Download Centre for Marketing - Research

Centre for Marketing
CONSUMER LOYALTY MARKETING IN
REPEAT-PURCHASE MARKETS
Mark D. Uncles*
Grahame R. Dowling
Kathy Hammond
Angelo Manaresi
Centre for Marketing Working Paper
No. 98-202
October 1998
Mark Uncles is Professor of Marketing, University of New South Wales, Sydney, Australia.
Grahame R. Dowling is Associate Professor of Marketing, Australian Graduate School of Management,
University of New South Wales, Sydney, Australia. Kathy Hammond is Director of the Future Media
Research Programme and a Research Fellow, London Business School, UK. Angelo Manaresi is a
Lecturer in Marketing, Dipartimento di Discipline Economico-Aziendali, Universita di Bologna, Italy.
*Correspondence to: Professor Mark Uncles, School of Marketing, University of New South Wales,
Sydney 2052, Australia, Ph: +612-9385-3510, Fax: +612-9663-1985, email: Error! Bookmark not
defined.
Acknowledgments
We thank Jack Cadeaux, Robert East, Byron Sharp, and Chris Styles for their constructive suggestions.
Also all those who commented on an earlier draft of this paper presented at the MSI Workshop on
Fundamental Issues and Directions in Marketing, with special thanks to the organisers, George Day and
David Montgomery.
London Business School, Regent's Park, London NW1 4SA, U.K.
Tel: +44 (0)171 262-5050 Fax: +44 (0)171 724-1145
Error! Bookmark not defined. Error! Bookmark not defined.
Copyright © London Business School 1998.
Consumer Loyalty Marketing in Repeat-Purchase Markets
Abstract
Extensive empirical research indicates that consumer loyalty in competitive repeat-purchase markets is
shaped more by habits, reinforcement effects, and situational influences than strongly-held attitudes.
This commonly results in divided-loyalty among a set of brands for individual consumers and doublejeopardy patterns among competing brands. To be effective, loyalty marketing initiatives need to take
account of these factors to strengthen preferences and reinforce purchasing habits. Failure to do so may
mean consumers come to see initiatives as irrelevant or as merely promotional rewards.
Keywords
Consumer Loyalty, Habit, Reinforcement, Divided Loyalty, Loyalty Programs
1
Consumer Loyalty Marketing in Repeat-Purchase Markets
Building and maintaining consumer loyalty lies at the heart of the marketing concept. Pursuit
of this goal implies that the organisation is not only interested in making the sale or achieving trial
purchase at any cost, but is focussed on how to achieve longer-term profitability through repeatpurchase and consumer retention. It is this loyalty that directly provides a continuing revenue stream for
the organisation. Managers of superbrands such as Nestle, Sony, Hitachi, BA, United, Disney, Kodak,
Kellogg's, Coca-Cola, Foster's, Budweiser and McDonalds are well aware of the importance of
maintaining and developing consumer loyalty. The longevity of these brands is a testament to their
success. Academic interest extends back to the earliest days of the discipline and continues into present
times (Copeland 1923; Jacoby and Chestnut 1978; Webster 1992; Dick and Basu 1994; Mellens et al.
1996; Oliver 1997; Uncles and Laurent 1997).
A major reason for this continuing interest is the belief that loyal consumers offer a wide range
of benefits, quite apart from providing a stream of revenue. Specifically, an organisation with loyal
consumers might expect to benefit from:
•
Reduced consumer acquisition costs and more cost-effective servicing costs (Blattberg and
Deighton 1996; Reichheld 1996; Palmer and Beggs 1997).
•
Greater consumer resistance to counter-offers (notably price-based offers) or counter-arguments
(from advertising and other external sources) (Belch 1981; Tellis 1988).
•
Fewer reasons for consumers to engage in extended information search among alternatives
(Holbrook 1978).
•
Word-of-mouth support and endorsement, which strengthens the commitment to a brand and
might even become a form of advocacy (Oliver 1980; Christopher et al. 1991; Bone 1995).
2
•
Possible brand extensions and line extensions that meet the needs of existing loyal consumers and
are also attractive to the trade (Aaker 1991; de Chernatony and McDonald 1998; Keller 1998).
•
The capacity to cross-sell related products and services (Pearson 1996).
•
Opportunities for new product testing among loyal lead-users (von Hippel 1986).
As organisations attempt to realise these "hoped-for" benefits the practice of marketing is
changing profoundly. Many academics and practitioners are coming to view the primary focus of our
discipline as "loyalty marketing". Three defining characteristics of this viewpoint are particularly
noteworthy:
First, brand management has developed to take advantage of new loyalty marketing vehicles.
To build and maintain consumer loyalty, brand managers are supplementing their mass-media
advertising with more direct communications, through direct and interactive methods, internet
communications, and other innovative channels of distribution (Pearson 1996; Peppers and Rogers
1997; Barwise and Hammond 1998). Simultaneously, however, brand managers are having to face
more threats to their brands, especially parity responses from competitors.
Second, enabling technologies for identifying and communicating with consumers have
developed dramatically in recent decades. Examples include automated data capture; increased
computer storage, retrieval and processing capabilities; and computer-mediated interactivity with
consumers. As the costs of these technologies fall, there has been a rapid growth in direct marketing,
targeted marketing, and one-to-one marketing. This has given added poignacy to familiar calls for
marketers to stay close to their consumers (Peppers and Rogers 1993; Petrison et al. 1993; Deighton
1994).
Third, an outcome of these two changes has been the development of so-called "loyalty
programs". These are now pervasive across industries as diverse as grocery retailing and air travel.
They take a variety of formats, based on some combination of incentives, rewards, membership and
3
other "relationship" benefits. Typical drivers behind the launch of these programs come from branding
goals (such as the desire to strengthen the brand by forming a relationship with consumers); consumer
goals (by rewarding and recognising loyal consumers, or by offering additional sources of value); and
competitive response (namely, the attempt to differentiate parity brands, pre-empt the competition, or
imitate competitors). The applied marketing literature suggests that there are very high expectations for
these loyalty-building initiatives (Reichheld and Sasser 1990; Nalebuff and Brandenburger 1996;
Reichheld 1996).
Speculation by academics, consultants and businesspeople is that marketing in the new century
will be very different from when much of the pioneering work on consumer loyalty was undertaken
(e.g. by Churchill 1942; Brown 1953; Cunningham 1956, 1961; Tucker 1964; Frank 1967). Yet we
believe many of the changes are more apparent than real and there is much to be gained from looking
backwards as well as forwards. Indeed, three themes have been of enduring concern and form the core
of this paper: how to conceptualise consumer loyalty, how to measure it, and how to manage it.
The purpose of this paper is, first, to appraise the concept and measurement of consumer
loyalty, and, second, to relate our understanding of loyalty and how it works to the emerging field of
loyalty marketing. The focus is on established repeat-purchase markets where there is direct
competition between branded products and services. These markets include most transport, retail, hotel,
media and personal services, as well as packaged goods, food and beverages, pharmaceuticals and
cosmetics. Many of these are routine and mundane purchases, but they are hugely important in terms of
the share of disposable consumer income for which they account, and will continue to be so.
In the next section we outline the conceptual and measurement aspects of consumer loyalty
and loyalty programs. There then follows a discussion of the implications of our framework. Alternative
perspectives are discussed, before concluding with an agenda for the future.
4
5
Consumer Loyalty Marketing: A Framework
Consumer loyalty is defined here as an on-going propensity to buy a brand. While there are
many different definitions of consumer loyalty we chose this generic definition because it: (a) highlights
the manifest nature of consumer loyalty in providing a revenue stream to the organisation, and which
therefore is of strategic importance to management; (b) is often used when studying low-involvement,
repeat-purchase consumer behavior; and (c) can apply to the study of an individual, a market or a
brand.
From this definition we describe individual consumer loyalty (Figure 1) and aggregate
consumer loyalty (Figure 2), and discuss a set of related measurement issues (Figure 3) for repeatpurchase markets. Each of these is discussed in turn.
Individual Consumer Loyalty
In Figure 1 we present a framework for looking at individual consumer loyalty in repeatpurchase markets. Current loyalty is seen as shaped by habits and reinforcement effects from past
behavior and attitudes, by personal preferences to features of the brand and any special initiatives to
maintain and build loyalty, and situational influences. Below we discuss the reasoning behind each
factor and the relationship between these components.
< Figure 1 >
6
Behavior
At the core of consumer loyalty is the manifest long-run propensity to buy a brand. That is, a
consumer reveals or states a propensity to buy a particular brand some proportion of the time. For
example, on the transatlantic air-route a typical frequent-flyer might choose BA 80% of the time.
The implication here is that loyalty is not something that can be attained exclusively for an
indefinite length of time. But rather, over time, consumers will tend to buy from a repertoire of brands
and have on-going propensities to buy each of the brands in these repertoires. Perhaps BA 80% of the
time and United 20%. Both proportions are seen as enduring over a certain pre-specified time horizon
(which would be longer than the length of the typical purchasing cycle), but on any specific purchase
occasion either BA or United could be chosen. Under these circumstances, not too much reliance
should be placed on trying to predict the next purchase in isolation, rather, it makes more sense to
attempt to predict both the repertoire of purchases and the shares of each brand in the category.
