You are on page 1of 13

Journal of Retailing and Consumer Services 7 (2000) 33 } 45

Store loyalty: its impact on retail revenue.


An empirical study of purchasing behaviour in the UK

Simon D. Knox*, Tim J. Denison


Institute for Advanced Research in Marketing, Cranxeld School of Management, Cranxeld University, Cranxeld, Bedford MK43 0AL, UK

Abstract

In this paper, we present some signi"cant empirical "ndings about store loyalty and consumer spending in the United Kingdom
across "ve retail sectors. Our "ndings are presented at two levels: Firstly, we compared loyalty levels across retail sectors in the UK
and identi"ed that home improvement stores generate the lowest levels of consumer loyalty. Secondly, by disaggregating the data by
loyalty types, we found that, while loyal shoppers tend to have smaller monthly budgets than switchers, they spend double the amount
in their &&"rst choice'' store. All our results highlight the importance of developing a corporate approach to managing customer loyalty
in retailing. ( 1999 Elsevier Science Ltd. All rights reserved.

1. Introduction e!orts to do just this are becoming increasingly com-


mon-place in the UK through loyalty clubs, cards
Manufacturers of fast moving consumer goods have and programmes, each designed to motivate the con-
long been aware of the importance of customer loyalty to sumer to spend more in one store group and less in
their brands. This has been the focus of numerous aca- others. However, absolute customer loyalty at store
demic and business models since the 1950s, in which links level is not a realistic proposition for retail marketers.
between brand loyalty and indicators of market perfor- While a consumer, conceivably, may retain the services
mance, such as market share, have been established of the same bank throughout his/her working life, pos-
(Ehrenberg et al., 1990). In contrast, retailers have tradi- sibly at the exclusion of all others, it is inconceivable
tionally placed less strategic importance on customer that they would show comparable loyalty to a grocery
loyalty (Wrigley and Dunn, 1984c). supermarket or clothing chain. Research suggests
However, even the most powerful of business enter- that most shoppers purchase on a portfolio basis,
prises cannot ignore the natural forces of change. In the switching from store to store at will (Kau and Ehrenberg,
UK, there is little doubt that competition among major 1984). There would seem to be very few shoppers
retailers has intensi"ed (Knee and Walters, 1985; Duke, that remain brand and store loyal and these are
1991). Despite this, their need is to raise sales volumes in generally light shoppers (Ehrenberg, 1993). For the
order to spread higher "xed cost investments, arising retailer, the challenge rests in raising the degree of
from such factors as the acquisition of larger sites (typi- customer loyalty rather than winning over life-time
cally, 30,000#ft2 of sales area for grocery superstores) exclusivity.
and the introduction of electronic checkout data and Marketing investment in long-term customer relation-
stock control systems. Though store location remains the ships is "nding favour across many industries, parti-
keystone to gaining customers, there is growing belief in cularly within the service sector where, traditionally,
the value of keeping them loyal. Retail management marketing e!ort has been concentrated on attracting new
customers rather than keeping existing ones. For
* Corresponding author. Tel.: #44 (0)234 751122; fax: #44 (0)234
example, the wisdom of customer retention strategies in
751806 the banking and insurance sectors is now beyond doubt;
E-mail address: s.knox@cran"eld.ac.uk (S.D. Knox) the high initial costs borne in setting up new accounts are

0969-6989/99/$ - see front matter ( 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 9 6 9 - 6 9 8 9 ( 9 8 ) 0 0 0 3 3 - 2
34 S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45

o!set through subsequent transactions over the life-time reason, the topic of store loyalty and shopping protocol
of the relationship. Retaining customer accounts in these has begun to attract research attention (Knox and de
circumstances is the key to pro"tability (Christopher Chernatony, 1990), though brand choice at point of sale
et al., 1991). Any increase in the life-time of the customer continues to be a dominant theme.
relationship with the "rm can lead to substantial im- So what is meant by store loyalty? In essence, it
provements in business performance. One US study refers to the consumer's inclination to patronise a
found that by reducing customer defections by ten per given store or chain of stores over time. While expres-
cent, pro"tability of a credit card company rose by over sions of store loyalty and customer retention are often
120% (Reichheld and Sasser, 1990). They also found that used as a surrogate for buying behaviour patterns,
a 5% cut in defections resulted in an 85% increase in they are rarely measured and used more precisely. Since
pro"ts for a bank's branch system and a 50% increase in consumers are very unlikely to show exclusive loyalty
an insurance brokerage. to a store group, store loyalty is a relative term in
However, the relationship between customer loyalty practice and so is more di$cult to measure precisely.
and business pro"tability in the retailing of fast Many analysts choose to use the measure of repeat
moving consumer goods or durables has not been widely store visits as a convenient expression for store loyalty,
studied. despite the obvious shortcoming of the disregard for
In this paper, we present some signi"cant "nd- the actual amount being spent in the store. Who is
ings about store loyalty and consumer spending patterns the more loyal shopper? Is it the person who visits
based upon an empirical study of shopping behaviour the same store on seven out of every ten grocery shop-
among di!erent types of retail outlets in the UK. ping trips or the person who only goes there every
We address two main issues. Firstly, we explore the fourth trip but on that occasion buys 80% of their
extent to which British consumers are loyal to particular food?
stores at a time when consumer mobility and store No simple means of best measuring store loyalty has
choice are greater than ever before. Secondly, we been established. This is re#ected in the various measures
examine the relationship between the levels of store of store loyalty that have been used in past studies. Some
loyalty and consumer spending across "ve retail sectors, have been originally developed for assessing brand
to establish whether or not they are linked in the retail- loyalty (e.g. Jacoby and Kyner, 1973; Jacoby and Chest-
ing of goods, as they appear to be in the retailing of nut, 1978), others are more original and tailored to store
services. behaviour. In a sense, the fact that multiple measures
exist at all re#ects the lack of a coherent de"nition. We
have classi"ed the various measures into four categories
2. Measuring store loyalty (see Fig. 1). Patronage Measures (P) focus on the number
of shopping visits made in one store relative to others
Many past studies of consumer loyalty and buyer whilst the Switching Ratio (S) re#ects the degree of suc-
behaviour have concentrated on the dual issues of cessive visits made to the store and subsequent switching
repeat purchase and brand choice (Chat"eld et al., patterns. Budget Measures (B) is the proportion of total
1966; Ehrenberg, 1988; Goodhardt et al., 1984). expenditure devoted to a preferred store and Composite
They re#ected the overriding power that manufacturers Measures (B-P-S), the fourth category of store loyalty,
enjoyed over retailers in facilitating consumer choice
during the 1960s and early 1970s. Consumer buying
behaviour was heavily in#uenced by the pull of manufac-
turers' proprietary brands rather than the push of any
retailer.
Nowadays, this situation has reversed in the UK;
channel power clearly lies with retailers (Knox and
White, 1991). Nowhere is this more apparent than in the
British grocery sector, where signi"cant market concen-
tration has taken place over the last 26 years. In 1970,
British supermarket chains shared 44% of total national
grocery sales; this "gure had risen to 82% by 1996 (Min-
tel, 1997). Nowadays, it appears that consumer buying
behaviour is in#uenced at least as much by the retailer as
the manufacturer. A further indication of this is the
growing market share of retailer own-label brands. By
1995, approximately 39% of packaged grocery goods
sales were own-label brands (TNAGB, 1997). For this Fig. 1. A typology of store loyalty measures.
S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45 35

