Journal of Information Technology (1999) 14, 375 386

Loyalty of on-line bank customers

L EIF B. MET HLIE and HE RBJ RN NY SVEE N
Department of Strategy and Management, Norwegian School of Economics and Business Administration,
Breiviksveien 40, 5045 Bergen, Norway

Electronic commerce changes the relationships between sellers and buyers dramatically. The new properties
of electronic markets offer customers added values. New customer value propositions have to be established
in most markets and new marketing strategies must be formulated. One of the business sectors most heavily
affected is banking. How can banks retain loyal customers when moving into electronic banking? Research
in marketing has unveiled several determinants of customer loyalty. However, this knowledge is based on
research on customers in traditional markets. In this paper we focus on customers loyalty in on-line banking.
The ndings indicate that determinants of loyalty in on-line banking environments are similar to those in
the physical market-place. However, customer satisfaction is found to have the most signi cant impact,
followed by brand reputation, while switching costs and search costs, although signi cant, have minor
explanatory power. The implications of these ndings on banks marketing strategies are discussed.

Introduction
On average, US companies lose 50% of their customers
in 5 years (Seybold, 1998). How many leads are
required to recruit a new customer? Surveys show that
it is up to six times as expensive to recruit new customers as it is to retain existing customers (Rosenberg
and Czepiel, 1983). Furthermore, loyal customers
are assumed to be less price sensitive (Krishnamurthi
and Raj, 1991) and the presence of loyal customers
provides the rm with valuable time to respond to
competitive actions (Aaker, 1991). Seybold (1998)
described the economics of customer loyalty based
on a pro tability model developed by Reichheld
(1996). The elements to achieving higher revenues via
customer retention are (1) base revenue the longer
you retain them, the more money you make, (2) cost
savings they cost you less to serve, (3) price premium
they pay more for your products, (4) acquisition
costs they cross-sell and up-sell themselves and (5)
they generate referrals.
On these grounds it is not surprising that so many
banks have embarked on customer loyalty programmes,
usually targeted at speci c market segments. Here they
seek to establish durable relationships by adding value
to services and offering favourable prices. Banks have
traditionally established strong brands and experienced
low turnover of their customer base. Bank customers
have generally been loyal to their main bank. Will online banking in electronic markets change this situation?
`Electronic commerce is transforming the marketplace by changing rms business models, by shaping
relations among market actors, and by contributing to
0268 3962 1999 The Association for Information Technology Trust

changes in market structure (OECD, 1998, p. 18).

Thus, on-line banking in electronic markets requires
new business models for banks. What are the new value
propositions of banks needed to meet customers
buying behaviour in electronic markets and how are
the customer values created? Rayport and Sviokla
(1994) claimed that customer values are created in the
content the functionality of the product or service
provided and in the infrastructure the channel of
delivery and the context the way the product or
service is presented to the customer. The World Wide
Web on the Internet, the dominant technological
platform of electronic commerce today, has unique
properties that enable new customer values to be
created in content, infrastructure and context (Methlie,
1998).
The Internet and electronic markets have imposed
a fundamental shift in business economics. Evans and
Wurster (1997) called it the new economics of information. It will change the way we do banking in all
core areas: production, distribution, trading and
payment (Llewellyn, 1997). It lowers transaction costs
(Bakos, 1991) and increases the economy of scale,
scope and speed in processing banking services to the
market. As a result, fundamental restructuring of the
value chains in banking is to be expected. Llewellyn
(1997) advocated a contract model of banking based
on vertical disintegration and outsourcing; Evans and
Wurster (1997) described it as `deconstruction of
the value chain (p. 74) and Hagel et al., (1997,
p. 43) called it `changing the entire value proposition
of retail banks. Disintegration of vertically integrated
value chains in banking offers cost ef ciency and added

376
exibility. `In this new world distribution will be done
by the phone company, statements by nancial
management software, facilitation by different kinds
of agent software, and origination by any number of
different kinds of product specialists (Evans and
Wurster, 1997, p. 77).
If customer loyalty is a key factor in achieving
pro tability in banking, as can be inferred from the
many loyalty programmes found in this sector and
electronic markets are changing the banking environment and customer relationships, then a study of
what determines customer loyalty in on-line banking
should be of great interest. Therefore, in this paper
we focus on factors that keep on-line customers loyal
to banks on the Internet.

Loyalty and the determinants of loyalty

Many de nitions of loyalty have been presented in the
literature. Jacoby and Chestnut (1978) gave a summary
of many of them. Their conclusion was that the
only way to reveal real loyalty is to measure customers
attitudes and intentions towards the brand in question. Based on this, Oliver (1997) de ned customers
loyalty as `a deeply held commitment to rebuy or repatronize a preferred product or service consistently in
the future, despite situational in uences and marketing
efforts having the potential to cause switching behavior
(p. 392). This means that customers must perceive
the bank as better than competitors and that customers
must like the bank better than competitors (affective
loyalty). Rebuying or repatronizing means that customers must have an intention to keep on using the
bank in the future (conative loyalty). In other words,
affective loyalty is related to customers attitudes
towards the bank while conative loyalty is related to
customers behavioural intentions towards the bank in
the future. Based on this, we see that loyalty is not
only rebuying or what Dick and Basu (1994) called
spurious loyalty but that loyalty towards a brand
also implies a positive attitude towards the brand in
addition to intention to rebuy the brand.
In this paper, we address the question of what
determines the loyalty of on-line banking customers.
As noted above, loyalty is a complex, aggregated
concept that cannot be de ned nor measured by
any single variable. From marketing literature, several
antecedents of loyalty have been revealed. Among the
determinants of loyalty towards a brand are customers satisfaction with the brand (Selnes, 1993;
Samuelsen et al., 1997), brand reputation (Selnes,
1993; Samuelsen et al., 1997), switching costs (Dick
and Basu, 1994), salesperson trust (Macintosh and
Lockshin, 1997), salesperson commitment (Macintosh