There is considerable empirical evidence to suggest that divided or "polygamous" loyalty is a
better description of actual consumer behavior for most people than either brand switching (a conscious
once-and-for-all change of allegiance to another brand) or promiscuity (the butterfly tendency to flit
from brand to brand without any fixed allegiance). Empirical studies of divided loyalty have been
undertaken for packaged goods, cosmetics, TV viewing, aviation fuel contracts, fast foods, grocery
retailing, and other repeat-purchase products and services in countries such as the US, Japan, UK,
Germany, Italy and Australia (Bass et al. 1976; Lattin and McAlister 1985; Ehrenberg 1988; Kahn et al.
1988; Uncles et al 1994, 1995; Blattberg et al. 1995; and reviewed in Ehrenberg and Uncles 1997).
Reasons for the generalized patterns of divided loyalty, or polygamy, include the fact that
consumers buy different brands for different usage occasions, or for variety, or because it was the only
one in stock, or it offered better value-for-money at the time of purchase because of a special deal. In
the case of BA and United the pattern of divided-loyalty might depend on whether the consumer is
7
travelling for business or pleasure, the precise route and schedule, special deals, offers, and perhaps
arrangements with third parties (such as car hire and hotel offers). We return to these effects later.
That on-going propensities differ across brands allows for the fact that a consumer may have a
favorite brand, a second brand, a third brand and so forth (BA might be the favorite brand and United
the second brand). Beyond these there may be no on-going propensity to buy other brands (in our
example, Delta, Air France and Virgin might not be chosen at all). The main exception - where
exclusive buying is observed - is among consumers who are light buyers of the product category and
therefore of any brand in the category. Among these consumers, apparent loyalty to BA or United may
merely reflect a very limited number of flying occasions.
The research also suggests that there is no difference between experience and credence goods
and services, both hold out the possibility of there being on-going propensities to buy from a repertoire
of brands. Nor, in the context of loyalty marketing, should it be assumed that a special program will
fundamentally alter these propensities. Indeed, all the evidence suggests that just as we have brand
repertoires so we also have program repertoires. The frequent-flyer with BA and United will take
advantage of both these airlines' loyalty programs. As for brand choice, only the light buyer will tend to
support just the one program (Dowling and Uncles 1997).
Attitudes
Many have argued that there must be "attitudinal commitment" for true consumer loyalty to
exist (Day 1969; Ajzen and Fishbein 1977; Jacoby and Chestnut 1978; Bagozzi 1981; Fishbein 1981;
Foxall and Goldsmith 1994; Mellens et al. 1996). This needs to take the form of a consistently favorable
set of stated beliefs toward the brands that are purchased. BA therefore is not only chosen 80% of the
time, but there is an on-going "attitudinal preference" to fly BA as well. Thus, flyers might believe it has
8
more comfortable seats, better executive lounges, tastier food, and friendlier airline staff, and
consistently hold these beliefs over a certain pre-specified time horizon.
However, having a consistent favorable set of stated beliefs about BA does not preclude
having a set of favorable beliefs about United. Nor does it preclude broadly similar evaluations of their
directly competing loyalty programs - there may in fact be very little to attitudinally differentiate one
brand or program from another. Typically, however, there exists a favorite brand within a personal
repertoire. This notion of a favorite brand might imply a real preference (perhaps based on the
perception that BA offers more than competing airlines) or simply signify that among salient brands it is
convenient to choose one brand more than another for much of the time (perhaps habitual use of BA
makes it easier to continue to deal with this airline or it offers financial advantages to the consumer)
(Cunningham 1961; Hammond 1997; Stern and Hammond 1998).
While empirical evidence for behavioral loyalty is strong and extensive, it is somewhat
equivocal for attitudinal loyalty. In large part this reflects the nature of buying in repeat-purchase
markets. Because many of the purchases are routine and mundane, attitudes about brands in the
consideration set may not strongly held. For the consumer who is repeatedly flying, shopping or buying,
it is not important to have a set of strongly held beliefs toward the brands that are purchased, as long as
they "do the job". To the extent that a consumer expresses a consistent set of attitudes, it is likely to be
based on frequent usage/purchase, frequent exposure to the brand, and familiarity with the
communications messages - a combination of learning, and in-use and emotional memory (Kuehn 1962;
Bird and Ehrenberg 1966, 1970a, 1970b; Lutz and Winn 1974; Lastovicka and Bonfield 1982; Hoyer
1984; Leong 1993; Rossiter and Percy 1997). BA, therefore, is known to have comfortable seats
because of past experience in flying with this airline, reinforced by the message content of BA's
advertising.
9
A further sign of the eqivocal nature of attitudes is that at an individual level, evidence shows
that attitudes can be quite unstable. Thus, a favorable belief about BA is not necessarily sustained over
time (even by those who continue to fly with this airline), raising questions about the nature of
attitudinal commitment. Moreover, those attitudes that are stable may simply describe the brand and the
communication messages, rather than offer an evaluation (Barwise and Ehrenberg 1985; Castleberry et
al. 1994; Dall'Olmo Riley et al. 1997).
The way we conceptualise these weak levels of attitudinal commitment and strong signs of
behavioral loyalty is to envisage a recursive relationship between the two, in that they interact over
successive purchase occasions (as suggested by the double-headed arrow in the left panel of Figure1).
The consumer will almost certainly have direct experience of the product category, and also of the
brand itself, so it is not as if choices are being made in ignorance. In these circumstances, it makes little
sense to assume attitudes cause behavior. If causality operates at all here it is more likely to be that
repeated behavior strengthens attitudes. Having flown with BA and United, or drunk Foster's and
Budweiser, the consumer is then prepared to state a liking for these brands. The role of a loyalty
program might then be to reinforce these actions and beliefs, in much the same way as advertising
(Ehrenberg 1974).
Brand Component
Brands in the repertoire offer functional value to the consumer. Transatlantic airlines enable
the traveller to fly from New York to London, in a timely, safe, comfortable and convenient way. An
ability to "do the job" is a necessary condition for the brand to be in a consumer's repertoire - assuming
that consumers are not forced through circumstance to buy something that fails to deliver or satisfy.
It is commonly argued that brands also offer psychological and social value (Aaker 1991;
Sheth et al. 1991; de Chernatony and McDonald 1998; Keller 1998). This is most easily appreciated for
10
unique luxury and lifestyle brands - DeBeers diamonds or YSL clothing. But in repeat-purchase
markets psychological and social value is inextricably bound up with attitudes and, as shown above,
these may reflect an accumulation of experiences (in-use and emotional memory) or simply playback the
message content of advertising (learning). These additional sources of value might, however, hold out
the possibility of providing a few points of difference between equally acceptable brands and this might
be useful to help the consumer justify a purchase (rather than admit to a reliance on convenience or
price even if these were the main reasons for the specific choice of brand).
In large measure, these sources of value will be known to consumers from past experiences.
Nevertheless, there is at least the possibility that thoughts and feelings about a brand can change prior
to a specific purchase occasion. Conventionally this would depend on the combined influence of the
following factors:
•
Cognition, based on new information about the brand.
•
Conative factors, including behavioral dispositions towards the brand based on perceived
switching costs, sunk costs, and consumer expectations.
•
Affective responses induced by mood states and emotions.
•
Social norms, in that favorable recommendations from peers, or from those who are seen as
credible (through advertising perhaps), might influence the consumer's assessment of the
functional, psychological and social value of the brand.
•
Personal norms that result in the individual having a pre-disposition to be brand-loyal or, by
contrast, deal-prone, and that will see the consumer choosing brands in a way that is compatible
with this proneness1.
The important point here is that for repeat-purchase markets these factors tend to be weak, in
contrast to the enduring impact of habit and the reinforcement effect of learnt behavior. For habitual
11
purchasing, problem-solving and extended decision-making sequences would have occurred some time
in the past - if at all (Olshavsky and Granbois 1979). In such circumstances, regarding thoughts and
feelings as formative antecedents of loyalty (in the sense used by Howard and Sheth 1969; Engel et al.
1993; Dick and Basu 1994) is not the most relevant approach to take.
Loyalty Program Component
Loyalty programs, like the brands with which they are associated, offer value to both the
organisation and the consumer. The value to the organisation is thought to accrue as much from how
they change the pricing dynamics among competitors as their ability to attract a larger pool of loyal
consumers (Nalebuff and Brandenburger 1996). Another way these programs might add value is by
providing an additional attribute to the product, which may be used in an attempt to provide at least
some points of difference among directly competing brands (Carpenter et al. 1994).
Many consumers appear to value loyalty programs because they offer an exotic reward (free
air travel) that is out of all proportion to the value of the item being purchased (such as food or
gasoline). Conceivably, these loyalty programs have the effect of inducing stronger thoughts and
feelings than is typically associated with the routine purchasing of brands. This might be seen in terms
of:
•
Cognition. More information is provided to the "valued consumer", through membership
newsletters, brochures and special events.
•
Conative factors. The consumer perceives a rise in switching costs, in that points will be forgone
and rewards unattained if the brand is not chosen.
1
In practice it may be hard to identify personal norms because people are inconsistent in their behavior and
attitudes across differing product categories.
12
•
Affective responses. The program is seen as creating a sense of belonging between the consumer
and the organisation, and this gives the person an emotional disposition to buy the particular
brand.
•
Social norms. There might also be social pressure to "join the club" or to achieve a certain status
within the program (such as BA Gold Class status), which motivates a purchase.
•
Personal norms. If the person has a pre-disposition to collect things or join clubs, it may be
attractive to join and use a membership-based loyalty program.
However, loyalty programs designed to support low-involvement, repeat-purchase products
and services, run the risk of generating more loyalty to the program than the brand. The crucial
question is whether the functional, psychological and social value of these programs is sufficient to
override the established patterns of loyalty described earlier. By contrast, the benefits of a program may
simply reward habitual purchasers (drawing in particular on the economic value of the program) or
reinforce existing behavior (a learning effect based on the different sources of value).