combines P and S with B. Each category is explained in This has been preferred by some researchers (e.g.
more detail below: Farley, 1968; Rao, 1969; Crouchley et al, 1982a, b) as it
overcomes one of the main weaknesses of the patronage
2.1. Measure 1: the patronage ratio (P) ratio, but su!ers from being less straightforward to
understand and enable quick conclusions to be drawn.
Store loyalty can be measured simply by comparing
the number of purchases made in one store for a particu- 2.3. Measure 3: the budget ratio (B)
lar product line relative to other stores (Kelley, 1967;
Thompson, 1967). Mathematically, it can be expressed as Other researchers have chosen to measure loyalty in
terms of the proportion of the consumer's total expendi-
n#1!p
P" i, ture on groceries made in the consumer's &&"rst choice''
i m store (Cunningham, 1956, 1961; Dunn and Wrigley,
where n is the number of stores available to the consumer 1984).
during the survey period, p the number of stores pa- e
i B " i,
tronised during the survey period by the ith consumer i E
and m the number of stores in the market. i
It is also the basis upon which store choice is con- where e is the amount spent in target store, or store
i
sidered in the stochastic models of buyer behaviour chain, during survey period by the ith consumer and
which have come to dominate brand choice and repeat E the total expenditure in all stores during survey period
i
buying studies (Jephcott, 1972; Wrigley, 1980; Wrigley by the ith consumer.
and Dunn, 1984a, b; Lamb and Goodhart, 1989). How- The great advantage of this measure is that it takes
ever, simply measuring store patronage over time fails to into account the relative level of spending, whereas any
capture any change of allegiance that may have taken &&analysis by purchase occasion'' (using either patronage
place during that time. For example, consumer A may or switching measures) does not di!erentiate between
regularly shop interchangeably between stores 1 and 2, &main' shopping trips and &top-up' trips.
say, over a six month period, showing little loyalty prefer- To many analysts, expenditure patterns are the most
ence between the two. Compare this to a situation in appealing single measure of store loyalty, but it is not
which consumer B shops continually at store 1 for three without its weaknesses. For example, were a consumer to
months and then becomes disillusioned with it, so aban- shop regularly for everyday clothes in store A but, occa-
doning it for store 2 during the remainder of the study sionally, buy expensive designer-wear elsewhere, an anal-
period. With regard to patronage measured over the full ysis by expenditure alone would not clearly establish
six month period, both exhibit an equal level of loyalty, underlying loyalty to store A.
but their behavioural patterns are very di!erent. These
di!erences ought to be re#ected in a true measure of 2.4. Measure 4: composite measures (B-P-S)
loyalty. This illustration demonstrates the problem of
any measure of store loyalty which is based upon fre- Recognising the shortcomings of the single measure,
quency of visit (or store patronage) alone. Clearly, this composite indices have also been used to measure store
weakness is more pronounced over longer study periods, loyalty, principally involving the number of stores visited
as consumers' long-term commitment to a particular and the proportion spent at the &&"rst choice'' store (Tate,
store diminishes and gives way to other preferences. 1961; Carmen, 1970; Enis and Paul, 1970).
The Enis and Paul measure is one such composite,
2.2. Measure 2: the switching ratio (S) which consists of an unweighted, geometric mean of
patronage, switching and budget measures applied to the
An alternative which re#ects the degree of &switching' &&"rst choice'' store or chain of stores (Burford et al.,
between favoured stores over time involves measuring 1971).
the number of successive visits or &runs' to the same store The loyalty, ¸, of the ith consumer towards a particu-
or store chain: lar store is expressed by their formula

C D
k#1!s b (k#1!s ) (n#1!p )
S" i, ¸ "100 i i i 1/3,
i m i m m
where k is the number of opportunities to switch during where b is the fraction of the budget allocated to the
i
the study period, s the number of switches from the "rst store during the survey period by the ith consumer, s the
i i
choice store or chain of stores, to some other store during number of switches from the favoured store to other
the survey period by the ith consumer and m the number stores during the survey period by the ith consumer,
of intervals, or buying occasions, in the survey period. p the number of stores patronised by the ith consumer
i
36 S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45