Methlie and Nysveen

and Lockshin, 1997) and affective antecedents such as
emotion, feeling and mood (Dick and Basu, 1994).
Furthermore, social norms and situational in uences
are assumed to affect loyalty (Dick and Basu, 1994).
In this article, we focus on loyalty among bank
customers on the Internet. They do not meet salespersons. Salesperson trust and salesperson commitment are therefore excluded from our analysis.
Furthermore, we have excluded affective antecedents
such as emotion, feeling and mood, assuming that
customer relationships are primarily based on cognitive evaluations of the bank due to the nature of the
services bought. Search costs are assumed to be
reduced when moving from the traditional marketplace to electronic markets (Bakos, 1991). Search costs
are also treated as a part of the switching cost construct
(Fornell, 1992). Since search costs are regarded as
an important part of the economics of electronic
markets, we chose to treat search costs as an explicit
determinant of loyalty here.
From the arguments above, we embark on a model
of customer loyalty including the following four determinants: (1) satisfaction with the brand, (2) brand
reputation, (3) switching costs and (4) search costs.
Moving from the market-place to an electronic
market changes customer relationships. It is therefore
assumed that some of the determinants of loyalty will
be affected. A reduction in search costs has
been predicted (Bakos, 1991, 1997; Strader and Shaw,
1997). This reduction is assumed to reduce loyalty and
it may also be argued that it will change the effect of
search costs on loyalty relative to other determinants.
Brand reputation is often used as a heuristic for the
evaluation of a brand. If a brand has a positive reputation, customers are supposed to be loyal to the brand.
Furthermore, better opportunities in searching for
information about a brand in electronic markets make
it easier for customers to evaluate brand attributes
before they buy. This reduces the importance
of brand reputation as a heuristic for brand quality.
A consequence of this could be that the relative
importance of brand reputation on loyalty is reduced.
Electronic markets also enable a wide range of new
seller and customer relationships to be set up, for
instance increased and customized product information, customer participation in product development
and reduction in order ful lment time (Methlie, 1998).
All this may improve customer satisfaction (Strader
and Shaw, 1997). By increasing customer satisfaction,
this variable may increase its importance as a determinant of loyalty. Based on these possible changes,
a fair proposition is that the values of the determinants of loyalty will change when moving bank
services from the physical market-place to an electronic
market.

377

Loyalty of on-line bank customers

From research in marketing we have some
knowledge about the determinants of loyalty in the
physical market-place. Taking this knowledge as our
basis we want to investigate whether determinants of
loyalty in the market-place hold in electronic markets
as well. We shall look at the following determinants
of loyalty: customer satisfaction with a brand, brand
reputation, switching costs and search costs.
Satisfaction
Customer satisfaction is a post-choice evaluation of
a speci c transaction (Selnes, 1993). Early studies
conceptualized customer satisfaction as customer
perception of product performance (Cardozo, 1965;
Anderson, 1973). Later studies showed that customer
satisfaction is a function of customer perception of product performance compared to a set of standards (e.g.
expectations, values and norms) the con rmation/
discon rmation paradigm (Yi, 1990). Besides the
direct effect of con rmation/discon rmation, both
expectation (the comparison standard) and perception
of brand performance are found to have direct effects
on customer satisfaction with a brand (Churchill and
Surprenant, 1982; Yi, 1990).
Satisfaction is considered to act as an antecedent
of loyalty, arising out of direct prior experience (Dick
and Basu, 1994). Several studies have found support
for the positive relationship between customer satisfaction with a brand and their loyalty towards that
brand (Fornell, 1992; Sandvik and Duhan, 1996;
Samuelsen et al., 1997). `Loyal customers are not
necessarily satis ed customers, but satis ed customers
tend to be loyal customers (Fornell, 1992, p. 7). A
satis ed customer has few incentives to change supplier
or brand. Therefore, `Customer satisfaction makes it
costly for a competitor to take away another rms
customer (Fornell, 1992, p. 10). Satisfying customers
should therefore be of great importance to vendors in
keeping their customers loyal.
Customers who are satis ed with their bank probably develop a positive attitude towards the bank. As
a result, their intention to stay with the bank in the
future is likely to be high. Based on this argument
we propose the following hypotheses.
H1a: Increasing customer satisfaction leads to higher
affective loyalty.
H1b: Increasing customer satisfaction leads to higher
conative loyalty.
Brand reputation
A brand is a name, term, symbol, sign or design to
identify the goods or services of one seller and to