Only a limited number of before-and-after studies have been published, but from these it
appears that on-going propensities are not altered much by the introduction of loyalty programs (e.g.,
Sharp and Sharp 1997; East et al. 1998; Sharp 1998). This suggests that habit and reinforcement
dominate the effects outlined above. This is also in line with evidence on the long-run effects of
promotions; namely, that there are no after-effects in terms of repeat-purchase loyalty following price
promotions, and that most consumers who take advantage of a promotion have bought the product
before (Raju and Hastak 1980; Rothschild and Gaidis 1981; Szymanski and Busch 1987; Totten and
Block 1987; Ehrenberg et al. 1994; Blattberg et al. 1995).
Our conclusion is that over the long-run, product values and loyalty program values will
interact. This we conceive as another recursive relationship. For instance, given the product category
decision to fly from London to New York, the main driver of brand choice might be the habit of
13
travelling with United, reinforced by satisfaction with previous experiences of flying with United, and
then also reinforced by the prospect of accumulating frequent-flyer points. Under this conception,
programs that add real value to the core brand and are consistent with the brand's positioning are more
likely to interact positively than ones that offer irrelevant value.
Situational Influences
That the consumer has a brand repertoire also means the decision made on a specific purchase
occasion may be subject to situational influences, without any serious risk (in that the consumer knows
from prior experience the alternatives in the repertoire are all satisfactory). These situational influences
are of three types:
•
Situational cognitive influences, such as competitive information at the point of purchase which
induces the consumer to re-think the purchase.
•
Situational conative influences where brands in the repertoire change their basic offer (such as BA
offers an up-grade, United flies at a more convenient time on this occasion, or Avis offers a
special deal in conjunction with BA, not United).
•
Situational affective influences include the consumer's general mood (having to contemplate air
travel) and emotional responses at the point of purchase.
These influences are specific to the situation and therefore are as likely to be responses to the
buying process as much as choice of brand or program. However, if similar choice situations repeatedly
occur then these situational influences may become engrained (a consumer might come to expect to buy
on offer, even if the person was not especially deal-prone, or a particular type of buying situation - like
air travel - might become associated with specific emotions, this becoming a form of "emotional
memory").
14
A loyalty program that plays on situational influences may be perceived as offering real value
to the consumer (hence, the possible attraction of frequent-flyer programs that reduce queuing times
and minimise check-in procedures, or retail schemes that have fast lanes for program members).
It is apparent that these situational influences are crucial to any understanding of consumer
loyalty in competitive markets. Hitherto they have been largely under-played, giving rise to views of
loyalty which fail to be realised in the marketplace. It is not uncommon to hear comments such as: "we
would have 100% loyalty if only it wasn't for out-of-stock conditions, distribution bottlenecks, pointof-sale promotions by our competitors, and so forth". Yet these conditions are facts of marketing. To
the extent that larger brands have better distribution and are more widely available, this provides some
explanation for the finding that larger brands have somewhat greater levels of loyalty (discussed more
fully in the next section). From a practical viewpoint, there is little to be gained from an abstract notion
of "pure loyalty", untroubled by competitive and situational influences, when these are always present in
the marketplace.
Figure 1 shows the collective, time-ordered influence of the purchase history, product and
loyalty program components, and the situational influences that govern current behavior and current
attitudes. In the context of repeat-purchase markets, this process is merely a snap-shot in a sequence of
purchases extending into the future: t+1, t+2, etc. Our argument is that it makes good sense to think of
loyalty as contingent on all these factors.
Aggregate Consumer Loyalty
So far, the focus has been on individual consumer loyalty, however any practical concept of
consumer loyalty must have a means of aggregating the process in Figure 1. This is true unless one
assumes that all consumers are homogenous (say within a very tightly-defined segment), or upholds an
extreme form of one-to-one marketing where it is possible to justify building complete models of the
15
loyalty of individuals and targetting customised marketing programs at these specific consumers. In
practice, consumers seldom show high levels of homogeneity (therefore many segments are fuzzy,
overlapping, or hard to discern empirically, Jain et al 1990; Hammond et al. 1995), and often the oneto-one targetting of large numbers of consumers is uneconomic (in this respect it is quite different from
having key accounts in business-to-business markets). Thus, for the types of markets studied in this
paper, we need a perspective on consumer loyalty that explicitly accommodates heterogeneity across
consumers.
At an aggregate level repeat-purchase markets typically have a well-defined structure, as
shown in Figure 2. Brands here are defined by their degree of loyalty and market shares. The
fundamental and overriding fact here is that most brands in an established, competitive market reflect
the "Double Jeopardy Law" that smaller market-share brands have fewer buyers than bigger brands and
these buyers buy them less often (and also typically say that they like them less) (McPhee 1963;
Ehrenberg et al. 1990). An inference here is that one way to instil loyalty is to become the market leader
and there is some empirical evidence for this interpretation (Ehrenberg 1988; Reibstein and Farris 1995;
Rust and Zahorik 1993; Szymanski et al. 1993; East and Hammond 1996; Dekimpe et al 1997).
< Figure 2 >
In theory and in practice there can be exceptions to these Double Jeopardy brands:
•
Big brands which exhibit signs of "super-loyalty" (that is, they have a higher degree of repeatpurchase or behavioral loyalty than we would expect after allowing for the DJ effect) (Fader and
Schmittlein 1993).
•
Niche brands which have a higher proportion of behaviorally loyal buyers for their level of market
share than predicted by the DJ line.
•
Change-of-pace brands which have a higher market share but a lower repeat-purchase rate than
the DJ line predictions (Kahn et al. 1988).
16
Being exceptions, these types are not observed often or consistently across markets.
Therefore their predictive use is limited. Nor do they appear to correlate systematically with
independent variables like price, advertising support, etc. (Bhattacharya 1997). Hence, managers will
typically find it very difficult to engineer these types of exceptional brand2.
Yet, much of the literature on loyalty programs suggests that their objective is to create a niche brand
or a super-loyal brand without the necessary increase in market share. Specifically, these programs try
to gain a higher proportion of behaviorally and attitudinally loyal buyers for a certain level of market
share (Baldinger and Rubinson 1997). In Figure 2 this goal is shown as A'. The existence of the Double
Jeopardy Law suggests that more likely outcomes lie between A (no change or a defensive "hold") and
A'' (a slight loyalty gain arising from a significant market share gain). Thus, if a loyalty program helps a
brand become more of a market leader, then loyalty might increase as a natural outcome of this
leadership. More than likely, a loyalty program will not achieve this by itself, but it might be a useful
contributory factor in the context of a set of initiatives (for an airline this might mean a wider range of
routes, more frequent flying times, greater comfort on board the aircraft, faster check-in procedures and
more reliable baggage handling).
Published studies are few in number, but a few examples illustrate the point: American
Airlines (first to launch a frequent-flyer program but its continuing success depends on market
dominance, Kearney 1990); Tesco (first to launch a sophisticated national grocery program in the UK,
but has gained share because of new store opening and because of what competitor Sainsbury failed to
do (East 1997; East and Hogg 1997; Mason 1998); AirMiles (which has had mixed fortunes, but what
appeal it has comes from the range of co-branding arrangements); and FlyBuys (where Australian
competitors not in the program have consistently performed as well, Sharp and Sharp 1997; Dowling
2
Given that investment analysts try to beat the random walk of Wall Street, perhaps it is not surprising to see
marketing managers trying to beat the norms of competitive markets.
17
and Uncles 1997). Successes appear to be among those who have increased market share and who can
offer broadly-based value, which is what we would expect for competitive markets.
Measurement Issues
The framework shown in Figure 1 is not an operational model - it goes beyond what can be
captured in a single tractable algorithm. Nevertheless, we can envisage a set of measures to capture the
main effects. This we show in Figure 3. In principle there are several competing operational models to
bring together these measures in repeat-purchase markets - most of which start with stochastic
formulations (e.g., Massy et al. 1970; Kahn et al. 1986; Colombo and Morrison 1989; Eliashberg and
Lilien 1993). Within this framework, consumer loyalty to a brand is thought of as if it were stochastic
(not that it is stochastic). Thus, for instance, the specific incidence of purchasing appears as if at
random, even though the consumer will have an on-going propensity to buy a brand. Furthermore,
individual attitudes themselves appear to be more unstable over time than is often assumed, even while
appearing reasonably stable in the aggregate (Dall'Olmo et al. 1997). An implication is that while
attitudes are not usually explicitly modeled within these stochastic models, there is a degree of
consistency between the behavioral and attitudinal representations. Consumers will have reasons for
their revealed behavior and stated attitudes but this may appear to be a random outcome on any
particular purchase occasion when looked at in the aggregate.
< Figure 3 >
Among the range of available stochastic models, the NBD-Dirichlet formulation appeals most
because of the considerable empirical support across a very wide range of conditions - no other model
has been as widely empirically validated in marketing (Bass et al. 1976; Goodhardt et al. 1984;
Schmittlein et al. 1985; Ehrenberg 1988) and it realistically captures the nature of consumer behavior in
competitive markets (including polygamous loyalty and the Double Jeopardy effect).
