ArticlePDF Available

Abstract and Figures

Referral programs have become a popular way to acquire customers. Yet there is no evidence to date that customers acquired through such programs are more valuable than other customers. The authors address this gap and investigate the extent to which referred customers are more profitable and more loyal. Tracking approximately 10,000 customers of a leading German bank for almost three years, the authors find that referred customers (1) have a higher contribution margin, though this difference erodes over time; (2) have a higher retention rate, and this difference persists over time; and (3) are more valuable in both the short and the long run. The average value of a referred customer is at least 16% higher than that of a nonreferred customer with similar demographics and time of acquisition. However, the size of the value differential varies across customer segments; therefore, firms should use a selective approach for their referral programs.
Content may be subject to copyright.
Journal of Marketing
Philipp Schmitt, Bernd Skiera, & Christophe Van den Bulte
Referral programs have become a popular way to acquire customers. Yet there is no evidence to date that
customers acquired through such programs are more valuable than other customers. The authors address this gap
and investigate the extent to which referred customers are more profitable and more loyal. Tracking approximately
10,000 customers of a leading German bank for almost three years, the authors find that referred customers (1)
have a higher contribution margin, though this difference erodes over time; (2) have a higher retention rate, and
this difference persists over time; and (3) are more valuable in both the short and the long run. The average value
of a referred customer is at least 16% higher than that of a nonreferred customer with similar demographics and
time of acquisition. However, the size of the value differential varies across customer segments; therefore, firms
should use a selective approach for their referral programs.
Keywords: customer referral programs, customer loyalty, customer value, customer management, word of mouth,
social networks
Philipp Schmitt is a doctoral student (e-mail: pschmitt@wiwi.uni-frankfurt.
de), and Bernd Skiera is Professor of Marketing and Member of the Board
of the E-Finance Lab at the House of Finance (e-mail:,
Schoo l of Business and Economics, Goethe University Frankfurt .
Christophe Van den Bulte is Associate Professor of Marketing, the Whar-
ton School, University of Pennsylvania (e-mail: vdbulte@wharton.upenn.
edu). The authors thank the management of a company that wants to
remain anonymous for making the data available and Christian Barrot,
Jonah Berger, Xavier Drèze, Peter Fader, Jeanette Heiligenthal, Gary
Lilien, Renana Peres, Jochen Reiner, Christian Schulze, Russell Winer,
Ezra Zuckerman, and the anonymous JM reviewers for providing com-
ments on previous drafts of this article.
&ord of mouth (WOM) has reemerged as an impor-
tant marketing phenomenon, and its use as a cus-
tomer acquisition method has begun to attract
renewed interest (e.g., Godes and Mayzlin 2009; Iyengar,
Van den Bulte, and Valente 2011). Traditionally, WOM’s
appeal has been in the belief that it is cheaper than other
acquisition methods. A few recent studies have documented
that customers acquired through WOM also tend to churn
less than customers acquired through traditional channels
and that they tend to bring in additional customers through
their own WOM (Choi 2009; Trusov, Bucklin, and Pauwels
2009; Villanueva, Yoo, and Hanssens 2008). Villanueva,
Yoo, and Hanssens (2008) further suggest that customers
acquired through WOM can generate more revenue for the
firm than customers acquired through traditional marketing
From a managerial point of view, these findings are
encouraging and suggest purposely stimulating WOM to
acquire more customers. However, there are concerns that
firm-stimulated WOM may be substantially less effective
than organic WOM in generating valuable customers
(Trusov, Bucklin, and Pauwels 2009; Van den Bulte 2010)
because (1) targeted prospects may be suspicious of stimu-
lated WOM efforts; (2) such efforts often involve a mone-
tary reward for the referrer, who, as a result, may seem less
trustworthy; (3) programs providing economic benefits tend
not to be sustainable (Lewis 2006); (4) unlike organic
WOM, stimulated WOM is not free, raising questions about
cost effectiveness; and (5) stimulated WOM is prone to
abuse by opportunistic referrers.