differentiate them from those of competitors (Keller,

1998). Brand reputation has been de ned as a perception of quality associated with the brands name
(Selnes, 1993). A strong brand is a brand with `a set
of assets linked to a brands name or symbol that adds
to the value provided by a product or service to that
rms customers (Aaker, 1996, pp. 7 8). A strong
brand is a signal of quality; it is a risk reducer for
customers and it reduces search costs (Keller, 1998).
A brand name or a brands reputation is often
used as a heuristic for choice when intrinsic cues or
attributes are dif cult or impossible to employ (Selnes,
1993). Brand reputation and satisfaction may look like
the same construct. In the literature, however, brand
reputation is seen more as a long-term and overall
evaluation of the product than customer satisfaction
(Selnes, 1993).
A positive relationship is expected between brand
reputation and customer loyalty to the brand (Sandvik
and Duhan, 1996). The reason for this is that a brands
reputation might function as a social norm and that
customers often rely on the beliefs of their reference
groups to a great extent. Another reason why brand
reputation is important for loyalty is that it may be
dif cult to evaluate satisfaction with a product directly
in some situations. In such situations the brand
reputation is used as a heuristic for evaluating the
product. However, an important point to note is that
customer satisfaction with a brand is a determinant of
brand reputation in some situations. In the insurance
industry, for instance, customer satisfaction with a
brand is found to have a positive relation to brand
reputation (Selnes, 1993). This underscores both the
direct and indirect effects of customer satisfaction on
customers loyalty. However, in this paper we are
focusing on the direct effect of customer satisfaction
on loyalty. The expected effects of brand reputation
on affective and conative loyalty are presented as the
following hypotheses.
H2a: Increasing brand reputation leads to higher
affective loyalty.
H2b: Increasing brand reputation leads to higher
conative loyalty.
Switching costs
Porter (1980) de ned switching costs as the costs
of switching from one suppliers product to another
suppliers product. Fornell (1992) included search
costs, transaction costs, learning costs, customers
habits, emotional costs and cognitive effort in the
switching costs construct. As can be seen, switching
costs include both economic and psychological values.
Switching costs is a common strategy to increase

378
loyalty (Dick and Basu, 1994). Currently banks are
eager to launch loyalty programmes where customers
obtain substantial bene ts by holding most of their
banking business within one bank (positive lock-in).
`Switching barriers make it costly for the customer
to switch to another supplier (Fornell, 1992, p. 10).
The implication of this proposition is a positive
relationship between switching costs and customer
loyalty. High switching costs retain customers from
changing banking relationships. Therefore, an increase
in switching costs will lead to an increase in loyalty.
Since this is a kind of `lock-in strategy, it probably
has a positive effect, mainly on the intention to keep
on as a customer (conative loyalty) rather than on
customers affective commitment towards the bank
(affective loyalty). Customers will probably perceive
high switching costs as something negative which
reduces their exibility to switch supplier or product.
Therefore, we predict a negative relationship between
switching costs and affective loyalty and argue the
following hypotheses.
H3a: Increased switching costs lead to lower affective
loyalty.
H3b: Increased switching costs lead to higher
conative loyalty.

Methlie and Nysveen

H4a: Increased search costs lead to lower affective
loyalty.
H4b: Increased search costs lead to higher conative
loyalty.
Affective and conative loyalty
Affective loyalty is a component of loyalty that focuses
on how much a customer likes the bank or how positive their attitude is towards the bank. Conative loyalty
focuses on the customers intention to keep on using
the bank in the future. Fishbein and Ajzen (1975)
argued that attitude towards a product predicts attitude towards buying the product. Attitude towards
buying the product is assumed to predict intention to
buy the product. In marketing literature, attitude
towards a product is often used as a direct predictor
of intention to buy the product (Homer, 1990; Brown
and Stayman, 1992). Based on this, it seems plausible
to predict a positive effect of affective loyalty on
conative loyalty. Therefore, we propose the following
hypothesis.
H5: Increased affective loyalty leads to higher
conative loyalty.
The model

Search costs
Fornell (1992) included search costs as part of
the switching costs construct. In this paper, search
costs are treated as a construct of their own since
search costs are an important aspect of the economics
of electronic markets and a key concept in studying
buyers behaviour. A standard de nition of buyers
search costs in the economic literature is `the cost
incurred by the buyer to locate an appropriate seller
and purchase a product (Bakos, 1997, p. 1677). A
major impact of electronic markets is that they
reduce the costs customers must pay to search for
information about products in the market (Bakos,
1991).
High search costs are assumed to make it expensive
for customers to shop around for the best offer in
the market. The implication of this is that it is costly
for customers to nd alternative brands or suppliers.
Therefore, they keep on using the same brand or
the same supplier. This implies that high search costs
lead to high conative loyalty. However, customers
are probably not satis ed with a situation where it is
dif cult to nd alternatives and, thus, having limited
exibility being imposed on them. Therefore, it is
assumed that the relationship between search costs and
affective loyalty is negative as for switching costs.

So far we have presented four determinants of

signi cant importance for loyalty, both affective loyalty
and conative loyalty in traditional markets. In this
paper, we want to study the effects of these determinants in the context of on-line banking. Figure 1
depicts the model of our study. Positive and negative
postulated relations are represented by + and ,
respectively.

Figure 1

The proposed model for the study.

Loyalty of on-line bank customers

The analysis will show whether the drivers for loyalty
are valid for both affective loyalty and conative loyalty.
The hypotheses are summarized in Figure 1.