during the survey period, m the number of total store patterns of consumer expenditure, choice and frequency
visits during the survey period, k (or m-1) the number of of visit, the Enis and Paul Index seemed to us the most
opportunities to switch and n number of stores available enabling in these circumstances. Our full research ration-
to the consumer to purchase product category goods ale and design is discussed in the next section.
during the survey period.
Their measure is calculated as a percentage "gure with
a theoretical upper limit of 100%, indicating exclusive 3. Research design and methodology
purchasing at one store throughout the study period. In
practice, values will rarely fall below 5% since the Leaving aside this problem of loyalty de"nition and
measure is applied to the store in which the consumer measurement, there seemed to us a number of compelling
spends most over the study period (i.e. &&"rst choice'' reasons why we should carry out this new, empirical
store). Measures made around this lower level indicate study at this time.
complete indi!erence between stores and characterise Firstly, it is now nearly 15 years since the last UK
continuous switching from one store to another. study of its kind was conducted (Dunn and Wrigley,
Enis and Paul de"ned &&"rst choice'' store preference by 1984). British retailing has evolved considerably during
budget. However, one could just as easily de"ne &&"rst this period. On the one hand, consumers contrive to
choice'' on a patronage basis, that is to say measuring become more mobile and better informed, which discour-
loyalty to the store one visits the most frequently, rather ages strong loyalty to individual stores. On the other
than the one in which most money is spent. In some hand, major retailers, in general, have invested heavily in
cases, therefore, this could mean that the consumer's site location and image building in order to di!erentiate
&&"rst choice'' store may di!er according to how &&"rst themselves from competing stores and to target consu-
choice'' is gauged. Since the theme of our research was mer groups more e!ectively (de Chernatony et al., 1992).
concerned with revenue measures, we, like Enis and Paul The physical shopping environment also continues to
before us, decided to anchor the index against budget change; away from inconvenient town-centre locations
preferences and not patronage. towards planned and specialised shopping malls de-
Though the multidimensional measure o!ers a more signed to encourage more pedestrian #ow between the
balanced model of loyal behaviour, it is also less straight- stores. It is unclear what overall impact these develop-
forward to interpret directly than the single measures. ments have had on store loyalty levels.
Consumers who are deemed 60% loyal to their &&"rst In our research design, we decided to measure store
choice'' store, using an expenditure measure, means they loyalty at ten major shopping malls to re#ect contempor-
spend 60% of their budgets in that store. However, using ary UK shopping behaviour, based around the car-borne
a composite measure no such direct conclusions can be shopper.
drawn. A high loyalty rating is generally indicative of Secondly, in most previous studies the researchers
a high budget percentage spend in the &&"rst choice'' store, have focused on grocery retailing in the USA or UK. It
patronage of relatively few stores and infrequent switch- struck us that there was a real need to broaden the scope
ing. Nevertheless, in theory, a very high value for one of of existing knowledge by examining store loyalty across
these measures can o!set a low value of another, as Fig. 2 a range of retail sectors, if only to con"rm that the
illustrates. conclusions reached from grocery-based studies apply
It has been argued that the value of any single or equally to other forms of retailing. So we designed our
composite loyalty measure rests on its usefulness in ap- research to cover stores across the retail sectors, namely:
plication (Charlton, 1973). Given that our research objec-
tive was to provide a comparative measure of store f Gasolene ,lling station companies (e.g. Texaco, Esso,
loyalty levels across "ve retail sectors with very di!erent Shell).
f Grocery retailers (e.g. Safeway, Sainsbury, Tesco,
ASDA).
f Home improvement stores (known in the UK as Do-It-
Yourself or &&DIY'', i.e. stores specialising in goods for
consumers who want to make their own repairs and
alterations around the home and garden. UK retailers
of this type include B and Q, Homebase and Do-It-
All).
f Mixed retail stores (i.e. stemming from specialist-store
origins but which now cover an extensive portfolio of
products, e.g. Boots, originally a drug store, but now
selling such diverse product lines as kitchen utensils,
Fig. 2. Characteristics of the Enis-Paul measure. electrical goods and CDs).
S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45 37

f Department stores (i.e. traditional variety stores, all f Home furnishings


anchored by clothing but also o!ering an extensive f Clothing
range of other product categories from sports equip- f Food/confectionery.
ment to home furnishings, e.g. John Lewis and House
of Fraser). These categories were determined according to the
most common line items purchases surveyed in the pilot
Thirdly, previous empirical studies have generally been
exercise.
based in one geographic location, such as the work carried
By measuring loyalty in terms of category purchases,
out by Wrigley and Dunn (1984a) in Cardi!, Wales. There
we were very careful to ensure that the comparisons were
were sound methodological reasons for their narrow
based on all like purchases, irrespective of store type, to
geographic focus, not least of which must have been the
re#ect actual shopping behaviour. After all, stores do not
logistics of monitoring and recoding behavioural data.
compete on a neat, &&sector-by-sector'' basis. For instance,
Nevertheless, there is also a very strong case for drawing
multiple grocery retailers now compete head-to-head
a sample from the national population to override any
with the likes of newsagents, #orists and clothing stores.
local factors that might disguise general trends.
Using the experience of our pilot study, we designed
Accordingly, we set out to design a study that would
a structured questionnaire that was intended for use as
allow us to reach some wider conclusions about cus-
shoppers left the retail store. The logic was to use that
tomer loyalty to retail stores and to overcome many of
particular store purchase as a benchmark for recalling
the limitations of previous studies.
where, when and how often they had shopped for the
To study store loyalty on a less geographically con-
product type at each named store chain during the course
"ned basis across di!ering retail sectors, we favoured
of the last month. The questionnaire also asked questions
using a broadcast survey, by conducting personal inter-
to apportion expenditure in the named stores chains over
views with shoppers as they exited the various store
the same time period. It included questions to check and
types. Our main concern with this approach was whether
cross-validate answers previously given. For example, we
or not respondents would be able to recall their shopping
asked respondents to estimate how much they had spent
behaviour with any degree of accuracy. To address this
in a particular store chain over the previous month, only
concern, we carried out a pilot exercise, by interviewing
to ask them subsequently about how much they tended
80 shoppers in two town centre locations. Using a semi-
to spend on a typical visit to the store and how many
structure questionnaire, we asked respondents to work
times they had visited the store over the previous four
backwards from the present and to tell us chronologi-
weeks. In the case of department and mixed retail stores,
cally where and when they had shopped for goods in two
we restricted the questioning to the buying behaviour
of our "ve retail sectors. Guided by their responses, we
relating to one of the "ve speci"ed product categories,
determined that memory fatigue and inaccuracies quick-
dependent upon what exiting shoppers had just bought.
ly set in when consumers are asked to recall the details of
Store loyalty was assigned on this basis rather than
their purchasing behaviours for longer than the previous
typical shopping basket values. This was due to the
month. Additionally, we found that asking about shop-
extreme variability of expenditure within these two retail
ping activities in more than one retail sector at a time is
sectors cited by individual respondents (e.g. a T.V. one
generally too demanding on their memory and goodwill.
visit, followed by a tube of toothpaste the next, as we
In exploring the shopping protocol of department and
discussed earlier). We did not interview consumers who
mixed retail stores, our pilot study revealed that measur-
visited a store without purchasing from any of our
ing store loyalty is altogether a more di!use task than it
chosen categories.
is in other sectors where product lines are more homo-
A second pilot survey was then carried out to check the
geneous (such as groceries or gasolene). For some prod-
viability of our protocol and questionnaire design. This
uct lines, such as cosmetics, respondents might be
involved interviewing a further 80 shoppers in the same
relatively loyal to a department store, but for many other
two shopping centres. In each instance the required data
lines sold there, they may be no more than occasional
was collected without any ambiguity or evident fatigue
buyers. In these types of stores, we concluded, therefore,
on the shopper's part. Satis"ed that the pilot studies had
that it was less meaningful to consider store loyalty, per
served their purpose of developing a survey method that
se, and more appropriate to consider loyalty concomi-
would provide us with reliable and accurate information,
tant with the purchase of products within particular
we then proceeded to the main "eldwork with a 25-item
categories. For reasons of simplicity of research manage-
questionnaire.
ment and analysis, we decided to limit the measurement
In our attempt to develop a nationally-representative
of store loyalty in these two sectors to "ve categories.
pattern of store loyalty, 750 consumers were interviewed
They were:
across ten major shopping centre sites in Britain. Prior to
f Personal care products this, managers at the main shopping malls in each of the
f Leisure goods ten retail regions were identi"ed and contacted. The "nal
38 S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45

