to individual
researchers' biases, hence a number of important constructs could be ignored'.
We advocate the use of discriminant analysis to identify constructs that
distinguish motivation and behaviour. Though the dynamics of this process
can swamp particular structural aspects of the situation, and limit the
effectiveness of the construct approach, Grover (1993) utilised factor
analysis to yield a 'parsimonious model' based on different categories
of constructs. To minimise such bias, we incorporated the criteria suggested
by Yin (1994) when building the research instrument: construct validity
(both external validity and internal validity), and reliability.
Distinguishing between Motivation and Behaviour
It is important to make clear distinctions
between customer behaviour and customer motivation in determining causes
and effects of loyalty, and to ensure the correct measures are taken.
Behaviour is mathematically linked
to financial outcomes (Caruana, 1993), whereas motivation is less directly
connected (Kotler, 1980). But much of the language in this area tends
not to do so (Dwyer, Schurr, and Oh, 1987). For example, loyalty can be
used to mean behaviour, for instance, repeat purchasing. Alternatively,
loyalty can refer to motivation, that is, feeling loyal towards a frequently
purchased brand.
These opposing views of the meaning of a key term have become embedded
structurally within marketing. The advertising industry prefers the motivational
definition, since it aligns with their activities, while direct marketers
choose the behavioural definition, which they are better able to address
and measure.
Constructs Validation
Construct validity was achieved by
using multiple sources of evidence and by establishing a chain of evidence
supported by a common and consistent research framework based on Runge's
(1985) taxonomy. All our interviews were based on the same structure,
which served to ensure that all the constructs were addressed by each
interviewee. Furthermore, the contextual ratings technique required the
informants to review the statements provided, as suggested by Yin (1994).
Furthermore, we distinguished between the external and internal validity.
External Validity. This assesses
the degree to which an instrument is measuring the construct it is purporting
to measure. The problem of external validity is concerned with providing
generalisation through each case study. As Yin (1994) explains 'case studies
do not provide statistical generalisation and the number of cases studied
is not relevant for that'. Validity is not an absolute characteristic
and has multiple aspects. Using the Grover (1993) model as a benchmark,
our research included the structures, relationships and characteristics
that potentially contribute to customer loyalty. In other words, measures
must truly measure the constructs they are intended to measure. Ray (1984)
argues that 'the development of measures which have been tested for validity
is a critical requirement for the advancement of knowledge in the social
sciences'. However, generalisation is not automatic, and findings must
be tested through replication in more cases. Yin (1994) argues that the
'use of multiple case studies enables the replication of the logic through
other cases'. Furthermore, in this research, the findings were analysed
in the light of existing explanations, based on our experiences working
in the field, which strengthens their external validity.
Internal Validity. Caruana (1993)
argues that internal validity can only be tested statistically. Therefore,
since this research is primarily dealing with qualitative data, it presented
us with a limitation in measuring internal validity because reliable tests
of convergent and nomological validity such as Campbell and Fiske's (1959)
multi-trait multi-method matrix cannot be performed without quantitative
data. However, we feel that internal validity was achieved thanks to the
clear patterns that emerged through the cross-analysis of the case studies.
Perhaps the strongest internal validation of this research was to assess
the convergence between the results from the three multi-variant data
analysis techniques of content analysis, contextual ratings and correspondence
analysis. If the correspondence is high, the researcher can be assured
that the results reflect the research problem as depicted in Section One
above. However, we are aware that this type of convergence does not address
the generalisability of the results to other samples of the population.
Hair et. al. (1998) note that owing to the characteristics of these three
techniques, the convergence of the results 'does provide some internal
validity' to the overall patterns of organisation positions vis-á-vis
the constructs influencing customer loyalty. This is supported by the
fact that most of the data was collected through semi-structured interviews,
enabling respondents to freely introduce into the discussion any issue
that they considered relevant. We feel that the fact that there are a
reasonable number of cases, in different industries, with similar patterns
reduce the probability of the phenomena under study being explained by
other constructs than the ones identified and used in the research.
Reliability. According to Yin (1994)
reliability refers to the degree an instrument is free from error and
yields consistent results. In other words, reliability is concerned with
minimising errors and biases in the study. This implies that a set of
procedures must be available to enable a later researcher to replicate
the same research and achieve the same results. Yin (1994) suggests the
use of a case study protocol and the development of a case study database.