The study
To test the hypotheses, we collected data from
customers of three banks delivering services on the
Internet. The study was carried out in January 1998.
As for most banks on the Internet today, the services
currently available are primarily account-based services
(bill payment, money transfers between accounts and
account information). Data were collected by a
questionnaire accessible on the home page of each of
the three banks for 10 days. The respondents lled
out the questionnaire on their computer screens and
returned the data by the Internet. The data collected
were organized in an Access database and copied to
LISREL for statistical analysis. Respondents could
win prizes by answering the questionnaire. Therefore,
we controlled for multiple responses from identical
persons by checking their Internet addresses. A total
of 1463 respondents answered the questionnaire. The
respondents were between 18 and 71 years old with a
mean age of 36 years compared to the mean age of
the population of 37 years. Of the respondents, 6%
were women, compared to 18% of the population of
Internet bank customers at the time of the study.
Therefore, women were under-represented in the
sample. However, after splitting the two groups men
and women we found no signi cant difference
between them with respect to the effect on the two
loyalty constructs discussed in this paper. Only 9% of
the sample had not nished high school. As many as
73% of the respondents had been to college for 3
or more years, of which nearly 60% had a degree in
engineering or business. From more general data, we
know that well-educated people are the strongest
users of the Internet. Overall, our sample was fairly
representative of the population except for the overrepresentation of men in the sample which, however,
as mentioned above, did not represent a problem in
our analysis.
Measures, reliability and validity
In the study we looked at search costs as a construct
separated from other switching costs. Search costs were
measured by two items: (1) `I used a lot of time
searching for information about other banks services
before I selected the bank and (2) `I used several information sources to collect information before I selected
the bank . Switching costs were measured by (1) `I
think I will lose money by quitting my relationship

379
with my bank and (2) `I think it will cost a lot of
money to change bank . For satisfaction we used (1)
`So far, my bank has satis ed my expectations and
(2) `I feel that my bank manage my loan and deposits
in a good way . Three items were used to measure
brand reputation: (1) `My bank has a good reputation
among my friends, (2) `My bank has a good reputation compared to other banks and (3) `My bank
gets a lot of new customers because it has a good
reputation. In discussing loyalty above, we argued that
loyalty includes both a deeply held commitment
towards the bank (affective loyalty) and an intention
to keep on using the bank (conative loyalty). Affective
loyalty was measured by (1) `My bank has personal
meaning to me and (2) `I feel that I belong to my
bank . Conative loyalty was measured as (1) `I intend
to keep on as a customer at my bank and (2) `I gladly
recommend my bank to other people that I know. All
items were measured on a seven-point scale from very
negative (1) to very positive (7). Reliability was
assessed by inspecting the item reliability and mean
variance extracted. As shown in Table 1, the values of
item reliability are between 0.55 and 1.0 indicating
reliable measures. The values of the mean variance
extracted varied from 0.84 to 0.70, all well above the
0.5 level suggested by Bagozzi and Yi (1988).
Consequently, we argue that reliability is assured (cf.
Table 1).
According to Anderson and Gerbing (1988), convergent validity can be assessed by determining whether
each indicators estimated pattern coef cient on its
underlying construct is signi cant. This is done by
checking the t-values for the indicators of the factor
loadings (see Table 2). The results show that all of
the pattern coef cients were signi cant. Consequently,
convergent validity is assured.
Discriminant validity is assessed by determining
whether the con dence interval ( 2 SE) around the
correlation estimate between two constructs includes
1.0 (absolute value). The results presented in Table 3
show that none of the con dence intervals included 1.0
(absolute value), thus indicating discriminating validity.
Although all constructs passed the Anderson and
Gerbing (1988) test for both convergent and discriminant validity, this test is admittedly a weak test of
discriminant validity, particularly with such a large
sample. Consequently, discriminant validity was also
assessed by the procedure suggested by Fornell and
Larcker (1981). They argued that the mean variance
extracted should be higher for each latent construct
than the squared correlation between the constructs.
From Tables 2 and 3, we see that none of the squared
correlations between the constructs were higher than
the mean variance extracted for each of the constructs.
Consequently, discriminant validity is assured.

380
Table 1

Methlie and Nysveen

Reliability

Construct

Item

Search costs

I used a lot of time searching for information about other banks

services before I selected the bank
I used several information sources to collect information before I
selected the bank
I think that I will lose money by quitting my relationship with my bank
I think that it will cost a lot of money to change bank
So far, my bank has satis ed my expectations
I feel that my bank manages my loan and deposits in a good way
My bank has a good reputation among my friends
My bank has a good reputation compared to other banks
My bank gets a lot of new customers because it has a good reputation
My bank has personal meaning to me
I feel that I belong to my bank
I intend to keep on as a customer at my bank
I gladly recommend my bank to other people that I know

Switching costs
Customer satisfaction
Brand reputation

Affective loyalty
Conative loyalty

Table 2

Item
Mean
reliability variance
extracted
0.93

0.70
1.00
0.55
0.73
0.66
0.71
0.79
0.60
0.76
0.91
0.72
0.81

0.82

0.78

0.70

0.70

0.84

0.77

Measuring model
Search
costs

I used a lot of time searching for information about

other banks services before I selected the bank
I used several information sources to collect
information before I selected the bank
I think that I will lose money by quitting my
relationship to my bank
I think that it will cost a lot of money to
change bank
So far, my bank has satis ed my expectations

0.96
(17.87)
0.84
(17.21)

Switch
costs

SatisfactionReputation

1.00
(27.94)
0.74
(23.27)

0.85
(37.46)
0.81
(38.08)

I feel that my bank manages my loan and

deposits in a good way
My bank has a good reputation among my friends

My bank has a good reputation compared to

other banks
My bank gets a lot of new customers because it
has a good reputation
My bank has personal meaning to me

I feel that I belong to my bank

I intend to keep on as a customer at my bank

I gladly recommend my bank to other people

that I know

Values are standardized estimates with t-values in parentheses.