The uncertainty about the benefits of stimulated WOM
in customer acquisition is frustrating for managers facing
demands to increase their marketing return on investment
and considering whether to use this method. Our study
addresses this managerial issue by investigating the value of
customers acquired through stimulated WOM and compar-
ing it with the value of customers acquired through other
methods. We do so in the context of a specific WOM mar-
keting practice that is gaining prominence, namely, referral
programs in which the firm rewards existing customers for
bringing in new customers. Although these programs are
typically viewed as an attractive way to acquire customers,
their benefits are often viewed to be their targetability and
cost effectiveness (Mummert 2000). We broaden this view
by assessing the value of customers acquired through these
types of programs.
Specifically, we answer four questions: (1) Are cus-
tomers acquired through a referral program more valuable
than other customers? (2) Is the difference in customer
value large enough to cover the costs of such stimulated
WOM customer acquisition efforts? (3) Are customers
acquired through a referral program more valuable because
they generate higher margins, exhibit higher retention, or
both? and (4) Do differences in margins and retention
remain stable, or do they erode? The answers to the last two
questions provide deeper insight into what might be driving
the value differential.
We answer these four questions using panel data on all
5181 customers that a leading German bank acquired
through its referral program (referred customers) between
January 2006 and December 2006 and a random sample of
4633 customers the same bank acquired through other
methods (nonreferred customers) over the same period. For
both groups of customers, we track profitability (measured
as contribution margin) and loyalty (measured as retention)
at the individual level from the date of acquisition until Sep-
tember 2008. The total observation period spans 33 months.
We use two metrics of customer value: (1) the present value
of the actually observed contribution margins realized
within the data window and (2) the expected present value
over a period of six years from the day of acquisition.
Although our study is limited to a single research site, as is
common for studies that require rich and confidential data,
the methodology and findings are of broad interest. Cus-
tomer referral programs are gaining popularity in many
industries, including financial services, hotels, automobiles,
newspapers, and contact lenses (Ryu and Feick 2007).
We make the following contributions: First, we provide
empirical evidence that a referral program, a form of stimu-
lated WOM, is an attractive way to acquire customers.
Referred customers exhibit higher contribution margins,
retention, and customer value. Second, building on our find-
ing that differences in contribution margin erode over time
whereas those in retention do not, we document that
referred customers are more valuable in both the short and
the long run. Third, we show that the referral effect need not
be present in every customer segment. Finally, we illustrate
how the type of analysis we conduct enables firms to calcu-
late the return on investment and the upper bound for the
reward in their customer referral programs.
We proceed by describing referral programs and develop-
ing our hypotheses. A description of the research setting,
the data, and the model specifications follows. Then, we
report the results. Finally, we discuss implications for prac-
tice, the limitations, and opportunities for further research.
;9:53,8!,-,88(2 85.8(39
Customer referral programs are a form of stimulated WOM
that provides incentives to existing customers to bring in
new customers. An important requirement for such pro-
grams is that individual purchase or service histories are
available so the firm can ascertain whether a referred cus-
tomer is indeed a new rather than an existing or a former
Referral programs have three distinctive characteristics.
First, they are deliberately initiated, actively managed, and
continuously controlled by the firm, which is impossible or
very difficult with organic WOM activities such as sponta-
neous customer conversations and blogs. Second, the key
idea is to use the social connections of existing customers
with noncustomers to convert the latter. Third, to make this
conversion happen, the firm offers the existing customer a
reward for bringing in new customers.