18
This model is also very parsimonious and routinely tractable, although complexity can be
added to explicitly model heterogeneity, state dependence and non-stationarity. For example,
independent variables (advertising spend, existence of a loyalty program, etc.) can be incorporated into
the estimation of the parameters. Also, the data may be split according to demographic and attitudinal
characteristics, or these effects can be incorporated directly into the estimation process (in Figure 3 we
distinguish these as "measurable variables") (Jain et al. 1990; Jones and Zufryden 1980; Heckman 1981;
Uncles 1988; Fader 1993; Russell and Kamakura 1994; Fader and Hardie 1996). More simply, these
influences are subsumed within the main effects as "unmeasurable variables". Specification of the
model, empirical evidence and applications are fully described in Ehrenberg and Uncles (1997).
There are several very significant specification factors underlying the depiction in Figure 3, all
of which are crucial to the NBD-Dirichlet formulation and would be equally important for any
competing model. Namely:
•
The object to which loyalty is being exhibited; for example, brand or vendor, SKU or pack-size,
place of purchase or time of day, or, indeed, the loyalty program.
•
The specified time horizon (t-2, t-1, t, t+1, t+2, etc. in our figure). As the observed time period
increases (relative to the length of the average purchase cycle), so there are increased
opportunities to buy competing brands and so also the scope for attitude change is greater.
•
The level of aggregation of consumers (i = 1, …, N). We have noted already that many
conceptions of loyalty essentially describe individuals, but for marketing management we need to
aggregate across individuals. This can be achieved by assuming a parametric form for
heterogeneity across the whole market or for segments of consumers (to the extent that
managerially useful segments can be identified).
•
Competitive intensity, as defined by the number of competing brands, the extent to which they
have differentiated attributes, and their impact on intentions at the point-of-purchase. Farley
19
(1964) showed market structure, such as number of brands available, is associated with measures
of loyalty.
These specification factors highlight why snapshot studies of consumer loyalty can be so
misleading and why a model, such as a first-order Markov process, is usually inappropriate (limited as it
is to t and t-1). To characterise loyalty - as against current purchase choice (which in fact is what many
models attempt to do) - behavior and attitudes need to be appraised over a certain time horizon. This
may be measured in terms of brand share for each individual consumer (United is flown on six out of
ten flights), discrete time periods (for BA the repeat-buying level from one six month period to the next
might be 60%), or survival rates (BA has been selected on six consecutive flights) (Massy et al. 1970;
Ehrenberg 1988; Helsen and Schmittlein 1993; Bharadwaj and Bhattacharya 1998; Voss 1998).
Typically we would want to use some form of longitudinal data to determine on-going loyalty and also
provide a basis for consumer lifetime value calculations.
In general, it makes little sense to speak of loyalty without reference to these specification
factors. Yet, in practice many studies fail to explicitly state them, and thus we see widely varying
conclusions about loyalty, and programs to instil loyalty (e.g., Christopher et al. 1991; Fournier 1998).
Studies that mainly sample outliers from a heterogeneous population of consumers might imply that
exceptional levels of loyalty can be realised, but how representative are these studies, do they take full
account of the object to which loyalty is being exhibited, or competitive intensity? Those responsible for
the future management of BA or United, for instance, have to consider the millions of consumers who
fly with them, not only exceptional cases.
20
Implications of the Framework
Conceptual Implications
The most important conceptual implications that flow from Figure 1 for loyalty marketing are
that:
•
A dynamic equilibrium tends to exist in repeat-purchase markets, such that the introduction of
something like a loyalty program may give rise to perturbations, but it is unlikely to disturb the
whole system. With so many of these being me-too or lookalike programs it is not hard to see
why this is the case.
•
Divided-loyalty, or polygamy, is the dominant characteristic of repeat-purchase markets. This
occurs for many different behavioral and attitudinal reasons, including the dictates of convenience;
the desire to overcome boredom by variety seeking; to meet the needs of different consumption
occasions; point-of-purchase situation effects, including out-of-stock conditions and personal
moods and emotions; a reluctance to feel "locked-in"; and greater cynicism than in the past with
regard to organisations' claims for their brands. Loyalty programs that are not designed to
accommodate these underlying reasons for divided loyalty are unlikely to meet with even limited
success. Our observations suggest that, to date, most initiatives do not acknowledge these issues,
let alone address them.
•
Consumers differ considerably in both their product-category and brand purchase frequencies.
Any brand generally has most of its buyers purchasing it infrequently and a much smaller number
buying it often (that is, the "80:20 rule" holds, as it does in most of the models, such as the NBDDirichlet). Of those who buy often and who therefore are more profitable, on average they buy a
larger number of different brands. In aiming at these heavy buyers competing loyalty programs
tend to target the same people - in much the same way as brand promotions do.
21
•
Given the variability and instability in attitudes for low-involvement, frequently purchased
products and services, it is dangerous to assume that the behavioral patterns can be disturbed by
attempts to change attitudinal loyalty. On the whole, attitudinal commitment appears to be weak,
and more strongly shaped by habit and reinforcement effects than cognitive/affective factors. This
implies that loyalty programs that build on habitual patterns or reinforce existing behavior will be
more successful than ones that attempt to bring about something largely unrelated to existing
purchasing patterns.
Measurement Implications
There are also a number of measurement implications that flow from Figures 2 and 3:
•
Measurement systems are needed that capture polygamous behavior. Many in-house databases do
not do this - they merely record information about own-company brands. This is where traditional
consumer panels have an advantage over in-house databases, in that panels enable a picture of the
whole purchase repertoire to be built up.
•
Any operational definition of loyalty must be explicit about the time-based nature of purchase
behavior. An advantage of traditional behavioral definitions, such as repeat-buying, share of
category requirements, brand persistence, etc. is that they are explicit about the treatment of time.
So too are lifetime value calculations - the discounted value of a consumer's purchases over some
stipulated time horizon. It is then possible to explore the impact on consumer buying or value of
differences in the assumed time horizon. Again, consumer panel data are relevant, so too are
attitude tracking surveys (Totten and Block 1987; Brown 1994; Russell and Kamakura 1994).
One-off attitude studies, which are by far the most common source of attitudinal information, are
very much a second-best approach.
22
•
Measures are needed that allow for consumer heterogeneity. This means essentially quantitative
measures, although qualitative research which describes both "average" and exceptional
consumers can also be considered under this framework - for instance, qualitative research on
mood as a partial explanation for commonly observed patterns of divided loyalty (Gordon 1994;
Fournier and Yao 1997).
•
Finally, we should aim for measures of loyalty marketing effectiveness - just as we have
effectiveness measures of branding, advertising, promotion, etc. Typically these should take the
form of before-and-after studies; although it might well be hard to detect measurable differences if
the program is essentially defensive or if it mainly reinforces existing behavior. Models such as the
NBD-Dirichlet can serve as a benchmark against which to assess any behavioral departures or
exceptions, although this model takes no account of the costs of creating and maintaining a
program which will be an important consideration in any comprehensive assessment of loyalty
marketing effectiveness.
Managerial Implications
Among the implications for management, the following are particularly significant:
•
The typical patterns of single brand and polygamous loyalty that have been established for brand
buying in repeat-purchase markets are likely to persist in a world of loyalty marketing. This is
evident already among non-frequent-flyers who belong to one program and frequent-flyers who
participate in several programs in order to retain schedule flexibility. Knowing polygamous loyalty
is so widespread helps us evaluate programs. For instance, those who exhibit divided loyalty
(because they are more frequent buyers) are likely to be more profitable and therefore could be
offered greater value in the hope of nudging them into buying somewhat more of the brand.
Single-brand loyal consumers may be being over rewarded for their patronage.
23
•
For management to build loyalty in a competitive market the brand and program must work
together to enhance the overall value proposition of the brand. Only then might established habits
be reinforced and situational influences mitigated. For instance, if convenience is the major
determinant of retail store choice, then a program which makes a retail chain more convenient
(such as on-line purchasing) is more likely to succeed than one that offers rewards unrelated to
the important choice attributes (such as frequent-flyer points). Moreover, the way to achieve this
may be with loyalty marketing, but not necessarily a formal program. The experiences of Lexus,
Readers' Digest, American Express, Singapore Airlines and State Farm show such a range of
initiatives which are not simply based on programs, rewards and points.
•
While Figures 1 and 3 contain a number of contingent and specification factors, it is still
appropriate to talk about "loyalty" as a general concept. Indeed, it is potentially misleading to
define some forms of loyalty as "latent" or "spurious" and assume that these are somehow less
valid or useful to management. Much money has been made from inertia-based loyalty and much
money has been lost by assuming satisfied customers will remain loyal. The design of loyalty
marketing initiatives needs to take more account of these factors, rather than rely on the massive
changes in attitudes expected in idealistic schema (such as the "loyalty ladder" that sees
"purchasers" being converted into "advocates").
These implications suggest that the structure of the market imposes constrains on managerial
actions. While these constraints might be unwelcome, most managers will recognise them as facts of
life. Specifically:
•
Situational factors invariably come into play. Stock-outs, accessibility at the time of purchase,
competitive offers, etc. are part of normal marketing, not exceptions.
•
Consumers are often constrained in when to buy, where, how, who they are with, etc. and by
variability in distribution coverage, etc. How these factors vary over time can have an important
24
impact on the pattern of polygamous buying. Also people have to do things they may not want to
do - airline travel, grocery shopping, banking, taking out insurance, etc. As a compensation
mechanism they often minimise the amount of search and thought that is devoted to such
"grudging" purchases.
•
Organisations are themselves constrained to offer at least parity with existing market offerings
(often giving rise to me-too brands, lookalike packaging, copy-cat advertising, parity pricing, and
imitative reward schemes).
•
Product attributes are similar, so there may be no valid grounds on which a buyer can make an
informed cognitive/affective choice.