Table 1 measure been chosen, we would have become dependent


UK Shopping mall interview locations on a small number of di!erent inputs which would have
signi"cantly increased the risk of inaccurate measure-
1. Liberty II, Lakeside, Thurrock
2. Milton Keynes Shopping Centre
ment. We subsequently tested the association and con-
3. Queensgate, Peterborough vergent validity of the three measures which proved to be
4. Fosse-Park, Haymarket, Leicester highly satisfactory: a high score with one measure was
5. Broadmead, Bristol met by high scores using the other two measures. The
6. St Davids, Cardi! mean loyalty level of shoppers towards their &&"rst
7. Merry Hill, Dudley
8. Metro Centre, Gateshead
choice'' store was calculated to be 60.4% with a standard
9. Arndale, Manchester deviation of the 16.1%. Like in the original Enis and Paul
10. Plaza, East Kilbride study, our spread of loyalty values approximate to a nor-
mal distribution. Therefore, we are able to con"rm that
the Enis and Paul Index can reasonably discriminate
interview location in each region was dependent upon various degrees of store loyalty among shoppers.
willingness of these managers to co-operate. In the event, It is interesting to make a brief comparison between
seven of the shopping malls in the sample constituted the the original Enis and Paul results from the US in 1970
principal site in the region and the remaining three were and our own data from the UK. Their study of grocery
second ranking by turnover in their respective regions retailing produced an average loyalty rating of 70.1%
(Table 1). Twenty-two of the consumer interviews were and a standard deviation of 16.2%, a remarkably similar
subsequently rejected for lack of consistency in shopping spread of values to our own (SD"16.1%). At the disag-
information provided by the respondents (based on the gregate level, the 161 shoppers in our study who were
cross-validation questioning). questioned about their grocery shopping behaviour, the
Sampling of individual consumers was conducted on average loyalty level to their &&"rst choice'' food store was
a quota basis; the target quota speci"cations related to found to be 60.7%. However, it would be improper to
location, (75 respondents per site), retail sector (150 re- make any categorical inferences from the comparison
spondents per sector) and socio-economic grouping1 (A: between the two studies since there are important
minimum of 100 respondents; B: minimum of 100 respon- methodological di!erences between research designs:
dents; C : minimum of 150 respondents; C : minimum of Enis and Paul calculated store loyalty levels over a ten
1 2
150 respondents; D and E: minimum of 100 respondents). week period from diary panel data, while our ad hoc
In all other respects, the survey comprised a representa- survey data recalled shopping behaviour over the last
tive cross-section of shoppers vis-à-vis their character- four weeks. Further, previous research has found that
istics (e.g. gender, age, family size) and their shopping store loyalty decreases over time (Wrigley and Dunn,
routines. Interviews, which lasted approximately 15 min, 1984b). They found an 11% reduction in loyalty to
were conducted from Monday to Saturday and through- grocery stores over a six week period compared to their
out the course of the day, which included late night one week measure and a 15% di!erence over 24 weeks.
openings. The interview team consisted of the authors So, although their research suggests that the rate of
and three ex-students each with an MSc in Marketing loyalty decline stabilises, we can only surmise that, were
from Cran"eld University. we to have measured store loyalty over the same time
Our "ndings from the survey are outlined in the fol- period as Enis and Paul, our average store loyalty "gure
lowing sections. would have been still lower, indicating an even wider
temporal drift in loyalty behaviour. We can fairly readily
begin to rationalise why it might be that UK grocery
4. How loyal are consumers in the 1990s? shoppers in the 1990s appear less loyal to their &&"rst
choice'' store than were their American counterparts
The Enis and Paul Index was used to calculate during the 1960s. Despite the e!orts by UK retail man-
a loyalty level for each of the 728 survey respondents, agement to engender loyalty among their customers,
based on their &&"rst choice'' store (in budgetary terms). greater choice of store, higher mobility levels and, per-
Part of the appeal of using the Enis and Paul Index lay in haps, even the transition to impersonal, self-service sys-
the fact that it was constructed around three separate tems may have encouraged consumers to become more
inputs, each calculated from independent strings of ques- "ckle in their shopping behaviour.
tions collected during the interview. Had a single
4.1. Are shoppers more loyal to stores in some
retail sectors than others?
1In the UK, the population is grouped according to "ve socio-
economic classi"cations: A"top professionals, B"professionals,
C "white-collar workers, C "skilled manual workers, D"semi- Whilst our study design prevented us from establishing
1 2
skilled workers, and E"unemployed and senior citizens. whether individuals have di!ering loyalty pro"les across
S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45 39