Although leveraging the social ties of customers with
noncustomers to acquire the latter is not unique to customer
referral programs, the three distinctive characteristics of
these programs set them apart from other forms of network-
based marketing (Van den Bulte and Wuyts 2007). Unlike
Referral Programs and Customer Value / 47
organic WOM, the firm actively manages and monitors
referral programs. Unlike most forms of buzz and viral mar-
keting, the source of social influence is limited to existing
customers rather than anyone who knows about the brand or
event. Unlike multilevel marketing, existing customers are
rewarded only for bringing in new customers. They do not
perform any other sales function (e.g., hosting parties) and
do not generate any income as a function of subsequent
sales. Consequently, referral programs do not carry the
stigma of exploiting social ties for mercantile purposes, as
multilevel marketing does (Biggart 1989).
In most referral programs, the reward is given regardless
of how long the new referred customers stay with the firm.
Such programs are prone to abuse by customers. Although
the firm does not commit to accept every referral, the incen-
tive structure combined with imperfect screening by the
firm creates the potential for abuse in which existing cus-
tomers are rewarded for referring low-quality customers.
This kind of moral hazard is less likely to occur with WOM
campaigns that do not involve monetary rewards condi-
tional on customer recruitment.
Existing studies of customer referral programs have
provided guidance about when rewards should be offered
(Biyalogorsky, Gerstner, and Libai 2001; Kornish and Li
2010), have quantified the impact of rewards and tie
strength on referral likelihood (Ryu and Feick 2007; Wirtz
and Chew 2002), and have quantified the monetary value of
making a referral (Helm 2003; Kumar, Petersen, and Leone
2007, 2010). The key managerial question of the (differen-
tial) value of customers acquired through referral programs
has not yet been addressed.
Because referral programs are a customer acquisition
method, an important metric to assess their effectiveness is
the value of the customers they acquire. Additional insights
come from investigating differences between referred and
nonreferred customers in contribution margins and retention
rates, the two main components of customer value (e.g.,
Gupta and Zeithaml 2006; Wiesel, Skiera, and Villanueva
Our hypotheses regarding these customer metrics of
managerial interest are informed by prior work in economics
and sociology on employee referral (e.g., Coverdill 1998;
Rees 1966), especially the work of Fernandez, Castilla, and
Moore (2000), Neckerman and Fernandez (2003), and
Castilla (2005) on the quality of employee referral programs.
These studies show that the benefits of such programs are
realized through distinct mechanisms, of which better
matching and social enrichment appear particularly relevant
to marketers. Better matching is the phenomenon that refer-
rals fit with the firm better than nonreferrals, and social
enrichment is the phenomenon that the relationship of the
referral to the firm is enriched by the presence of a common
third party (i.e., the referrer).
Customer and employees referral programs are likely to
be subject to similar mechanisms because they share the
three distinctive characteristics of having active manage-
ment, using the social connections of existing contacts, and
offering rewards with the risk of abuse. Selecting a new
employer or bank both are high-involvement decisions
involving a fair amount of uncertainty. Although some basic
banking products, such as checking accounts, are well
known to most customers, the wider set of financial services
that banks provide are considered experience goods rather
than search goods (e.g., Bolton, Freixas, and Shapiro 2007;
Parasuraman, Zeithaml, and Berry 1985). The recurrent
losses of many private investors indicate that many people
are not very skilled at assessing complex bank offerings.
We use the better matching and social enrichment
mechanisms to develop and motivate our hypotheses. How-
ever, our goal is to document managerially relevant differ-
ences in contribution margin, retention, and customer value
rather than to test those specific mechanisms. The mecha-
nisms are only possible explanations for the differences we
Differences in Contribution Margin
Several characteristics of social dynamics in human net-
works (e.g., Van den Bulte and Wuyts 2007) imply that
referred customers may match up with the firm better than
other newly acquired customers. The first is reciprocity.