Many discussions of loyalty ignore these factors - they focus on just the brand of interest,
imply the consumer has unconstrained choice, and yet chooses to adhere faithfully to the one brand this is a wishful view of markets and marketing. Our approach, which includes competitive effects and
situational influences, provides a more realistic focus for managers and articulates much of what tends
to go on in brand strategy meetings.
25
Consumer Loyalty Marketing: Alternative Perspectives
While the theoretical and empirical basis of the framework presented here is strong, many
practitioners and academics have in mind alternative perspectives of consumer loyalty. We discuss these
briefly.
Behaviorist Frameworks
A strongly behavioral viewpoint contends that current loyalty is the result of circumstances
that applied to previous loyalty, and that any such loyalty will be positively reinforced if the brand
continues to "do the job". The best guide to future loyalty, therefore, is past loyalty. Accordingly,
attitudes are at best seen as descriptive of the brand, informed by past behavior, and learnt responses to
familiar circumstances (Skinner 1953; Tucker 1964; Rothschild and Gaidis 1981; Foxall 1987).
The strong behavioral view is that a loyalty program will merely change the circumstances, to
which consumers will respond in a non-evaluative manner. Thus, frequent-flyers might describe pointsschemes as a form of reward associated with stated airlines, but would not be thinking long and hard
about these points prior to choosing an airline or prior to flying. In its extreme form this viewpoint does
not do justice to either the importance of habit nor to any evaluative processes which consumers might
go through, particularly in response to the introduction of something new (such as a loyalty program).
A weaker notion of behavioralism, however, underlies much of our conception of loyalty in
this paper. In general, past behavior is a very robust predictive guide of future behavior and, to the
extent that loyalty marketing initiatives have an impact, they do so largely through the reinforcement of
existing behavior. This is consistent with the view that for loyalty marketing to succeed it must raise
market share by building on existing patterns of behavior and associated attitudes, rather than imagine it
can bring about totally new patterns.
26
Cognitive/Affective Frameworks
In contrast to the behaviorist perspective, many of those involved in the development and
analysis of loyalty marketing have in mind a strong cognitive/affective concept of consumer loyalty.
Namely, that through the introduction of a program or some other special initiative, consumers will reevaluate their attitudes and change their subsequent behavior. In so doing they will come to display
faithful adherence to a brand and may even become advocates.
This viewpoint echoes the cognitive/affective framework of consumer behavior that is
prevalent throughout much of the marketing discipline (Howard and Sheth 1969; Engel et al. 1993;
Dick and Basu 1994). From a managerial viewpoint, it holds the promise of very desirable outcomes - a
(large) pool of faithful adherents who will be profitable. However, as the foregoing discussion has
shown, the effect of beliefs and attitudes may be quite weak in practice. There are a number of reasons
for this.
First, the empirical evidence shows that most people in repeat-purchase markets do not exhibit
strong adherence to brands, or associated loyalty programs. Those that do tend to be few in number
and are not necessarily heavy or profitable buyers. In the specific case of airline choice, the appearance
of faithful adherence may come from corporate policies which favor one carrier over another - but the
frequent flyer benefits typically accrue to individuals who will display varying levels of heavy flying and
profitability.
Second, focusing on attitudinal commitment without allowing for habit and circumstance will
be misleading for most consumers. Attitudinal commitment to icon brands such as Porsche is not
uncommon, however few can afford to buy such brands. There also may be attitudinal commitment to
heritage brands, such as Spam, Jell-O, Marmite, and Aeroplane Jelly, but quite often the functional need
to buy these brands limits their consumption and so, once again, attitudes overstate purchase behavior.
27
All-in-all, an exclusively cognitive/affective perspective on consumer loyalty may merely result
in many consumers being classified as "disloyal switchers". This is not a very helpful classification when
so many of these consumers clearly have on-going propensities to buy certain brands and not others.
Satisfaction-Based Frameworks
Satisfaction is an important post-purchase evaluation which has the potential to reinforce
existing behavior and attitudes, and therefore impact on continuing purchasing as well as word-ofmouth recommendation (Oliver 1997). As a predictor of (especially single-brand) loyalty, however,
satisfaction is an imperfect measure.
First, loyalty is so contingent on prevailing habits and situational influences that very often
there is only a weak connection between retention and consumer satisfaction (Jones and Sasser 1995;
Hennig-Thuau and Klee 1997). An exception would be following an extremely dissatisfying experience
that prompts a re-evaluation of previous choices (Dube and Morgan 1998). Second, satisfaction, as it is
currently measured, is not discriminating among groups of acceptable brands (Kayande 1997;
Edwardson 1998).
Given these problems in relating satisfaction to loyalty more thought needs to be given as to
how it is measured, and its role in the recursive relationship between behavior and attitude (Oliver
1998). Our view is that satisfaction impacts loyalty through habit and purchase reinforcement.
Relationship-Based Frameworks
Relationship-based perspectives of consumer loyalty have their origins in business-to-business
markets and focus on relationships between people, and the networks of communication, activity, trust
and friendship that arise among these people. These conditions are very different from the markets we
are addressing in this paper, although we comment on this work because many brand managers have
28
tried to replicate relationship-based loyalty in repeat-purchase consumer markets (Peppers and Rogers
1993, 1997; Morgan and Hunt 1994; Fournier 1998).
Most consumers will not want a relationship with an ATM, brand of toothpaste, fast food
outlet or airline. Some might, however, claim to have a relationship with a bank teller, supermarket
check-out operator or travel agent (although even here many would not make this claim). The issue
here has more to do with the quality and authenticity of a relationship between people, than repeatpatronage. In principle, loyalty marketing could shift the focus towards building relationships between
people, but in practice much of the effort does not do this. In fact, there is considerable evidence of the
"industrialisation" of loyalty-building initiatives, involving the appearance of personal service and the
rhetoric of relationship marketing. This is particularly evident with rewards-based programs, but it even
features in membership-based initiatives which ostensibly play on more deeply-seated associations
between the consumer and the brand, such as affinity cards (Worthington and Horne 1996). Some have
recognised this trend and pointed out the need to move beyond notions of loyalty to the creation of
genuine and authentic relationships (Barnes 1994; Iacobucci and Ostrom 1996; Fournier et al. 1998).
However, in trying to move beyond loyalty marketers will need to recognise that consumers
are polygamous. For instance, consumers know that the sales assistants in one store are every bit as
friendly as in another because they routinely patronise both stores (albeit with differing on-going
propensities). Moreover, the reasons for polygamy are such that it will not cease to occur even if there
are genuine relationships.
Analogies with personal relationships, political affiliation, and religious adherence are
unrealistic - most ordinary brands will not generate enough motivation or attachment for buyers to fight
for or defend the brand (notwithstanding a few egregious examples, like Nike Air Jordan shoes, which
have been associated with street fighting, Smothers 1993). Perhaps the only relevance of religious and
29
political analogies is to draw attention to the prevalence of agnosticism and floating-voters in
contemporary society, which parallels some aspects of consumer behavior.
The exception to this scenario may be some lifestyle brands. Some consumers talk about
having a relationship with Harley-Davidson, but this is a most unusual brand with no direct competitors
and a community of riders who participate in group activities - it is more akin to a product category
decision than brand choice (Schouten and Alexander 1995).
Non-Traditional Repeat-Purchase Markets
Historically many sectors have been viewed as non-repeat purchase markets. Consumers
would join a public utility just the once, or they would buy a life insurance policy with the expectation
that it would be for life. In these cases the question of how to instil loyalty was largely irrelevant - the
switching costs and penalties were such that most consumers stayed with the organisation first chosen.
More likely, the focus was on issues such as managing post-purchase satisfaction, creating more
frequent use, word-of-mouth recommendations and cross-selling.
More recently, the privatisation of utilities and the deregulation of monopolies has offered
consumers the opportunity to either switch or hold a repertoire of brands. As a result many
conventional non-repeat purchase markets, such as personal banking, insurance, telephony, and energy
supply are beginning to look more like repeat-purchase markets with consumers exhibiting patterns of
divided loyalty (for example, different telephone companies for domestic and international calls).
Typically loyalty programs in these marekts are not radically innovative. They simply add one
or two attributes to the product offering. If well-considered, these attributes might be perceived as
offering some value to the consumer and might help to raise market share (in combination with other
marketing activities such as improved distribution or more incisive advertising). As competition
30
increases and repeat-purchasing occurs, however, these attributes are easily copied or improved upon
(while most programs can be branded, there can be little protection for the ideas behind most programs.
In these circumstances, loyalty marketing can be seen as an attempt to re-impose switching
costs, and those that offer real value may have some measure of success, but it is essentially a response
to the advent of competition and one that many consumers will not want to accept. They do not want
to feel "locked-in" having now been given a measure of consumer sovereignty.
Really New Markets
In newly emerging markets, such as the introduction of a new high-tech product or patent
medicine, the conception described in Figures 1 and 2 might not be appropriate. At the very least,
buying habits would not have been established, consumers will lack experience of the product, and the
market may be pre-competitive. Therefore, there is an opportunity to establish a pattern of loyalty
based on pioneer and first-mover advantages before the market settles down into a dynamic equilibrium
(Robinson and Fornell 1985; Szymanski et al 1995). Under these special conditions many aspects of
marketing may be quite different from those which exist once a competitive market has settled down
(Glazer and Weiss 1993).