Fig. 3. Store loyalty by retail sector.

retail sectors, we have been able to make comparisons acy could have had a dulling e!ect on consumer loyalty
across sectors at an aggregate level. to individual home improvement stores in the UK.
Our survey shows that store loyalty levels in the UK Secondly, home improvement retailers have also been
across the grocery, mixed retail and gasolene sectors are failing to di!erentiate themselves through store image
very similar and do not vary signi"cantly (see Fig. 3). and perceived value. During our interviews, consumers
t-tests showed no signi"cant di!erences from the overall commented that they were not usually conscious of the
mean. Why this might be so is very intriguing in itself. In particular home improvement store they were in; all the
all three sectors, consumers visit these stores regularly retailers appeared to share very similar merchandising
and, in doing so, they are developing habit-forming pur- practices which failed to create in-store individuality.
chase patterns. Over time, they conceivably become con- Judging by their recent television advertising campaigns,
ditioned to the outlets they frequent. By contrast, British these retailers seem content to compete on &&bleeding
consumers visit home improvement and department edge'', price-slashing tactics, rather than by creating
stores less often. Interestingly, it is in these two sectors a &&value added'' proposition. On the evidence of our
that loyalty levels are signi"cantly di!erent (with asso- research, it would appear that this approach fails to
ciated probability levels of 0.001 and 0.05 respectively). create strong loyalty ties with consumers. It seems some-
Loyalty among consumers towards their &&"rst choice'' what ironic to acknowledge that retailers in this sector
home improvement stores is signi"cantly lower than have also been pioneers of store loyalty cards in the UK.
for the other sectors. It would seem that UK retailers It remains to be seen whether their use of this devise at
in this sector are struggling to gain any competitive a tactical level (it has to be viewed separately and as
advantage over one another, and failing to di!erentiate a slightly dissonant activity to the overall discount strat-
themselves. egy), will be su$cient to instill customer loyalty to the
We can suggest two primary reasons for this. Firstly, store over any signi"cant length of time.
the 1970s were boom years in the home improvement By contrast, department stores attracted stronger
market in the UK. This encouraged rapid and wide- loyalty among respondents than any other retail sector.
spread construction of home improvement superstores, This "nding is intuitively appealing since shopping
located in out-of-town developments, with competitors in department stores has certain elitist and psycho-
clustered in close geographical proximity. This has gener- social qualities, rather like a &&lifestyle badge'' that con-
ally minimised the opportunity for site advantages be- sumers are proud to wear. We associate this more with
tween retailers within the sector. Location has long been consumer socio-economic and psychological character-
recognised as a key ingredient to retailing success and istics than with either environmental or marketing reasons.
closely associated with repeat buying behaviour (Brown Department stores tend to sell many leading brand
1991). The absence of any signi"cant locational suprem- lines, and resist own-label derivatives. Whether this is
40 S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45

a contributing factor in helping them achieve higher much by elitist values, perhaps, as the promise of quality
loyalty levels is a moot point and one which justi"es produce. Other loyalties lay in between these
future exploration. Interestingly, it was the one retail dichotomous extremes.
sector where respondents did not list location as being In this section we have re-a$rmed the value of the Enis
the driving force for shopping at one department store in and Paul Index as a means of measuring store loyalty
preference to another. It appears that convenience and of discriminating consumers' shopping habits. We
(through location) is not a consumer priority in this have demonstrated that store loyalty levels do di!er
context, which challenges Reilly's long standing law or between certain retail sectors and product categories. We
retail gravitation (1929) as being too simplistic, as other have also shown that shoppers are particularly prone to
researchers have also done in the past (Thompson, 1971). switching between stores in the home improvement sec-
Re-analysis of our database by product category across tor and between mixed retail/department stores when
department and mixed retail stores, shows that store buying food. In contrast, loyalty to department stores is
loyalty levels do not di!er markedly when calculated on particularly high with regard to certain product catego-
clothing, leisure goods and home furnishings purchases ries, such as personal care products.
(Fig. 4). Our main purpose has been to establish whether store
In the case of personal care products, respondents were loyalty levels do di!er among retail sectors rather than to
noticeably more store-loyal in their buying habits. Per- explain why levels might di!er. The why of loyalty is
haps people tend to frequent the stores in which they a topic which has been addressed in the literature (e.g.
know particular brands are sold or specialist advice can Goldman, 1977) and is deserving of more discussion than
be sought. Whatever the root-causes may be (on which we are able to devote to it in this paper. However, it is
there is no known research), personal care products seem perhaps appropriate to refer to the principal factors that
to be a store loyalty generator among mixed retail and Enis and Paul (1970) identi"ed as being the main deter-
department stores. minants of store loyalty which also seem to have validity
For those buying food/confectionery from mixed retail in this study. These are: environmental conditions; con-
and department stores, loyalty to their &&"rst choice'' store sumer socio-economic characteristics; consumer psycho-
was particularly low. Nevertheless, there was noticeably logical characteristics; and retailer marketing strategies.
more spread around the mean loyalty level. A possible In the next section where we discuss patterns of expen-
explanation can be o!ered from unstructured customer diture by consumer loyalty types, we are constrained by
responses during interview: some, it appeared, bought the same limitations in our methodology. Consequently,
food there as an occasional treat or as an emergency we also seek to quantify and post signi"cant di!erences
fall-back, but not as part of their regular routine. Others in behaviours rather than to qualitatively explain our
favoured such stores as a matter of course, attracted as observations.