Because referring customers receive a reward, norms of reci-
procity may make nonopportunistic customers feel obliged
to bring in new customers who they think may be valuable
to the firm (Gouldner 1960). This process explains Necker-
man and Fernandez’s (2003) finding that referrals for which
the referrer claimed a fee had lower turnover than referrals
for which no fee was claimed. The second social dynamic
likely to be at work is triadic balance. If the main function
of the program is simply to nudge customers into making
referrals without much consideration for the monetary
reward (Thaler and Sunstein 2008), principles of triadic bal-
ance will make existing customers more likely to bring in
others who they believe would match well with what the
firm has to offer. A third social dynamic likely to be at work
is homophily—the tendency for people to interact with
people like them. Whereas reciprocity and triadic balance
imply that referrers are diligent and active in screening and
matching peers to firms, homophily implies that customers
are likely to refer others who are similar to themselves.
Because existing customers have an above-average chance
of being a good match (otherwise, they would not be cus-
tomers), firms may benefit from referral programs through
“passive” homophily-based matching rather than only
deliberate “active” screening-based matching by referrers
(Kornish and Li 2010; Montgomery 1991; Rees 1966).
Acquisition through referral can also result in informa-
tional advantages, making referred customers more prof-
itable than other customers. Referred customers are likely
to have discussed the firm’s offerings with their referrer. As
a result, they are likely to use its products more extensively
than novice customers who take a more cautious approach
in building involvement. Informational advantages to the
firm can also result if people refer others similar to them-
selves on dimensions that are relevant to the enjoyment of
the product or service but are not immediately observable to
the firm (Kornish and Li 2010). Examples for financial ser-
vices include risk aversion and a sense of fiscal responsibil-
48 / Journal of Marketing, January 2011
ity. In these situations, the firm can make inferences from
the observed behavior of the referrers about the products in
which the referred customers will be most interested (e.g.,
Guseva 2008). As a result, the firm is able to serve the
referred customer in a tailored way early on, something that
takes time to learn for other newly acquired customers.
Because of this informational advantage, the firm should be
able to generate a higher contribution margin from referred
customers at the beginning of the relationship.
However, the advantages of better matching and better
information should gradually vanish. As nonreferred cus-
tomers accumulate experience with the firm, they should
become equally well informed about the firm’s offerings
and procedures. Likewise, the firm should be able to use the
purchase and service history of the nonreferred customers
to serve them better. Furthermore, nonreferred customers
who are not a good match for the firm are more likely to
leave. Consequently, both revenues and costs of referred
and nonreferred customer should converge, eliminating the
difference in contribution margin over time. Thus:
H1: (a) The average contribution margin of a customer acquired
through a referral program is higher than that of a cus-
tomer acquired through other methods, but (b) this differ-
ence becomes smaller over time.
Differences in Retention
Social enrichment is another mechanism that may increase
the value of referred customers. The argument is that the
relationship with the firm is enriched because a family
member or friend is a customer of the same firm (Castilla
2005; Fernandez, Castilla, and Moore 2000). Having a per-
son close to oneself in a similar position (i.e., being a cus-
tomer of the same firm) should increase the person’s trust in
the firm and strengthen his or her emotional bond with it, as
both balance theory and social closure theory predict (Van
den Bulte and Wuyts 2007). This prediction is also consistent
with findings that customers reflecting on their affect toward
a firm mention friends who are customers with the same
firm (Yim, Tse, and Chan 2008). Such relationships should
be particularly relevant in a banking context, in which emo-
tions and trust play important roles in the customer–firm
relationship (e.g., Edwards and Day 2005; Fleming, Coff-
man, and Harter 2005). In short, referred customers are
likely to have a stronger sense of commitment and attach-
ment to the firm. This implies that referred customers are
less likely to churn than nonreferred customers, provided
that their referrer does not churn either. The latter condition
is likely to hold: Referrers typically have a greater long-
term likelihood of staying, which is why intention to refer is
frequently used as an indicator of loyalty (Gupta and Zeit-
haml 2006).
Although the informational advantage of a referred cus-
tomer decreases over time as direct experience substitutes
for social learning, there is no reason to expect erosion in
the affective bonding underlying the social enrichment
mechanism. Consequently, the erosion of the differential
expected in contribution margin need not apply to retention.