That said, people are likely to have some behavioral basis for their choices. They might have
bought or used something similar before, and would have experience of a similar buying process
(Wilkie and Dickson 1985). Even here, then, a strongly cognitive/affective conception of loyalty may be
misleading (in fact the absence of experience and brand familiarity makes it hard to see how wellfounded attitudes could be formed prior to purchase and the fact that so many products fail after initial
trial is some indication of the problem).
Conceivably, a loyalty program (perhaps in the form of a user-support group, or a contract
pricing structure, as seen among mobile phone companies and Internet access providers) might create a
31
sense of loyalty, with pioneer and first-mover advantages to the organisation that drives the process
forward. However, the evidence is of something far from sustainable.
32
Future Agenda for Consumer Loyalty Marketing
Based on the framework in Figure 1, and the review of alternative perspectives of consumer
loyalty, there are several areas where further work would contribute to our understanding of loyalty
marketing:
•
We would expect to see more research on the relationship between attitudinal loyalty and patterns
of purchase behavior in the context of the widespread existence of loyalty marketing initiatives.
This research is likely to proceed by further exploiting consumer panel data and tracking studies,
and establishing market norms. This work would need to incorporate point-of-sale influences as
well as address issues of market concentration and the intensity of competition.
•
More work is also needed to explore habitual decision-making and how this might be influenced
(if at all) by special loyalty-building initiatives - habits are not the absence of thought, but nor are
they extended decision-making.
•
In keeping with the emphasis on brand value, we see a role for further research on how loyalty
programs add value to a brand for different types of consumer (such as heavy versus light users).
Some loyalty programs build this thinking into the structure of their programs, but adopt a more
general approach.
•
Discussion of what are the most appropriate measures is likely to continue. We have emphasised
revealed behavior and stated intentions as being the most appropriate when considering loyalty to
brands, stores, or programs. These are most reliable and directly related to revenue. For the study
of professional service and business-to-business markets, relationships between people also may
be relevant - commitment, trust, friendship, and intimacy. However, only in exceptional
circumstances will these apply to prosaic product brands in repeat-purchase markets or more
"industrialized" forms of service provision.
33
•
Most attempts to instil exceptional loyalty are susceptible to nullifying actions by competitors - as
they too create databases, establish rewards programs, initiate dialogue with their (shared)
consumers. More evidence needs to be gathered on these competitive reactions. At present the
best advice from Figure 2 about how to achieve greater loyalty seems to be a very traditional one:
to increase market penetration and increase share of category requirements. We caution that the
use of short-term promotions does not seem to have been a cost-effective strategy to achieve
either share or repeat-purchase loyalty.
Ultimately, improved understanding and measurement of consumer loyalty might also present
a challenge for management. Any behavioral or attitudinal justification for the development and
maintenance of a loyalty program rests on having clear and precise measures of effectiveness, yet this
probably means managers will become more accountable for specific (measurable) outcomes. There are,
however, signs of reluctance to assess the cost-effectiveness of loyalty marketing, rewards programs,
promotions, databases, etc. (Piercy 1996).
There is now considerable momemtum behind loyalty marketing, but if this is to become the
marketing mantra for the new millennium it needs to be informed by what is already know about
behavioral and attitudinal loyalty. This has not happened thus far, and therefore many of the claims for
loyalty marketing are inflated, unrealistic and in danger of discrediting many of the more recent
marketing initiatives such as database, relationship and direct marketing.
34
References
Aaker, David A. (1991), Managing Brand Equity. New York: The Free Press.
Ajzen and Fishbein (1977), "Attitude-Behavior Relations: A Theoretical Analysis and Review of
Empirical Research," Psychological Bulletin, 84, 888-918.
Bagozzi, Richard P. (1981), "Attitudes, Intentions and Behavior: A Test of Some Key Hypotheses,"
Journal of Personality and Social Psychology, 41, 607-627.
Baldinger, Allan L. and Rubinson, Joel (1997), "The Jeopardy in Double Jeopardy," Journal of
Advertising Research, November-December, 37-49.
Barnes, James G. (1994), "Close to the Customer: But is it Really a Relationship?," Journal of
Marketing Management, 10 (7), 561-570.
Barwise, Patrick and A.S.C. Ehrenberg (1985), "Consumer Beliefs and Brand Usage," Journal of the
Market Research Society, 27, 81-93.
Barwise, Patrick and Kathy Hammond (1998). Predictions: Media. London: Phoenix Paperback.
Bass, Frank, Abel Jeuland and Gordon P. Wright (1976), "Equilibrium Stochastic Choice and Market
Penetration Theories", Management Science, 22, 1051-1063.
Belch, George (1981), "An Examination of Comparative and Non-Comparative Television
Commercials: The Effects of Claim Variation and Repetition on Cognitive Response and Message
Acceptance," Journal of Marketing Research, 18 (August), 333-349.
Bharadwaj, Sundar and C.B. Bhattacharya (1998), "Persistent Superior Performance: The Role of
Brand and Category Level Correlates", Emory Working Paper
Bhattacharya, C.B. (1997), "Is Your Brand's Loyalty Too Much, Too Little, or Just Right?: Explaining
Deviations in Loyalty from the Dirichlet Norm", International Journal of Research in Marketing, 14
(5), 421-435.
Bird, Michael and A.S.C. Ehrenberg (1966), "Intentions-to-Buy and Claimed Brand Usage,"
Operational Research Quarterly, 17, 27-46 and 18, 65-66.
Bird, Michael and A.S.C. Ehrenberg (1970a), "Consumer Attitudes and Brand Usage," Journal of the
Market Research Society, 12 (4), 233-247.
Bird, Michael and A.S.C. Ehrenberg (1970b), "Brand Image and Brand Usage", Journal of Marketing
Research, 7 (August), 307-314.
Blattberg, Robert C. and John Deighton (1996), "Manage Marketing by the Customer Equity",
Harvard Business Review, July-August, 136-144.
35
Blattberg, Robert C., Richard Briesch and Edward J.Fox (1995), "How Promotions Work", Marketing
Science, 14 (3), Part 2 of 2, G122-G132.
Bone, P.F. (1995), "Word-of-Mouth Effects on Short-Term and Long-Term Product Judgements,"
Journal of Business Research, 32, 213-223.
Brown, George H. (1953), "Brand Loyalty - Fact or Fiction," Advertising Age, Vol. 24. Reprinted in
A.S.C. Ehrenberg and F.G. Pratt, Consumer Behaviour. Middlesex: UK: Penguin, 1971, 28-35.
Brown, Gordon (1994), Plumbing and Shirts. Warwick, UK: Millward Brown International.
Carpenter, Gregory S. and Don Lehmann (1985), "A Model of Marketing Mix, Brand Switching and
Competition", Journal of Marketing Research, 22 (3), 318-329.
Carpenter, Gregory S., Rashi Glazer and Kent Nakamoto (1994), "Meaningful Brands From
Meaningless Differentiation: The Dependence on Irrelevant Attributes" Journal of Marketing Research,
31 (3), 339-350.
Castleberry, Stephen B., Neil Barnard, Patrick Barwise, A.S.C. Ehrenberg and Francesca Dall'Olmo
Riley (1994), "Individual Attitude Variations Over Time," Journal of Marketing Management, 10, 153162.
Christopher, Martin, Adrian Payne and David Ballantyne (1991), Relationship Marketing, Oxford:
Butterworth-Heinemann.
Churchill, H. (1942), "How To Measure Brand Loyalty," Advertising and Selling, 35, 24.
Colombo, Richard A. and Donald G. Morrison (1989), "A Brand Switching Model with Implications
for Marketing Strategies," Marketing Science, 8, 1, 89-99.
Copeland, Melvin T. (1923), "Relation of consumers' buying habits to marketing methods," Harvard
Business Review, 1 (3) April, 282-289.
Cunningham, R.M. (1956), "Brand Loyalty - what, where, how much?", Harvard Business Review, 34
January-February, 116-128.
Cunningham, R.M. (1961), "Customer Loyalty to Store and Brand", Harvard Business Review,
November-December, 127-137.
Dall'Olmo Riley, Francesca, A.S.C. Ehrenberg, Stephen Castleberry, Patrick Barwise and Neil R.
Barnard (1997), "The Variability of Attitudinal Repeat-Rates", International Journal of Research in
Marketing, 14 (5), 437-450.
Day, George S. (1969), "A Two-Dimensional Concept of Brand Loyalty," Journal of Advertising
Research, 9, 29-35.
de Chernatony, Leslie and Malcolm McDonald (1998), Creating Powerful Brands. Oxford:
Butterworth-Heinemann.
36
Deighton, John (1994), "The Future of the Marketing Communications Industry: Implications of
Integration and Interactivity". In Marketing Communications Structures Today and Tomorrow:
Integration, Allocation and Interactive Techniques. MSI Report No. 94-109, 40-43. Cambridge, MA:
Marketing Science Institute.
Dekimpe, Marnik G., Martin Mellens, Jan-Benedict Steenkamp and Piet Vanden Abeele (1997),
"Decline and Variability in Brand Loyalty," International Journal of Research in Marketing, 14 (5),
405-420.
Dick, Alan S. and Kunal Basu (1994), "Customer Loyalty: Toward an Integrated Conceptual
Framework," Journal of the Academy of Marketing Science, 22 (2), 99-113.
Dowling, Grahame R. and Mark Uncles (1997), "Do Customer Loyalty Programs Really Work?,"
Sloan Management Review, 38 (4), 71-82.
Dube, Laurette and Michael S. Morgan (1998), "Capturing the dynamics of in-process consumption
emotions and satisfaction in extended service transactions," International Journal of Research in
Marketing 15 ( ), 309-320.