Affective Conative
loyalty
loyalty

0.84
(38.08)
0.89
(41.33)
0.78
(33.85)

0.87
(39.15)
0.95
(44.21)

0.85
(38.64)
0.90
(42.24)

381

Loyalty of on-line bank customers

Table 3

Estimated correlation coef cients between latent constructs

Search costs

Search costs
Switching costs

Satisfaction

Reputation

Affective loyalty

Conative loyalty

1.00
0.06a
0.03b
2.34c
0.05
0.03
1.67
0.06
0.03
2.26
0.10
0.03
3.49
0.02
0.03
0.74

Switch costs

Satisfaction

Reputation

1.00

0.34
0.03
12.08
0.28
0.03
10.38
0.30
0.03
11.27
0.30
0.03
11.04

1.00

0.64
0.02
31.70
0.54
0.02
23.46
0.82
0.01
56.50

1.00

0.50
0.02
22.41
0.67
0.02
36.28

Affective
loyalty

1.00

0.67
0.02
37.13

Conative
loyalty

1.00

Estimated correlation coef cients.

Standard deviation.
c
t-values.
b

Results
The results of the analysis are presented in Figure 2.
As can be seen, we analysed the effects of the
antecedents on both affective and conative loyalty. Due
to the x 2 s well-known dependency on sample size, the

Figure 2

The results of the study

model t was deemed to be satisfactory even though

the x 2 p-value was signi cant. The sample size of 1463
was considered to be large and, consequently, we based
the t assessment on other indices. The root mean
square error of approximation (RMSEA) was 0.047,
which is below the recommended 0.05 level (Browne
and Cudeck, 1992). Both the goodness of t index
(GFI = 0.98) and the comparative t index (CFI
= 0.99) were satisfactory.
Figure 2 illustrates the relationships between the
antecedents and affective loyalty and conative loyalty,
respectively. From the squared multiple correlations
we infer that the model predicts the two kinds of loyalty
fairly well, with 76% of the variance explained for
conative loyalty and 36% of the variance explained
for affective loyalty.
The results show that customer satisfaction and
brand reputation are the two most important determinants for both affective loyalty and conative loyalty.
For conative loyalty, customer satisfaction (0.57) is the
main antecedent but brand reputation also has a signi cant effect (0.16). The effects of switching costs
( 0.02) and search costs (0.01) on conative loyalty are
not signi cant. The effect of affective loyalty on conative loyalty is positive and signi cant.
For affective loyalty, the results support three of
the four hypotheses presented in Figure 1. The effect
of customer satisfaction (0.33), brand reputation
(0.27) and search costs ( 0.14) are all signi cant and
in the predicted direction. The relationship between
switching costs and affective loyalty, however, is
positive and signi cant (0.12), rather than negative as

382
hypothesized. The implication of this is that an increase
in switching costs leads to an increase in affective
loyalty.
Based on these results, we conclude with support
for the satisfaction hypotheses and the brand reputation hypotheses for both affective and conative loyalty
(H 1a, H1b, H2a and H2b). The predicted effect of
search costs on affective loyalty is supported (H4a),
while the postulated effect of search cost on conative
loyalty is not signi cant. None of the switching costs
hypotheses were supported. The results show support
for the predicted effect of affective loyalty on conative
loyalty (H5).

Implications of the results

The ndings of this study support the hypothesized
effects of customer satisfaction and brand reputation
on affective and conative loyalty. The results indicate
stronger support for satisfaction and reputation than
for switching costs and search costs as determinants
of loyalty. This applies to both affective and conative
loyalty. Since customer satisfaction and brand reputation are the main determinants of loyalty, banks on
the Internet should struggle to satisfy their customers
and build strong brands.
Satisfaction
Customer satisfaction is associated with an overall
performance of the banking relationship that customers
experience through repeated transactions with the
bank. As described above, customer satisfaction is
determined by two indicators: expectations (comparison standards) and peformance perception. As shown
in Figure 2, customer satisfaction explains more of the
variation in conative loyalty than in affective loyalty.
The effect of customer satisfaction on loyalty con rms
previous ndings that satis ed customers tend to be
loyal (Fornell, 1992; Sandvik and Duhan, 1996;
Samuelsen et al., 1997).
In order to achieve loyal customers banks must know
how to satisfy their customers. Thus, focus on the
customers must be given highest priority in on-line
banking. However, a critical issue follows: how can
banks remain in control of customers access to the
distribution channels? In the physical market-place
banks have dedicated links to their customers through
their branch network. In electronic markets, services
are mediated through a distribution channel with
several intermediaries participating in the value chain:
context providers, network operators, Internet access
providers, business service providers (portals) and, to
an increasing extent, brokers and (software) agents.