Therefore, we state the following:
H2: (a) The average retention of a customer acquired through a
referral program is higher than that of a customer acquired
through other methods, and (b) this difference does not
become smaller over time.
Differences in Customer Value
If H1and H2hold and if the erosion of contribution margins
does not outweigh the initial difference in margins and the
persisting difference in retention, the following should hold
as well:
H3: The average value of a customer acquired through a refer-
ral program is higher than that of a customer acquired
through other methods.
H3can hold even when H1and H2do not, because it is pos-
sible for the differences in both margins and retention to be
small but for their combined effect to be sizable and signifi-
cant. Another reason to test H3on customer value, in addi-
tion to H1and H2on margins and retention, is that customer
value is what managers should care about most.
Although we base our prediction on sound theoretical
arguments, the posited effects are not as obvious as they
may seem because there are several competing forces at
work. First, the prospect of earning a referral fee may
induce referrers to pressure their peers to become customers
without giving much consideration to whether the firm
actually matches their peers’ needs. Second, the relationship
between the referred customer and his or her referrer, which
is necessary for social enrichment to operate, comes with an
inherent risk: When referrers defect, the customers they
brought in may become more likely to leave as well.
Although it seems unlikely that referrers as a whole are
more prone to churn than the average customer, the risk of
contagious defection should not be ignored altogether.
Third, an abuse of the referral program by customers who
are interested only in the monetary reward is probably the
most important reason for practitioners’ skepticism. This is
illustrated by TiVo’s termination of its referral program as a
result of frequent abuses (ZatzNotFunny 2008).
Support for our hypotheses would allow us to conclude
that the positive effects outweigh the negative ones. In addi-
tion, the empirical analysis provides not only a test of the
hypotheses but also an estimate of the size of the customer
value differential. Firms can use the latter to put a maxi-
mum value on the reward to be paid out as part of their
referral program.
Research Setting
We use data from a leading German bank, whose name we
do not divulge for confidentiality reasons. The data capture
all customers acquired through the bank’s referral program
between January 2006 and December 2006 and a random
sample of customers acquired through other methods (e.g.,
direct mail, advertising) over the same period. The latter
group may include customers affected by organic WOM for
which the bank did not pay any fee. To the extent that the
value of customers acquired through organic WOM is equal
Referral Programs and Customer Value / 49
to or greater than that of customers acquired through the
referral program, our results underestimate the value differ-
ential between WOM and non-WOM customers. Regard-
less, we correctly estimate the value differential between
customers acquired through the referral program and all
other customers for whom no referral fee was paid.
The observation period spans from January 2006 to
September 2008 (33 months), and the data on each individ-
ual customer include the day of acquisition, the day of leav-
ing the bank (if applicable), the contribution margin in each
year, and some demographics. In total, we have data on
5181 referred and 4633 nonreferred customers. Because the
referral program was used only in a business-to-consumer
context, all customers are individual people.
The referral program was communicated to existing
customers through staff and flyers in the branches and
through mailings.1The procedure was straightforward:
Every existing customer who brought in a new customer
received a reward of 25 euros in the form of a voucher that
could be used at several well-known German retailers.2
Except for opening an account, the referred customer did not
need to meet any prerequisites (e.g., minimum amount of
assets, minimum stay) for the referrer to receive the reward.
In addition, 2006 was not an unusual year in terms of
customer acquisition, and the bank’s management was con-
fident that findings about customers acquired in 2006 would
be applicable to customers acquired in earlier and later
years. Proprietary information of the bank shows that its
customers are similar to those of other leading European
banks. Regarding the usage of its referral program and the
response of its customers to it, no differences with other
firms are apparent either.