East, Robert. (1997), Consumer Behaviour: Advances and Applications in Marketing. London:
Prentice Hall.
East, Robert and Kathy Hammond (1996), "The Erosion of Repeat-Purchase Loyalty," Marketing
Letters, 7 (2), 163-171.
East, Robert and Annik Hogg (1997), "The Anatomy of Conquest: Tesco versus Sainsbury," Journal of
Brand Management, 5 (1), 53-60.
East, Robert, Annik Hogg and Wendy Lomax (1998) "The Future of Retail Loyalty Schemes," Journal
of Targeting, Measurement and Analysis for Marketing, 7 (1), 11-21.
Edwardson, Michael (1998), "Measuring Consumer Emotions in Service Encounters: An Exploratory
Analysis," Australasian Journal of Market Research, 6 (2), 34-48.
Ehrenberg, A.S.C. (1974), "Repetitive Advertising and the Consumer," Journal of Advertising
Research, 14, 25-34.
Ehrenberg, A. S.C. (1988), Repeat-Buying: Facts, Theory and Applications. New York: OUP. 2nd
edition.
Ehrenberg, A. S. C., Gerald J. Goodhardt and Patrick Barwise (1990), "Double Jeopardy Revisited,"
Journal of Marketing, 54 (3), 82-91.
Ehrenberg, A.S.C., Kathy Hammond and Gerald J. Goodhardt (1994), "The After-Effects of PriceRelated Consumer Promotions," Journal of Advertising Research, July-August, 1-10.
37
Ehrenberg, A.S.C. and Mark Uncles (1997), "Dirichlet-Type Markets: A Position Paper," Working
Paper, London: South Bank Business School; Sydney: University of New South Wales.
Eliashberg, Josh and Lilien, Garry L. (1993), Handbooks in Operations Research and Management
Science: Volume 5 - Marketing. Amsterdam: North-Holland.
Engel, James F., Roger D. Blackwell and Paul W. Miniard (1993), Consumer Behavior. Orlando, FL:
Dryden Press (7th edition).
Fader, Peter S. (1993), "Integrating the Dirichlet-Multinomial and Multinomial Logit Models of Brand
Choice," Marketing Letters, 4 (April), 99-112.
Fader, Peter S. and Hardie, Bruce (1996), "Modeling Consumer Choice Among SKUs," Journal of
Marketing Research, 33 (November), 442-452.
Fader, Peter S. and David Schmittlein (1993), "Excess Behavioral Loyalty for High-Share Brands:
Deviations from the Dirichlet Model for Repeat Purchasing," Journal of Marketing Research, 30 (4),
478-93.
Farley, John U. (1964), "Why Does 'Brand Loyalty' Vary Over Products?," Journal of Marketing
Research, 9-14.
Fishbein, M. (1981), "The Search for Attitudinal-Behavioural Consistency". In Harold Kassarjian and
Tom Robertson (Eds.) Perspectives in Consumer Behavior. Glenview, IL: Scott Foresman and
Company.
Fournier, Susan and Julie L.Yao (1997), "Reviving Brand Loyalty: A Reconceptualization Within the
Framework of Consumer-Brand Relationships," International Journal of Research in Marketing, 14
(5), 451-472.
Fournier, Susan (1998), "Consumers and Their Brands: Developing Relationship Theory in Consumer
Research," Journal of Consumer Research, 24 (March), 343-373.
Fournier, Susan, Susan Dobscha and David Glen Mick (1998), "Preventing the Premature Death of
Relationship Marketing," Harvard Business Review, January-February (pages??)
Foxall, Gordon R. (1987), "Radical Behaviorism and Consumer Research: Theoretical Promise and
Empirical Problems," International Journal of Research in Marketing, 4 ( ), 111-129.
Foxall, Gordon R. and R.E. Goldsmith (1994), Consumer Psychology for Marketing. London:
Routledge.
Frank, Ronald E. (1967), "Is Brand Loyalty a Useful Basis for Market Segmentation?," Journal of
Advertising Research, 7 (2), 27-33.
Glazer, Rashi and A.M. Weiss (1993) "Marketing in Turbulent Environments: Decision Processes and
the Time-Sensitivity of Information," Journal of Marketing Research, 30 (4), 509-521.
38
Goodhardt, Gerald G., A.S.C. Ehrenberg and Chris Chatfield (1984), "The Dirichlet: A Comprehensive
Model of Buying Behaviour", Journal of the Royal Statistical Society A, 147, 621-655,
Gordon, Wendy (1994), "Taking Brand Repertoires Seriously", Journal of Brand Management, 2 (1),
25-30.
Hammond, Kathy, A.S.C. Ehrenberg and Gerald J. Goodhardt (1995), "The Lack of Brand
Segmentation," European Journal of Marketing, 30 (12), 39-49.
Hammond, Kathy (1997), "Brand Loyalty for Frequently-Bought Goods", Unpublished PhD Thesis.
London: London Business School, University of London.
Heckman, James J. (1981), "Statistical Models for Discrete Panel Data," in Structural Analysis of
Discrete Data with Econometric Applications. Eds C. Manski and Daniel McFadden. Cambridge, MA:
MIT Press (114-178).
Helsen, Kristiaan and David C. Schmittlein (1993), "Analyzing Duration Times in Marketing: Evidence
for the Effectiveness of Hazard Rate Models," Marketing Science, 12 (Fall), 395-414.
Hennig-Thurau, T. and Klee, A. (1997), "The Impact of Customer Satisfaction and Relationship
Quality on Customer Retention: A Critical Reassessment and Model Development", Psychology and
Marketing, 14 (8), 737-764.
Hoyer, W.D. (1984), "An Examination of Consumer Decision-Making for a Common Repeat-Purchase
Product," Journal of Consumer Research, 11 (3), 822-829.
Holbrook, Morris (1978), "Beyond Attitude Structure: Toward the Informational Determinants of
Attitude," Journal of Marketing Research, 15 (November), 545-556.
Howard, John A. and Jag N. Sheth (1969), The Theory of Buyer Behavior. New York: John Wiley.
Iacobucci, Dawn and A. Ostrom (1996), "Commercial and Interpersonal Relationships: Using the
Structure of Interpersonal Relationships to Understand Individual-to-Individual, Individual-to-Firm,
Firm-to-Firm Relationships in Commerce," International Journal of Research in Marketing, 13, 53-72.
Jacoby, Jacob and Robert Chestnut (1978), Brand Loyalty Measurement and Management. New
York: John Wiley.
Jain, Dipak, Frank M. Bass and Y.M. Chen (1990), "Estimation of Latent Class Models With
Heterogeneous Choice Probabilities: An Application to Market Structuring," Journal of Marketing
Research, 27, 94-101.
Jones, Morgan J. and Fred Zufryden (1980), "Adding Explanatory Variables to a Consumer Purchase
Behavior Model: An Exploratory Study," Journal of Marketing Research, 17, 323-334.
Jones, T.O. and Sasser, W.E. (1981), "Why Satisfied Customers Defect," Harvard Business Review
(November-December), 88-99.
39
Kahn, Barbara, M. U. Kalwani and Donald G. Morrison (1988), "Niching versus Change-of-Pace
Brands: Using Purchase Frequencies and Penetration Rates to Infer Brand Positionings," Journal of
Marketing Research, 25 (4), 384-390.
Kayande, Ujwal (1997), "Theory of Generalizability and optimization of Marketing Measurement,"
Unpublished PhD Thesis. Alberta: University of Alberta.
Kearney, T.J. (1990), "Frequent Flyer Programs: A Failure in Competitive Strategy, with Lessons for
Management", Journal of Consumer Marketing, 7 (Winter), 31-40.
Keller, Kevin (1998) Brand Management
Kuehn, A.A. (1962), "Consumer Brand Choice as a Learning Process", Journal of Advertising
Research, 2, 10-17.
Lastovicha, J.L. and E.H. Bonfield (1982), "Do Consumers Have Brand Attitudes?," Journal of
Economic Psychology, 2 (1), 57-75.
Lattin, James M. and Leigh McAlister (1985), "Using a Variety-Seeking Model to Identify Substitute
and Complementary Relationships Among Competing Products," Journal of Marketing Research, 22
(2), 330-339.
Leong, S.M. (1993), "Consumer Decision-Making for a Common Repeat-Purchase Product: A Dual
Replication," Journal of Consumer Psychology, 2 (2), 193-208.
Lipstein, Benjamin (1959), "The Dynamics of Brand Loyalty and Brand Switching". Proceedings of
the 1959 ARF Conference. New York: Advertising Research Foundation.
Lutz, Richard J. and Paul R. Winn (1974), "Developing a Bayesian Measure of Brand Loyalty: A
Preliminary Report." In R.C. Corhan, ed., Proceedings of the 1974 AAM Conference. Chicago:
American Marketing Association.
McPhee, William (1963), Formal Theories of Mass Behavior. New York: New York Press.
Mason, T. (1998), "The Best Shopping Trip? How Tesco Keeps the Customer Satisfied", Journal of
the Market Research Society, 40 (1), 5-12.
Massy, William F., David B. Montgomery and Donald G. Morrison (1970), Stochastic Models of Buyer
Behavior. Boston: MIT Press.
Mellens, Michael, Marnik G. Dekimpe and Jan-Benedict E. B. Steenkamp (1996), "A Review of BrandLoyalty Measures in Marketing", Tijdschrift voor Economie en Management, 41 (4), 507-533.