Fig. 4. Store loyalty in department and mixed retail stores by product class.
S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45 41

5. Are loyal shoppers more pro5table to the retailer?

In other industries, particularly in the services sector,


a clear link has been made between the gearing e!ect of
customer loyalty and account pro"tability, where modest
improvements in customer retention can signi"cantly
improve business pro"tability (Reichheld and Sasser, 1990;
Buchanan and Gillies, 1990, Reichheld, 1996).
We set out to establish whether similar conclusions
can be reached in retailing. Given that the Enis and Paul
Index includes a budget ratio input, we would expect
loyal shoppers to spend relatively more of their budget Fig. 5. Monthly sector spending levels for loyal and switcher shoppers.
(%) in their &&"rst choice'' store than those less loyal.
However, whether these consumers spend more in
absolute terms (C) depends on the size of their C budget
(i.e. the total amount they have to spend across all stores distribution of loyalty values. We then labeled the top
or by store category). Previous studies have found that third as &&loyal'' shoppers and the bottom third as
store loyalty is seemingly independent of the total &&switchers'' and compared their monthly sector spending
amount consumers have to spend. However, in each case, levels. We found that loyal shoppers have approximately
these studies have been con"ned to the grocery sector seven per cent less to spend per month on average (across
(Cunningham, 1961; Enis and Paul, 1970). In the past, the sample of 728) whilst switchers had thirteen percent
researchers have expressed surprise at this "nding (Dunn more (Fig. 5). A t-test indicates the di!erence to be signi"-
and Wrigley, 1984) and, like them, we reasoned that one cant (p"0.05).
might reasonably expect store loyalty and total expendi- When our data are disaggregated across retail sectors
ture to be related. and product categories, this pattern is generally repeated;
a signi"cant di!erence in monthly spending level (C)
5.1. Store loyalty and monthly spending levels between loyal shoppers and switchers is found in all but
the gasolene retail sector where little di!erence was detec-
In direct contrast to these historical "ndings, for the ted. This pattern of spending was found to be most
"rst time we are able to report data which show an noticeable in mixed retail stores where switchers spent
association between store loyalty and the allocated three times as much as loyal shoppers each month.
monthly budget within a retail sector. Across the sample In direct contrast to this general trend and as the
as a whole, the linear correlation co-e$cient is low "ndings from the correlation analysis suggest, loyal
(r"!0.103) but, nonetheless, statistically signi"cant (at grocery shoppers are found to spend signi"cantly more
the 0.05 level). If the total sample is broken down into the per month than the switchers. On average, they spend
"ve retail sectors, the same inverse association is found 50% more on grocery products!
within four of the sectors, being strongest and most
signi"cant in mixed retail (r"!0.295) and home im- 5.2. Store loyalty and budget allocation
provement shopping (r"!0.303). Since the correlation
co-e$cient is negative, it signi"es that the higher the While we have shown that generally loyal shoppers
consumer's monthly budget (C) for the sector becomes, tend to have smaller monthly budgets, by all expendi-
the lower the level of loyalty attached to his/her &&"rst ture-based measures of loyalty we might expect loyal
choice'' store. shoppers to spend proportionately more of their budget
In the one remaining sector, the grocery sector, we found (%) in their &&"rst choice'' store than switchers. Not only
the relationship to be positive and signi,cant (r"0.239) for did we "nd this to be the case in this research, we also
grocery shopping, bucking the general trend found across found that the degree of di!erence was surprisingly large,
all other retail sectors. Here, the greater the consumer's which is in line with Enis and Paul's observations of
monthly budget for groceries, the more loyal they are shopping behaviour in the 1960s. Across the two groups,
likely to be towards their &&"rst choice'' store. This dispar- loyal shoppers allocate over twice as much of their
ate association helps explain the relatively low negative monthly budget (%) than switchers to their &&"rst choice''
correlation co-e$cient for the sample as a whole. store, as Fig. 6 indicates. Although the general trend is
In order to explore how &&"rst choice'' store loyalty and not unexpected, we were surprised at the magnitude of
spending level varies by shopping behaviour, we sub- disparity. Whilst we can only speculate as to why loyals
divided the sample into three loyalty bands. Each of these devote ninety percent of their monthly spend to a store
bands contained approximately equal numbers of re- chain, it is likely to be due to a psychological predisposi-
spondents as there were no natural break points in the tion, conditioned by location, store atmospherics, range,
42 S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45

Fig. 6. Percentage of monthly spend in &&"rst choice'' store for loyal and switcher shoppers.

Fig. 7. Monthly spending levels in &&"rst choice'' stores for loyal and
switcher shoppers.

service and value for money. In other words, the intan-


gible bene"ts associated with store selection have a sig- Fig. 8. The double indemnity e!ect in grocery retailing.
ni"cant part to play in building loyalty for some
shoppers.
This scale of di!erence was sustained across all "ve sectors. Rather than counter each other, the two trends,
retail sectors and product categories without exception. therefore, work in partnership: larger budgets and larger
So we have been able to establish that loyal shoppers proportions of budgets are spent by store-loyal cus-
tend to spend less in total (C) over the course of a month tomers in the grocery sector. This double leveraging
by sector (with the important exception of grocery store e!ect in grocery retailing we have termed the double
shopping), but allocate more of their budget (%) to their indemnity e+ect. It signi"es that customers loyal to a par-
&&"rst choice'' store. Combining these two contra-trends, ticular grocery store tend to spend up to four times as
we are able to conclude that loyal shoppers generally much in this store as switchers (see Fig. 8). Why grocery
spend more per month in absolute terms (C) as well as retailing should stand out in this way may be due to their
relatively (% of budget allocated) in their &&"rst choice'' success in market positioning and customer segmenta-
stores (Fig. 7). tion practices. Generally speaking, they deploy very
In fact, our study shows that loyal shoppers tend to sophisticated marketing techniques which may give them
spend twice as much as switchers in their &&"rst choice'' the edge on other retailers in targeting and serving cus-
store chain. It is our belief that this "nding is of great tomers, thus giving them less reason to shop around.
strategic importance to retailers in general and grocery
retailers (and manufacturers) in particular. In the latter
case, the loyalty bene"t is leveraged further due to the 6. Implications for retailers
fact that loyal grocery shoppers, as we pointed out
earlier, tend to have larger monthly budgets than fellow Although our "ndings are substantially more conclu-
shoppers, not smaller budgets as we found in the other sive and broader ranging than those of the Enis and Paul
S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45 43