Dependent Variables
Daily contribution margin is the individual contribution
margin on a daily basis. It is the total contribution margin
the customer generates in the observation period, divided by
the total number of days the customer was with the bank
(duration). This per diem scaling ensures the comparability
of the contribution margin of customers with different
observed (and possible censored) durations. The contribu-
tion margin equals revenue (interest and fees) less direct
costs (e.g., interest expenses, sales commissions, brokerage,
trading costs). The acquisition costs are not subtracted from
the contribution margin. We also compute a time-varying
version of daily contribution margin by dividing the contri-
bution margin generated by the customer in a particular
year (2006, 2007, 2008) by the number of days the cus-
tomer was with the bank in that year.
1These mailings went to the referring customers. Mailings to
which the nonreferred customers responded were sent directly to
2Although confidentiality concerns preclude us from reporting
the average cost of acquisition for referral and nonreferral cus-
tomers, we can report that the total acquisition cost for referred
customers (including not only the referral fee but also the addi-
tional administrative costs of record keeping, paying out, and so
on) is on average approximately 20 euros lower than that for non-
referred customers acquired through mailings.
Duration is the total number of days the customer was
observed to be with the bank. It is the basis for analyzing
We calculate two measures of customer value, one
based only on observed data and the other based on both
observed data and predictions. Observed customer value is
the present value of all actual contribution margins the cus-
tomer generated during the whole observation period (e.g.,
33 months for retained customers acquired in January
2006). This metric is affected by both contribution margin
and retention because a customer generates no margins after
leaving the bank. Our second metric, customer lifetime
value, is the present value of all contribution margins, both
actual and predicted, of the customer within the six-year
span following the day of acquisition.3For customers who
churned during the observation period, customer lifetime
value equals observed customer value because their pre-
dicted value is 0. The formulas are as follows:
(1) Customer Lifetime Valuei= Observed Customer Valuei
+ Predicted Customer Valuei,
where OMis is the observed monthly contribution margin of
customer i in the sth month after acquisition (calculated
from the observed annual contribution margin and the
observed duration), Duriis the customer’s observed lifetime
with the bank in months, diis a dummy censoring variable
that indicates whether the customer was still with the bank
by the end of the observation period, PMis is the predicted
monthly contribution margin of customer i in the sth month
after acquisition, PAis is the predicted probability that cus-
tomer i is still “alive” (i.e., with the bank) in that month,
and r is the firm-specific annual discount rate of 11.5%.4
The present value reflects what the customer is worth at the
day of acquisition.
Independent Variables
The independent variable of central interest is referral pro-
gram, a binary indicator that takes the value 1 for referred
customers (i.e., those who were acquired through the referral
program) and 0 for nonreferred customers. To improve the
comparability of referred and nonreferred customers, we con-
trol for the demographic variables age, sex (dummy variable,
( ) ( ) ,2 112
Observed Customer Valuei
Dur OM
( ) ( )
Predicted Customer Valuei i
is is
== +
50 / Journal of Marketing, January 2011
with women coded as 1 and men coded as 0), and marital
status (dummy variables for married, divorced/separated,
single, and widowed, with no answer as the base category).
We also control for the customer’s month of acquisition (11
dummy variables for each month, with December 2006 as
the base category).
To assess the robustness of the difference in customer
value, we also conduct separate analyses for the two key
segments of the bank: retail customers with standard finan-
cial needs and nonretail customers with significant assets or
requiring more sophisticated financial advice. This segmen-
tation scheme the bank uses is based on a comprehensive
analysis of both financial data (e.g., assets invested with the
bank, monthly checking account balance) and demographic
information (e.g., profession, place of residence). The seg-
ments form the basis for all strategic customer-related deci-
sions of the bank.