In this paper, the Research Instrument is appended at the back, including
data about the profile of each Maltese organisation taking part, contacts,
interview time spent, names of the participants and commercial data gathered
from secondary sources, plus other relevant data about the fieldwork.
These documents would enable another researcher to replicate the study
in order to test the research findings.
Measures must not vary unreasonably
because of irrelevant factors such as the way questions are asked, respondent
fatigue, and the like. Reliability of constructs was tested by Cronbach's
alpha (1951), which is a measure of reliability that ranges from 0 to
1, with values of .60 to .70 deemed the lower limit of acceptability.
However, for this qualitative research, a longitudinal study testing the
same constructs with the same group of Maltese organisations, will provide
a better construct reliability test. Such a longitudinal study was beyond
the means of this research.
During this research, the quality
of the measures adopted in the design were closely examined, by being
rigorous to construct-operationalisation efforts, especially where organisations
were faced with linking customers' behaviour with their motivating thoughts
and feelings; a very important area where measurement is underdeveloped.
This deficiency has been noted in various investigations including Parsons
(1983), who pointed out a lack of commonly accepted guidelines or measurement
frameworks; Treacy (1986), who outlined the critical importance of clearly
defining constructs and operationalising them in reliable and valid ways;
and Wiseman (1985), who included measurement in his agenda of issues which
need to be addressed in this area.
According to Reich and Benbasat (1990)
this presents significant difficulties since behaviour, such as repeat
purchasing, is not readily associated with thoughts, because the customer
does not exert great amounts of cognitive effort in thinking about the
purchasing decision.
Low cognitive effort can occur for several reasons
The choice is of relatively
low importance to the consumer, or they lack time to consider it. This
applies to many consumer products where repeat purchasing occurs through
inertia, for example, staple grocery items.
The choice is boring, even though
it is important. This can apply to financial services, utilities, income
tax returns, and the like.
The choice is based more on feelings and emotions than on thinking, even
though the purchase is important. This often applies to fashion goods,
clothing, automobiles, etc. A further complication is that customers may
engage in extensive and elaborate searches and information gathering,
even though the ultimate decision is primarily an emotional choice.
Getting the Measure of Behaviour
and Motivation
In most marketing departments, motivation
has tended to dominate, especially with regards to advertising. Alongside
the measures of imagery and brand perception, which have been used, the
growth of satisfaction measures also needs to be embraced. Customer satisfaction
indicators should not be left outside mainstream management, but incorporated
into key performance indicators (Reicheld, 1996).
Behavioural measures have become easier
to take through the growth of data and computing power. Tracking purchases,
in particular, can now be used to identify customer value and propensity
to purchase. Important issues need to be addressed about the period over
which such measures are taken and the way customer segments are identified.
Different product categories will
also require careful consideration of how to separate motivation from
behaviour. High-impulse items have been shown to involve very little cognition,
for example. Purchasing fragrance may be almost entirely driven by the
consumer's perceptions than by any need, and may be highly influenced
at point of sale.
The highly cognitive process of deciding
to buy a car is very hard to influence at the point of purchase. To ensure
that the right balance is struck, this paper outlines an approach, which
combines the full range of measures to provide a holistic view of the
customer relationship.
Populating the Database
Our model shows that the very first
step in formulating and implementing a customer loyalty scheme is triggered
by the fact that the company wants to acquire more knowledge about its
customers. An essential difference between one-to-one marketing schemes
and other forms of direct marketing is that loyalty schemes are data-driven.
Without the ability to formulate a single holistic view of the customer
it is impossible to enact any customer strategy and events which flow
from this way of thinking.
The first stage in building one-to-one
relationships is simply identifying as many customers as possible (Peppers
and Rogers 1993). In our opinion, loyalty schemes are very effective mechanisms
to entice customers to depart with their information.
We call this the "Bribing Phase"
in our model.
Leveraging Existing Customer
Knowledge
Once the first step is successfully
executed, a company undertaking to implement a one-to-one marketing strategy
will have to also put in place either some form of data mart or customer
marketing database, and some processes for analysing and modelling customer
behaviour. Our model combines three constructs: (1) customer behaviour,
(2) customer future value, and (3) customer needs and wants, in order
to enable a company to start formulating a one-to-one relationship.