study in certain respects, we are bound by the same we made no attempt to address the temporal nature of
observation which seems both timeless and acultural: loyalty in this study. Loyalty seems to diminish over
loyal customers spend substantially more money in their time, but the actual dynamics of store loyalty erosion
"rst choice stores than do switchers. Furthermore, as remains completely unexplored here. Fourthly, we have
these loyal customers are likely to be no more expensive not even begun to address individual consumer loyalty
to serve, we can realistically conclude that loyal custo- pro"les across retail sectors, nor by site location.
mers are potentially more pro"table to retailers. On the With regard to the "rst two limitations cited, we be-
evidence of this research, British retailers are fully justi- lieve that the data are signi"cantly robust to warrant
"ed in their attempts to increase customer loyalty in their further analysis and, consequently, will be the subject of
stores. Quite how successful card schemes are in en- future communications. Both the other limitations we
couraging store loyalty is quite a separate issue and one have mentioned highlight the frailty of our study and
which is beyond the scope of this paper. o!er signi"cant challenges to the research community at
Since our study suggests that loyal customers are more large.
pro"table, retailers would bene"t from developing their
marketing strategy around either converting more of
their shoppers into loyal customers or concentrating 8. Summary conclusions and future research directions
their e!orts on satisfying the needs and wants of their
loyal customers. Such a loyalty-reward strategy would, In this paper we have discussed two main issues arising
nevertheless, only be of strategic interest if loyal custo- from our empirical study of store loyalty and spending
mers were to comprise a substantial segment of the cus- patterns in the UK. Firstly, we have found that loyalty
tomer base and share common traits to make them levels are surprisingly similar across a number of retail
accessible beyond their in-store behaviour. For instance, sectors in the UK and any top-league British retailer in
whether loyal shoppers share similar socio-graphic char- the grocery, gasolene or mixed retail sector can expect to
acteristics and shopping protocols. In the absence of such enjoy a 60% loyalty level across a composite measure of
an analysis, customer retention strategy is likely to be budget, patronage and switching factors. We have con-
developed around loyalty programmes and frequent cluded from comparisons with previous studies, that
shopper promotions. While both these tools are currently store loyalty levels have fallen somewhat over the last 25
enjoying considerable interest at an operational level in years and highlight greater mobility, store choice
the UK, the strategic approach to customer retention, and awareness as possible root-causes. The one sector
based on di!erentiation through behavioural segmenta- in the UK to su!er from particularly low levels of
tion, is really only being explored by grocery retailers. store loyalty is home improvement retailing. These
The opportunity to communicate directly with con- retail marketers resort to price discounting on the
sumers using this type of swipe-card mechanism enables one hand, to try and secure competitive advantage,
the development of one-to-one relationships between while on the other, they are using loyalty schemes which
retailer and customer. We strongly believe that this are unlikely to be triggered primarily on the basis of
approach has the potential to be a far more cost-e!ective price cues.
means of engendering store loyalty, rather than continu- The second issue addressed in this paper is the rela-
ing to use mass-media advertising campaigns which are tionship between store loyalty and customer expenditure
inherently less selective. which our study identi"es for the "rst time to be signi"-
cantly associated. Thus, while we have found that loyal
shoppers generally spend less in total per month (C) by
7. Limitations sector than switchers, they do, however, spend more of it
} about twice as much } in their &&"rst choice'' store. This
While we are con"dent about the conclusions we is particularly true in gasolene and home improvement
have drawn and what their implications are for UK retailing.
retailers, our work has many limitations that have In grocery retailing, not only do loyal shoppers
to be recognised. Firstly, we have rather arti"cially desig- allocate proportionally more of their budget to their
nated the loyalty label without exploring salient determi- &&"rst choice'' store (%), as in other sectors, but they also
nants. Secondly, our data were dependent upon spend more on groceries (C) per se than their fellow
the accuracy of shopping behaviour recall by consumers shoppers. In practice, the combined e!ect means that
rather than from diary panels adopted by the previous loyal shoppers can spend up to four times as much (C) in
researchers cited. Nonetheless, after testing the their &&"rst choice'' store as their switcher counterparts.
accuracy of recall during the pilot stages and by We have designated this anomaly the double indemnity
incorporating a cross-validation mechanism into the e!ect.
main study, we are of the belief that the data accu- These loyalty-spending "ndings lead us to the inevi-
rately recorded buying behaviour patterns. Thirdly, table conclusion that loyal consumers represent the most
44 S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45