Descriptive Statistics
The sample includes several customers with an extremely
high daily contribution margin that is up to 80 standard
deviations above the mean and median. Such extreme data
points can influence comparisons of means and regression
results, so we purify the data using the standard DFBETA
and DFFIT criteria to identify influence points (Belsley,
Kuh, and Welsch 1980). This procedure led to the deletion
of 172 referred customers (3.3% of the original 5181
referred customers) and 147 nonreferred customers (3.2%
of the original 4633 nonreferred customers). As we report in
the subsection “Robustness Checks,” testing the hypotheses
without deleting the influence points results in larger differ-
ences and provides stronger support for the hypotheses. Yet
the size estimates obtained without the influence points are
more informative.
Table 1 presents the means, standard deviations, and the
correlation matrix for the purified sample of 9495 cus-
tomers. As the nonzero correlations between the referral
program variable and the various demographic and time of
acquisition variables indicate, the groups of referred and
nonreferred customers are not perfectly matched on demo-
graphics and time of acquisition. Thus, it is desirable to
control for these differences.
Figure 1 plots the average daily contribution margin for
the referred and nonreferred customers of the purified sam-
ple, for 2006, 2007, and 2008. The pattern is encouraging.
Referred customers tend to generate higher margins, and
the margins tend to erode more quickly among referred cus-
tomers, such that the margin differential is narrower in 2008
than in 2006 (8 cents versus 18 cents per day). Similarly, as
Figure 2 shows, after about a year, the retention rate of
referred customers is higher, and this is the case regardless
of duration. However, controlling for differences in demo-
graphics and time of acquisition is necessary to draw con-
clusions more confidently.
Statistical Models
To estimate the difference in contribution margin between
acquisition modes (H1a), we use a regression model with
the following specification:
3This way, we do not need to predict margins and retention rates
beyond four years after the end of the data window, and the result-
ing customer lifetime values are unlikely to be overly affected by
forecasting error (Kumar and Shah 2009).
4We base the discount rate on the capital asset pricing model.
We assume a risk-free interest rate of 4.25% (Svensson 1994), a
5% market risk rate based on the Institute of German Accountants,
and a firm-specific beta of 1.45 based on Thomson Financial
Referral Programs and Customer Value / 51
$40:9 "             
1. Referral
program 0–1 .53 .50 1.00
2. Observed
value Euros 210.66 336.15 .02 1.00
3. Customer
lifetime value Euros 255.75 338.95 .01 1.00 1.00
4. Daily
contribution Euros
margin per day .33 .50 .04 .98 .98 1.00
5. Duration Days 751.05 119.48 –.17 .18 .21 .09 1.00
6. Age Years 42.90 17.47 –.20 .10 .09 .10 –.01 1.00
7. Female 0–1 .54 .50 .07 .01 .01 .01 .01 .05 1.00
8. Married 0–1 .39 .49 –.15 –.02 –.03 –.02 –.03 .43 .01 1.00
9. Single 0–1 .44 .50 .16 –.05 –.04 –.05 .01 –.65 –.10 –.70 1.00
10. Divorced 0–1 .10 .30 .00 .03 .03 .03 .02 .13 .06 –.26 –.29 1.00
11. Widowed 0–1 .05 .22 –.05 .11 .11 .10 .03 .36 .14 –.18 –.20 –.07 1.00
12. Acquired
January 2006 0–1 .03 .17 –.17 .07 .08 .03 .31 .02 .00 .01 –.03 –.00 .04 1.00
13. Acquired
February 2006 0–1 .03 .18 –.18 .02 .03 –.01 .27 .05 –.02 .04 –.04 –.00 .01 –.03 1.00
14. Acquired
March 2006 0–1 .06 .24 –.18 .04 .04 .01 .29 .07 –.01 .05 –.04 –.00 .00 –.04 –.05 1.00
15. Acquired