Morgan, Robert M. and Shelby D. Hunt (1994), "The Commitment-Trust Theory of Relationship
Marketing", Journal of Marketing, 58 (July), 20-38.
Nalebuff, Barry J. and Adam M. Brandenburger (1996), Co-opetition. London: Harper-Collins
Business.
40
Oliver, Richard L. (1980), "Cognitive Model of the Antecedents and Consequences of Satisfaction
Decisions," Journal of Marketing Research, 17 (November), 460-469.
Oliver, Richard L. (1997), Satisfaction: A Behavioral Perspective on the Consumer. New York:
McGraw-Hill.
Oliver, Richard L. (1998), "Consumer Post-Usage Evaluation: Whence Loyalty?," MSI Conference on
Fundamental Issues and Directions in Marketing. Boston, MA: Marketing Science Institute.
Olshavsky, R.W. and Granbois, D.H. (1979), "Consumer Decision Making - Fact or Fiction?," Journal
of Consumer Research, 6, 93-100.
Palmer, Adrian and Rosalind Beggs (1997), "Loyalty Programmes: Congruence of Market Structure
and Success," Proceedings of the Academy of Marketing Conference. Manchester: Manchester
Metropolitan University.
Pearson, Stewart (1996), Building Brands Directly. London: Macmillan Business.
Peppers, Don and Martha Rogers (1993), The One to One Future: Building Relationships One
Customer at a Time. New York: Currency Doubleday.
Peppers, Don and Martha Rogers (1997), Enterprise One to One: Tools for Competing in the
Interactive Age. New York: Currency Doubleday.
Petrison, L.A., Robert C. Blattberg and Paul Wang (1993), "Database Marketing: Past, Present and
Future”, Journal of Direct Marketing, 7 (3), 27-43.
Piercy, Nigel (1996), Market-Led Strategic Change: Transforming the Process of Going to Market.
Oxford: Butterworth-Heinemann. 2nd edition.
Raju, P.S. and M. Hastak (1980), "Consumer Response to Deals: A Discussion of Theoretical
Perspectives", in J.C. Olson, ed., Advances in Consumer Research, Vol. 7, Provo, UT: Association for
Consumer Research.
Reibstein, David J. and Paul W. Farris (1995), "Market Share and Distribution: A Generalization, A
Speculation and Some Implications," Marketing Science, 14 (3), Part 2 of 2, G190-G202.
Reichheld, Frederick F. and Sasser, William E. (1990), "Zero Defections: Quality Comes to Services,"
Harvard Business Review, 68 (September-October), 105-111.
Reichheld, Frederick F. (1996), The Loyalty Effect: The Hidden Force Behind Growth, Profits and
Lasting Value. Boston, MA: Harvard Business School Press.
Robinson, W.T. and Claus Fornell (1985), "Sources of Market Pioneer Advantages in
Consumer Goods Industries" Journal of Marketing Research, 22 (3), 305-317.
Rothschild, L. and W.C. Gaidis (1981), "Behavioral Learning Theory: Its Relevance to Marketing and
Promotions", Journal of Marketing, 45 (2), 70-78.
41
Rossiter, John R. and Larry Percy (1997), Advertising Communications and Promotion Management,
New York: McGraw-Hill. 2nd edition.
Russell, Garry J. and Wagner Kamakura (1994), "Understanding Brand Choice Using Micro and Macro
Scanner Data," Journal of Marketing Research, 31 (May), 289-303.
Rust, Roland and A.J. Zahorik (1993), "Customer Satisfaction, Customer Retention, and Market
Share," Journal of Retailing, 69 (Summer), 193-215.
Schmittlein, David C., Albert C. Bemmaor and Donald G. Morrison (1985), "Why Does the NBD
Model Work? Robustness in Representing Product Purchases, Brand Purchases and Imperfectly
Recorded Purchases," Marketing Science, 4 ( ), 255-266.
Schouten, John W. and James H. Alexander (1995), "Subcultures of Consumption: An Ethnography of
the New Bikers," Journal of Consumer Research, 22 (June), 43-61.
Schultz, Don E. and William A.Robinson (1982), Sales Promotion Essentials. Chicago: Brian Books.
Sharp, Byron (1998), "Comment on the Future of Loyalty Schemes: Can Loyalty Schemes Reduce
Customer Vulnerability?," Journal of Targeting, Measurement and Analysis for Marketing, 7 (1), 2224.
Sharp, Byron and Sharp, Anne (1997), "Loyalty Programs and Their Impact on Repeat-Purchase
Loyalty Patterns", International Journal of Research in Marketing, 14 (5), 473-486.
Sheth, Jagdish, B. Newman and B. Gross (1991), "Why We Buy What We Buy: A Theory of
Consumption Values", Journal of Business Research, 22 (2), 159-170.
Skinner, B.F. (1953), Science and Human Behaior. New York: Free Press.
Smith, Robert E. and William R. Swinyard (1983), "Attitude-Behavior Consistency: The Impact of
Product Trial Versus Advertising," Journal of Marketing Research, 20 (August), 257-267.
Smothers, N. (1993) "Can Products and Brands Have Charisma?" in Brand Equity and Advertising
edited by David A. Aaker and Alexander L. Biel (Lawrence Erlbaum Associates: Hillsdale, New Jersey)
Chapter 7, 97-111.
Stern, Phillip and Kathy Hammond (1998), "Modelling Long-Run Loyalty," Working Paper. Warwick:
Warwick Business School and London: London Business School.
Szymanski, David and Paul Busch (1987), "Identifying the Generics-Prone Consumer: A MetaAnalysis", Journal of Marketing Research, 24 (November), 425-432.
Szymanski, David, Sundar G. Bharadwaj, and P.Rajan Varadarajan (1993), "An Analysis of the Market
Share-Profitability Relationship", Journal of Marketing, 57 (July), 1-18.
Tellis, Gerald J. (1988), "The Price Elasticities of Selective Demand: A Meta-Analysis of Econometric
Models and Sales," Journal of Marketing Research . (.), 331-341.
42
Totten, John C. and Martin P. Block (1986), Analyzing Sales Promotion: Text and Cases. Chicago, IL:
Commerce Communications Inc.
Tucker, William T. (1964), "The Development of Brand Loyalty," Journal of Marketing Research, 1
(3), 32-35.
von Hippel, Eric (1986), "Lead Users: A Source of Novel Product Concepts", Management Science,
32 (7), 791-805.
Uncles, Mark (1988) (Ed.), Longitudinal Data Analysis: Methods and Applications. Pion, London
Uncles, Mark (1994), "The Seven Deadly Perils of Loyalty Programs", The Marketing Society Review,
Autumn, 18-20.
Uncles, Mark and Laurent, Gilles (1997), "Editorial: Loyalty Marketing", International Journal of
Research in Marketing, 14 (5), 399-404.
Uncles, Mark, A.S.C. Ehrenberg and Kathy Hammond (1995), "Patterns of Buyer Behavior:
Regularities, Models and Extensions", Marketing Science, 14 (3), Part 2 of 2, G71-G78.
Uncles, Mark, Kathy Hammond, A.S.C. Ehrenberg and R.E. Davis (1994), "A Replication Study of
Two Brand-Loyalty Measures", European Journal of Operational Research, 76 (2), 375-384.
Voss, Kevin (1998), "The Expected Value of a Customer: A Model and Application," Unpublished
Dissertation Abstract. Tri-Cities: Department of Marketing, Washington State University.
Webster, F.E. (1992), "The Changing Role of Marketing in the Corporation", Journal of Marketing, 56
( ), 1-17.
Worthington, Stephen and Susan Horne (1996) "Relationship Marketing: The Case of the University
Alumni Affinity Credit Card", Journal of Marketing Management, 12, 189-199.
Wilkie, William L. and Peter R. Dickson (1985), Shopping for Appliances: Consumers' Strategies and
Patterns of Information Search. MSI Report 85-108. Cambridge, MA: Marketing Science Institute.
43
Figure 1. A Framework: Individual Consumer Loyalty
Habit
t-1, t-2, …
Past
Behaviour
Cognitive
Factors
Cognition
Brand
Component
Affective
Responses
Situational
Influences
Habit
t+1, t+2, …
Current
Behaviour
Loyalty
Program
Intentions
Past
Attitudes
Reinforcement
t-1, t-2, …
Functional
Value
Psychological
Value
Social
Norms
Current
Attitudes
Social
Value
Personal
Norms
Situational
Influences
Reinforcement
t+1, t+2, …
Figure 2. A Framework: Aggregate Consumer Loyalty
Double Jeopardy Line
under dynamic
equilibrium
1.0
A´
Level of
Loyalty
A´´
A
0.0
1%
Market Share
A
A´
A´´
A-A´´
Current level of loyalty at current share
Desired level of loyalty at current share
Level of loyalty at increased share
Range of most likely levels of loyalty
100%
Figure 3. A Framework: Measurement Issues
Measurable Variables
(e.g. repeat purchase, consumer attitudes,
value of loyal program, socio-demographics
of brand buyers and program members)
Observed
Purchase
Sequence
t-1, t-2, …
Measurable
Situational
Influences
(e.g. competitive offers)
Past
Behavior
Current
Behavior
Brand
Component
Past
Attitudes
Loyalty
Program
for i = 1, ….N
Intentions
for i = 1, ….N
Current
Attitudes
for i = 1, ….N
for i = 1, ….N
Unmeasurable Variables
(e.g. personal tastes)
Unmeasurable
Situational
Influences
(e.g. fleeting emotions)
Predicted
Purchase
Sequence
t+1, t+2, …