pro"table core of shoppers in Britain, a fact that has Charlton, P., 1973. A review of shop loyalty. Journal of Market Re-
been established already in industry sectors outside search Society, 15 (1), 35}51.
retailing. Chat"eld, C., Ehrenberg, A.S.C., Goodhardt, G.J., 1966. Progress on
a simpli"ed model of stationary purchasing behaviour. Journal of the
Though this paper is based upon an empirical study of Royal Statistical Society Series A 129, 317}67.
shopping behaviour in the UK, we believe that many of Christopher, M., Payne, A., Ballantyne, D., 1991. Relationship Market-
the patterns, associations and underlying conclusions ing: The Integration of Quality, Customer Service and Marketing.
drawn have wider applicability. In other service sectors, Heineman London.
such as banking and direct retailing, the pro"tability Crouchley, R., Pickles, A., Davies, R.D., 1982a. Dynamic models of
shopping behaviour: testing the linear learning model and some
gains from reallocating resources to improve marginal alternatives. Geogra"ska Annaler B (64), 27}33.
customer retention and loyalty are considered universal Crouchley, R., Pickles, A., Davies, R.D., 1982b. A re-examination of
(Reichheld, 1996). Burnett's study of Markovian models of movement. Geographical
Less certain is whether the similarities and di!erences Analysis 14, 260.
we have found in loyalty levels across di+erent retail Cunningham, R.M., 1956. Brand loyalty } what, where, how much?.
Harvard Business Review 34, 116.
sectors in the UK can be transposed to other countries Cunningham, R.M., 1961. Customer loyalty to store and brand. Har-
where competition and market structures are di!erent. vard Business Review 40, 127}137.
Further research into loyalty patterns across a larger de Chernatony, L., Knox, S.D., Chedgey, M., 1992. Brand pricing in
number of retail sectors in other geographical settings a recession. European Journal of Marketing 26 (2), 5}14.
will help resolve this issue. Duke, R., 1991. Post-saturation competition in UK grocery retailing.
Journal of Marketing Management, 7 (1), 63}75.
Our study clearly provides retailers with an incentive Dunn, R., Wrigley, N., 1984. Store loyalty for grocery products: an
to re-examine ways to make loyal customers more loyal empirical study. Area 16, 307}314.
to their stores. The research imperatives must be, "rstly, Ehrenberg, A.S.C., 1993. If you're so strong, why aren't you bigger?.
to explore the exact nature of the loyalty/pro"tability Admap 13}14.
relationship, in order to appreciate the gearing e!ect that Ehrenberg, A.S.C., Goodhardt, G.J., Barwise, T.P., 1990. Double jeo-
pardy revisited. Journal of Marketing 54, 82}91.
a marginal change in store loyalty can bring to pro"tabil- Ehrenberg, A.S.C., 1988. Repeat Buying: Fact, Theory and Applica-
ity. Secondly, we must establish the means to access loyal tions, 2nd ed., Oxford University Press, New York.
customers at the strategic level to in#uence the &&"rst Enis, B.M., Paul, G.W., 1970. Store loyalty as a basis for market
choice'' decision, thereby acting synergistically with segmentation. Journal of Retailing 46 (3).
loyalty programmes that operate at store level. This Farley, J.V., 1968. Dimensions of supermarket choice patterns. Journal
of Marketing Research 5, 206}8.
means exploring new consumer segmentation ap- Goodhardt, G.J., Ehrenberg, A.S.C., Chat"eld, C., 1984. The Dirichlet
proaches based upon purchasing behaviours and prefer- model } a comprehensive model of buying behaviour. Journal of the
ences on a portfolio basis across retail sectors and within Royal Statistical Society Ser A, 147.
product categories. The developments in data fusion and Jacoby, J., Chestnut, R.W., 1978. Brand Loyalty: Measurement and
information warehousing across retail markets will make Management. Wiley, New York.
Jacoby, J., Kyner, D.B., 1973. Brand loyalty vs. repeat purchasing
this a realistic research goal for academics and retail behaviour. Journal of Marketing Research 5, 1}9.
management alike. Jephcott, J.St G., 1972. Consumer loyalty } a fresh look. Proceedings of
Despite our limitations, we believe the paper makes the Annual Conference. The Market Research Society, London.
a timely contribution to knowledge by underlining the Kau, A.K., 1981. Patterns of store choice. Ph.D. thesis, London University.
importance of loyal customers in the retailing of fast Kau, A.K., Ehrenberg, A.S.C., 1984. Patterns of store choice. Journal of
Marketing Research XX1, 399}409.
moving consumer goods and durables. It has been based Kelley, R.F., 1967. Estimating ultimate performance levels of new retail
on a substantial, British empirical study and brings atten- outlets. Journal of Marketing Research 4, 13}19.
tion to a subject that we believe will continue to be the Knee, D., Walters, D., 1985. Strategy in Retailing: Theory and Applica-
battleground of retailing throughout the 1990s both in tion, Philip Allan, Oxford.
Europe and North America. Knox, S.D., de Chernatony, L., 1990. A buyer behaviour approach to
merchandising and product strategy. International Journal of Retail-
ing and Distribution Management 18 (6), 21}30.
Knox, S.D., White, H., 1991. Retail buyers and their fresh produce
suppliers: a power or dependency scenario?. European Journal of
References Marketing 25 (1), 40}52.
Lamb, T.J., Goodhart, G.J., 1989. A comparison of brand loyalty and store
Brown, S., 1991. Retail location } the post hierarchical challenge. loyalty. Working Paper No. 93, City University Business School, London.
International Review of Retail Distribution and Consumer Research Mintel, 1997. Food Retailing. Mintel International, London.
1 (3), 367. Rao, T.R., 1969. Consumers' purchase decision process: stochastic
Buchanan, R., Gillies, C., 1990. Value managed relationships: the key to models. Journal of Marketing Research 6, 321}29.
customer retention and pro"tability. European Marketing Journal, Reichheld, F.F., 1996. The Loyalty E!ect. Harvard Business School
8 (4), 523}525. Press, Boston, MA.
Burford, R.L., Enis, B.M., Paul, G.W., 1971. An index for the measure- Reichheld, F.F., Sasser, W.E. Jr., 1990. Zero defections: quality comes to
ment of consumer loyalty. Decision Sciences 2, 17}24. services. Harvard Business Review 105}111.
Carmen, J.M., 1970. Correlates of brand loyalty; some positive results. Tate R S., 1961. The supermarket battle for store loyalty. Journal of
Journal of Marketing Research 7, 67}76. Marketing 25, 8}13.
S.D. Knox, T.J. Denison / Journal of Retailing and Consumer Services 7 (2000) 33}45 45

Thompson, B., 1967. An analysis of supermarket shopping habits in Wrigley, N., Dunn, R., 1984b. Stochastic panel-data models of
Worcester, Massachusetts. Journal of Retailing 43, 17}29. urban shopping behaviour: 2 Multi-store purchasing patterns
TNAGB Superpanel, 1997. Own-label share of trade, 1977}1995. and the Dirichlet model. Environment and Planning A 16,
Westgate, London. 759}778.
Wrigley, N., 1980. An approach to the modelling of shop-choice patterns: Wrigley, N., Dunn, R., 1984c. Stochastic panel-data models of
an exploratory analysis of purchasing patterns in a British city. In: urban shopping behaviour: 3 The interaction of store choice
Herbert, D.T., Johnson, R. (Eds), Geography and the Urban Environ- and brand choice. Environment and Planning A 16,
ment. Progress in Research and Applications, vol. 3. Wiley New York. 1 221}1 236.
Wrigley, N., Dunn, R., 1984a. Stochastic panel-data models of urban Wrigley, N., Dunn, R., 1988. Models of store choice and market analy-
shopping behaviour: 1 Purchasing at individual stores in a single city. sis. In: Wrigley, N. (Ed.), Patterns of Store Choice, Store Location
Environment and Planning A 16, 629}650. and Market Analysis. Routledge, London.

You might also like