Methlie and Nysveen

From having a dedicated link to their customers
through branches, banks, as the ultimate product
providers, are pushed backwards in the value chain
with the danger of losing control of customer access.
Disintermediation and loss of customer control mean
reduced opportunities for fostering customer satisfaction and building strong brands. To achieve this
control, we predict that cooperation with new actors
in the value chain will be needed. This may also be
supplemented by focused acquisitions and licensing of
software products (Gordon, 1996).
To satisfy customers, it is important to listen to their
needs and desires (Sterne, 1998). Therefore, it is of
vital importance to give customers the opportunity to
respond to offers. Responding can include questions,
comments on services, suggestions for improvements,
demands, etc. and can be mediated through the e-mail
system. Another way to satisfy customers is to personalize services. Loyal customers can be obtained by
providing them with personal attention (Allen et al.,
1998). Personalized bank services may be offered by
customized information about savings, funds and credit
offers. `Frequently asked questions (FAQ), organized
discussion groups and the use of push technology to
move personalized messages to customers are other
means by which added value for the customers can be
created. Furthermore, value can be added by providing
support tools such as personal nancial systems (PFS)
to help customers manage their family economy and
nancial matters. Value-added services are assumed to
make customers more satis ed.
Brand reputation
Branding and brand reputation have been discussed
in the marketing literature over the last 10 years.
Reputation seems to be an important determinant of
loyalty when products are dif cult to evaluate
(Samuelsen et al., 1997). In markets with many
suppliers, complex services and unclear values and
preferences, rational choices are dif cult to make.
Mechanisms to simplify buying decisions are needed.
Brand reputation is one such mechanism where
information gathering and processing are reduced by
substituting attribute evaluation with heuristics the
brand. Our study indicates that brand reputation is a
signi cant determinant of loyalty, although not as
strong as customer satisfaction.
To explain why customer satisfaction seems to be a
stronger determinant of loyalty than brand reputation
we have to look at the bank services currently available from most of the Internet banks. Brand reputation is thought of as being a strong determinant of
loyalty in markets with complex products where simplifying buying decisions is needed. The bank services

383

Loyalty of on-line bank customers

available to the customers in our study may be
perceived as standardized services, although with rather
complex cost structures (fees, interest rates, etc.). This
may explain why satisfaction seems to be a more
signi cant determinant of loyalty conative loyalty in
particular than brand reputation. However, we anticipate that the importance of brand reputation will
increase as new services such as purchases of stocks,
mutual funds, loans, etc., become more common on
the Internet.
The results indicate that a strong brand is an important determinant of loyalty. `Building brand equity
requires creating a brand that consumers are aware
of and with which consumers have strong, favorable
and unique brand associations (Keller, 1998, p. 68).
To build brand equity, banks should choose brand
elements (brand, slogans and/or logo) that identify and
differentiate the brand. Furthermore, marketing programmes that include price strategy, product strategy,
communication strategy and channel strategy should
be implemented to build strong, favourable and unique
associations. In this process, alternative communication options such as media advertising (television,
radio, newspapers, magazines and the Internet), direct
response advertising, outdoor advertising, sponsorship
and event marketing should be used.
Switching costs
According to our results, switching costs have a positive
effect on affective loyalty, in contrast to the negative
relationship postulated in the hypothesis. A possible
explanation is that customers may feel that they
are saving money by being loyal to one bank instead
of focusing on the negative effects of being `locked
in. By keeping one primary banking relationship
customers save themselves from switching costs.
Customers appreciate this and two lines of reasoning
may explain this. Either customers focus on the added
value of a one-stop banking relationship or customers
regard the investment in a banking relationship as a
sunk cost, thus disregarding the negative effects of
being locked in. An appreciation of added value may
be associated with intangible assets related to personal
services, better nancial control of the account, etc.
On the other hand, the positive relationship may also
be explained by a rather unre ected attitude: either
the switching costs are regarded as insigni cant or
customers do not have the feeling of being exploited
by the bank. It might also imply that customers look
at switching barriers as a necessity something that is
inherent in the services (Andreassen, 1998). This can
be explained by the nature of the banking services
commonly available on-line today primarily accountbased services.

Although switching barriers can be built on negative

elements such as lock-in factors it seems more appropriate to build switching barriers on elements that
provide customers with added value. Many of the
loyalty programmes offered today are based on such
values, for example reduced interest rates, reduced
fees, etc. In this way, being locked in to one bank can
be perceived as a positive relationship strengthening
loyalty to the bank. Figure 2 shows that switching
barriers have a signi cant effect on affective loyalty
but not on conative loyalty. The reason for this could
be that changing banking relationships on the Internet
today is perceived as easy.
Search costs
The search costs in this study focus on the time used
to establish an Internet relationship with a bank. As
postulated in the hypotheses, the effect of search costs
on affective loyalty is found to be negative implying
that reduced search costs lead to increased affective
loyalty. This may indicate that the customers value
their time highly and that the alternative costs are
high. This can be explained by demography. Internet
bank customers are well-educated and have salaries
above average. The time they spend on searching
for product information reduces the time available for
other activities such as working or leisure. They are
not willing to spend their valuable time searching for
and evaluating alternative offers. Thus, the effect
of search costs on affective loyalty is negative and a
reduction in search costs leads to increased affective
loyalty (Andreassen, 1998).
An implication of this nding is that banks should
struggle to reduce search costs to increase affective
loyalty. Internet customers have the opportunity of
achieving reduced search costs (Bakos, 1991, 1997).
By taking advantage of technology such as search
engines and intelligent agents, search costs may be
reduced. For banks, however, such agents and engines
may imply a competitive risk. Sophisticated search
agents and engines give the customers a unique
opportunity of comparing offers of similar products
from several suppliers. If customers take advantage
of this opportunity, it will lead to pressure on prices
and pro t margins. Theoretically, this should lead
to a reduction in conative loyalty. However, the ndings of this study indicate no effect of search costs
on conative loyalty. Studies show that the services
offered by banks on the Internet are very similar
(Nettavisen, 1998). Since searching for alternative
services does not lead to signi cantly better offers, this
may explain why search costs are not an important
determinant of conative loyalty to banks on the
Internet.

384
Affective loyalty
The positive effect of affective loyalty on conative
loyalty found in this study underlines the importance
of building personal relations with customers to
increase their intention to keep on as customers in the
bank. As discussed earlier, this may be done by the
use of several communication techniques offered on
the Internet, for example e-mail. Personal attention
and personalized services will probably be of great
importance for customers attitudes towards the bank
and, through this, also of great importance for conative loyalty.