Based on the combination of results
that emerge from these constructs, the model leads the user to list a
series of business rules about which customers should be contacted with
which offer. By creating business rules based on analysis of those constructs
that influence customer loyalty, users of this model are guaranteed that
they will be on their way to develop long lasting profitable relationships
with their wanted customers. For instance, the model guides users to develop
a measure of customer relationships based on the change in value year-on-year.
Given the difficulties many companies have in identifying their most valuable
customers or in calculating their value to the business, this model provides
an insight to which customers are worth keeping and which are not. Markets
are an aggregate of customers. Markets change quickly and customers change
their lifestyles according to the life-stage they happen to be in. Various
marketing researchers have shown that these have an impact on the type
of products/services customers will purchase (Payne, 1991). The methods
used to segment the customer base need to be able to reflect these dynamics
and need themselves to be dynamically developed to reflect shifts in the
market.
We call this stage the "Adoption
Phase".
Segmentation
Within the context of developing customer
knowledge as part of one-to-one marketing and CRM, our experience in working
with some of Malta's leading organisations shows that many executives
find they are able to leverage customer knowledge as part of their direct
marketing initiatives. Our model is underpinned by geo-demographic segmentation
(WHO the customers are) followed by behavioural segmentation (WHAT they
buy or HOW they buy it). This simple model draws on data, which is commonly
held on the most recent purchase, the frequency of purchase and the value
of purchases. Applying this tool in our model, we were able to produce
ranking lists from best to worse (see Appendices).
Our research revealed that where individual
profitability and / or lifetime value can not be calculated, RFM is a
useful substitute (practical proxy), not least because it is based on
actual, known behaviour rather than inferred characteristics.
Life-Time Values
Life-Time Value (LTV) is another important
calculation underpinning our model. We have deployed this in a host of
companies who are working with us to segment their customer base. The
true figures are hard to produce since real costs over time will vary.
The financial outputs should follow logically within the cause-and-effect
chain that the model proposes. Significantly, the model is intended to
be applied for each customer segment. If this is done, then marketers,
especially within the context of relationship marketing, will be able
to justify their actions and investments towards building profitable relationships.
This is especially important given that many relationship marketing initiatives
have historically failed to make the connection between the activity and
the effect on profits based on the potential future earnings from each
customer.
Summary
In this paper we concluded that
our Customer Loyalty Model proposes a fundamental shift in the dependent
construct which measurement tools are intended to reveal. For the last
twenty years, the construct being sought has been customer satisfaction.
The new process needs to identify a new dependent construct that relates
directly to profit. Instead of asking the customer "how was the service",
customer-centric organisations need to find out "how the customer
will behave". If this can be achieved, marketing will finally be
able to demonstrate its direct effect on revenues and the bottom-line.
_________
References:
Alter, S.L. (1975) 'A Study
of Computer Aided Decision Making in Organisations' in D. Runge, Using
Telecommunications for Competitive Advantage, Oxford: Oxford Institute
of Information Management.
Reich, B.H. & Benbasat, I. (1990) An Empirical Investigation of Constructs
Influencing the Success of Customer-Oriented Strategic Systems, British
Columbia: The Institute of Management Sciences.
Grover, V. (1993) An Empirically Derived Model for the Adoption of Customer-based
Interorganizational Systems, Columbia: Decision Sciences Vol. 24, No.
3.
Ray, M.L. (1984) 'The Critical Need for a Marketing Measurement Tradition:
A Proposal' in Peter, J.P. & Ray, M.L. (1984) Measurement Readings
for Market Research, Chicago: American Marketing Association.
Parsons, G.L. (1983) Information Technology: A New Competitive Weapon,
Volume 24(1): Sloan Management Review.
Treacy, M.E. (1986) 'Towards a Cumulative Tradition of Research on Information
Technology as Strategic Business Factor' in Sethi, V. & King, W.R.
(1991) Construct Measurement in Information Systems Research: An Illustration
in Strategic Systems, Volume 22: Decision Sciences.
Wiseman, C. (1985) Strategy and Computers: Information Systems as Competitive
Weapons, Homewood: Dow Jones Irwin.
|