Limitations of the study

In this study we have argued for causal relationships
between the antecedents of loyalty and affective and
conative loyalty. Demonstration of causality involves
three distinct operations: demonstrating co-variation,
eliminating spurious relations and establishing the
time order of the occurrences (Frankfort-Nachmias
and Nachmias, 1992). A cross-sectional study, a
correlational design, was used in this survey. We can,
therefore, demonstrate co-variance between the variables. Elimination of spurious relations includes the
use of a homogenous sample and a set of control
variables. In this study the sample is homogeneous
with respect to the banking relationship all respondents were users of Internet banks. However, the age
of the respondents varied from 18 to 71 years, 94%
were men and 6% were women and they differed in
education. Other control variables were not measured
and could therefore not be used as co-variates. Time
order is not possible to establish by the use of a
cross-sectional study. Therefore, loyalty may have been
established prior to the measurements of the antecedent variables and not the other way around as
argued above (cf. Figure 1). However, the assumed
relations are based on theory and logic that indicate
that the antecedent variables are established before the
criteria variables.
The questionnaire was available for all customers of
the three banks using the Internet as a channel for
bank services. The customers themselves chose
whether they wanted to ll out the questionnaire or
not. This procedure leads to self-selection, which may
cause a biased sample. However, the demography of
the respondents compared to the demography of the
population of Internet users of the three banks indicates that the sample is fairly representative except for
the over-representation of men in the sample. We
analysed the model for women and men separately.
The results show that there are no signi cant differ-

Methlie and Nysveen

ences between the two groups for any of the relationships. This indicates that the over-representation of
men in the sample does not bias the results.

Concluding Remarks
`Gauging and communicating what your products
and services are worth to customers has never been
more important (Anderson and Narus, 1998, p. 53).
This proposition is even truer when it comes to
commerce in electronic markets. In this paper we have
described the results of some empirical ndings of what
determines loyalty when bank services are moved from
the market-place to on-line banking. In this study, we
have tested the determinants known from marketing
literature to hold in traditional markets to see whether
these determinants also hold in the new markets of
electronic banking. The ndings indicate that this is
true to some extent. Customer satisfaction, however,
is found to have the greatest impact, followed by brand
reputation, while switching costs and search costs have
less explanatory power of loyalty.
At the core of all successful business relationships
are satisfaction and trust. Trust can be built by brands.
Satisfaction and trust can be expressed in customer
values. There is a growing interest in being able to
express customer values in electronic commerce. Banks
need to reassess their customer value proposition when
moving into electronic markets. In addition, there is
an increasing focus on loyalty programmes in retail
banking today. The ndings of this study should be
of particular interest in forming marketing strategies
for banks in general and for these loyalty programmes
in particular.
We have outlined above some limitations of this
study. However, despite these limitations we claim that
the results found are valid and relevant. We should,
however, be aware that the ndings are associated with
the banking services available at the three banks in our
study. These services are the most common banking
services on the Internet today and, as stated above,
are primarily account-based services. With banking
services of the more `shopping types , such as buying
consumer loans, mortgages, stocks, etc., the situation
may change. Therefore, it will be interesting to do a
similar empirical study in 2 3 years time.

Acknowledgements
This study is part of a research programme on multimedia banking at the Foundation for Research in
Economics and Business Administration funded
jointly by the Norwegian Research Council and three

Loyalty of on-line bank customers

business partners: Telenor, Schibsted and Sparebanken
Hedmark. The authors wish to thank Einar Breivik,
associate professor of the Norwegian School of Economics and Business Administration, for support in
the statistical analysis of the data using LISREL.

References
Aaker, D.A. (1991) Managing Brand Equity (The Free Press,
New York).
Aaker, D.A. (1996) Building Strong Brands (The Free Press,
New York).
Allen, C., Kania, D. and Yaeckel, B. (1998) Internet World.
Guide to One-to-one Web Marketing (John Wiley & Sons,
Inc., New York).
Anderson, J.C. and Narus, J.A. (1998) Business marketing:
understand what customers value. Harvard Business
Review November December 76(6), 53 65.
Anderson, J.C. and Gerbing, D.W. (1988) Structural equation modelling in practice: a review and recommended
two-step approach. Psychological Bulletin 103(3), 411 23.
Anderson, R.E. (1973) Consumer dissatisfaction: the effect
of discon rmed expectancy on perceived product
performance. Journal of Marketing Research February X,
38 44.
Andreassen, M. (1998) Loyalty and bank services on
Internet. Thesis, The Norwegian School of Economics
and Business Administration, Bergen, Norway.
Bagozzi, R.P. and Yi, Y. (1988) On the evaluation of structural equation models. Journal of the Academy of
Marketing Science 16(1), 74 94.
Bakos, J.Y. (1991) A strategic analysis of electronic marketplaces. MIS Quarterly September 15(3), 1676 92,
295 310.
Bakos, J.Y. (1997) Reducing buyer search costs: implications
for electronic marketplaces. Management Science 43(12)
1676 92.
Brown, S.P. and Stayman, D.M. (1992) Antecedents and
consequences of attitude toward the ad. A meta-analysis.
Journal of Consumer Research 19, 34 51.
Browne, M.W. and Cudeck, R. (1992) Alternative ways of
assessing t. Sociological Methods and Research 21, 230 58.
Cardozo, R.N. (1965) An experimental study of customer
effort, expectation, and satisfaction. Journal of Marketing
Research II, 244 9.
Churchill Jr., G.A. and Surprenant, C. (1982) An investigation into the determinants of customer satisfaction.
Journal of Marketing Research XIX, 491 504.
Dick, A.S. and Basu, K. (1994) Customer loyalty: toward
an integrated conceptual framework. Journal of the
Academy of Marketing Science 22(2), 99 113.
Evans, P.B. and Wurster, T.S. (1997) Strategy and the new
economics of information. Harvard Business Review,
September October 75(5), 71 82.
Fishbein, M. and Ajzen, I. (1975) Belief, Attitude, Intention,
and Behavior (Addison-Wesley, Reading, MA).
Fornell, C. (1992) A national customer satisfaction barometer: the Swedish experience. Journal of Marketing 56,
6 21.

385
Fornell, C. and Larcker, D.F. (1981) Evaluating structural
equation models with unobservable variables and measurement error. Journal of Marketing Research 18, 39 50.
Frankfort-Nachmias, C. and Nachmias, D. (1992) Research
Methods in the Social Sciences (Edward Arnold, London).
Gordon, N. (1996) Homes for orphans. The Banker January
146(839), 68 9.
Hagel, J., Hewlin, T. and Hutchings, T. (1997) Retail banking: caught in a web? The McKinsey Quarterly 2, 42 55.
Homer, P. M. (1990) The mediating role of attitude toward
the ad: some additional evidences. Journal of Marketing
Research XXVII, 76 86.
Howcroft, B. (1987) Retail banking in the UK: a change in
organisational form. Journal of Retail B anking 9(3).
Jacoby, J. and Chestnut, R.W. (1978) B rand Loyalty
Measurement and Management (Wiley, New York).
Keller, K.L. (1998) Building, Measuring, and Managing Brand
Equity (Prentice Hall, Englewood, New Jersey).
Krishnamurthi, L. and Raj, S.P. (1991) An empirical analysis
of the relationship between brand loyalty and consumer
price elasticity. Marketing Science 10(2), 172 83.
Llewellyn, D.T. (1997) Banking in the 21st century: the transformation of an industry. Bulletin Economique et Financier
11(49) 5 27.
Macintosh, G. and Lockshin, L.S. (1997) Retail relationship
and store loyalty: A multi-level perspective. International
Journal of Research in Marketing 14, 487 97.
Methlie, L.B. (1998) A Business Model for Electronic Commerce
(Norwegian School of Economics and Business
Administration, Bergen, Norway).
Nettavisen, Z. (1998) D dt l p mellom bankene,
http://www.nettavisen.no, 13 August.
OECD (1998) The Economic and Social Impacts of Electronic
Commerce (OECD Paris).
Oliver, R.N. (1997) Satisfaction. A Behavioral Perspective on the
Consumer (McGraw-Hill Inc., New York).
Porter, M.E. (1980) Competitive Strategy (The Free Press,
New York).
Rayport, J.F. and Sviokla, J.J. (1994) Managing in the marketspace. Harvard Business Review November December
72(6), 141 50.
Reichheld, F.F. (1996) The Loyalty Effect (Harvard Business
School Press, Boston, MA).
Rosenberg, L.J. and Czepiel, J.A. (1983) A marketing
approach to customer retention. Journal of Consumer
Marketing 2, 45 51.
Samuelsen, B.M., Gr nseth, B.O. and Sandvik, K. (1997)
Customer loyalty: effects of customer satisfaction and
brand reputation BETA (1), 36 46.
Sandvik, K. and Duhan, D. (1996) The effects of performance
on brand reputation, satisfaction and loyalty, in The 25th
EMAC Conference, Budapest 14th 17th May, pp. 983 99.
Selnes, F. (1993) An examination of the effect of product performance on brand reputation, satisfaction and loyalty.
European Journal of Marketing 27(9), 19 35.
Seybold, P.V. (1998) Customer.com (Times Books, New
York).
Sterne, J. (1998) World Wide Web Marketing. Integrating the
Web into Your Marketing Strategy ( John Wiley & Sons,
Inc., New York).

386
Strader, T.J. and Shaw, M.J. (1997) Characteristics of electronic markets. Decision Support Systems 21, 185 98.
Yi, Y. (1990) A critical review of customer satisfaction, in
Review of Marketing, Zeithaml, V.A. (ed.) (American
Marketing Association, Chicago), 68 123.

Biographical notes
Leif B. Methlie is a professor of information management in the Department of Strategy and Management,
Norwegian School of Economics and Business
Administration, Bergen, Norway. He holds a BSc in
mechanical engineering from the University of
Strathclyde, UK (1962) and a degree in business
administration from Stockholm University, Sweden
(1967). Since 1974 he has conducted research on
information systems to support management and
decision makers based on management, cognitive
and systems theory. Since 1996 electronic commerce
has been his major research area and he is currently
managing several research projects concerning electronic banking and tourism. He teaches graduate

Methlie and Nysveen

courses on electronic commerce, IT-support for
management and knowledge management. He is a
former president of the Norwegian School of Economics and Business Administration (1990 1995).
Herbj rn Nysveen is a research scholar at the
Norwegian School of Economics and Business
Administration, Department of Strategy and Management, Bergen, Norway. He is a PhD student working
on his dissertation on the Internet and marketing
communication. He is currently engaged in several
research projects in electronic commerce concerning
electronic banking and tourism. He teaches a graduate
course on marketing and interactive multimedia.

Address for correspondence: Leif B. Methlie,

Norwegian School of Economics and Business
Administration, Department of Strategy and
Management, Breiviksveien 40, N-5045 Bergen,
Norway.