April 2006 0–1 .06 .23 .02 .04 .04 .02 .24 –.01 –.00 –.03 .02 .02 .01 –.04 –.05 –.06 1.00
16. Acquired
May 2006 0–1 .07 .26 .03 .03 .04 .01 .22 –.02 –.01 –.01 .02 –.01 –.02 –.05 –.05 –.07 –.07 1.00
17. Acquired
June 2006 0–1 .08 .28 –.01 .02 .02 .01 .14 .02 .02 –.01 –.01 .00 .04 –.05 –.06 –.08 –.07 –.08 1.00
18. Acquired
July 2006 0–1 .10 .30 .00 .01 .01 .01 .08 .02 –.00 .00 –.02 .03 –.00 –.06 –.06 –.09 –.08 –.09 –.10 1.00
19. Acquired
August 2006 0–1 .11 .31 .06 –.00 –.00 –.00 –.01 –.08 .00 –.06 .05 .02 –.02 –.06 –.06 –.09 –.08 –.09 –.10 –.12 1.00
20. Acquired
September 2006 0–1 .08 .27 .07 .00 .00 .01 –.08 –.06 .01 –.03 .04 –.00 –.02 –.05 –.06 –.08 –.07 –.08 –.09 –.10 –.10 1.00
21. Acquired
October 2006 0–1 .12 .33 .04 –.03 –.04 –.01 –.22 .01 .01 .02 –.00 –.01 –.01 –.07 –.07 –.09 –.09 –.10 –.11 –.13 –.13 –.11 1.00
22. Acquired
November 2006 0–1 .12 .33 .04 –.05 –.06 –.02 –.31 –.00 –.03 .01 –.01 –.01 –.00 –.07 –.07 –.09 –.09 –.10 –.11 –.13 –.13 –.11 –.14 1.00
23. Nonretail
customers 0–1 .12 .32 –.03 .27 .27 .27 –.00 .03 .00 .01 –.01 –.03 .00 .02 –.00 .01 .01 –.04 .00 –.01 –.02 –.02 –.03 .00
Notes: N = 9495. All correlations with absolute value of .02 or higher are significant at the 5% level. Differences in observed duration across customers are strongly affected by differences in the
month of acquisition. As a result, the zero-order correlations of duration with other variables that are also correlated with month of acquisition can be misleading. For example, the correla-
tion between duration and referral program changes from –.17 to .03 after we control for month of acquisition.
where i indexes the customer, DCM is daily contribution
margin over the observation period, RP is the binary variable
representing the referral program, X represents the control
variables, and the errors eihave a mean of zero and are
independent of the included covariates. We use ordinary least
squares to estimate the coefficients and compute Huber–
White heteroskedasticity-consistent standard errors (Breusch–
Pagan test, p< .001). The size of our sample implies that
we do not need to assume that the random shocks are nor-
mally distributed for statistical inference using t- and F-sta-
tistics (e.g., Wooldridge 2002, pp. 167–71).
To test whether difference in margin decreases the longer
the customer has been with the bank (H1b), we use a fixed-
effects specification estimated with ordinary least squares:
( ) ,4 1
i i k
ik i
= + + +
α β β ε
( ) ,5 2 3
it i it i it t it
= + + × + +α β β η ε
52 / Journal of Marketing, January 2011
where i indexes the customer; t indexes the year (t = 1, 2, 3);
DCMitis the daily contribution margin of customer i in year
t (i.e., the total contribution margin generated by customer i
in year t divided by the number of days the customer was
with the firm during year t); Titis the cumulative number of
days customer i had been with the bank by the end of year t;
htis a year-specific fixed effect; and the customer-specific
intercepts aiare not constrained to follow any specific dis-
tribution but capture all individual-specific, time-invariant
differences, including the effect of acquisition through the
referral program (RP) and that of the control variables X.
The errors eithave a mean of zero and are independent of
the covariates. The b3coefficient captures the proper inter-
action effect because the b1effect of RP is now captured
through the fixed effects. Again, we use the robust
Huber–White standard errors (Breusch–Pagan test, p< .001).
To assess the difference in retention between acquisition
modes, we use the Cox proportional hazard model. This
enables us to analyze right-censored duration data and to
exploit the fine-grained measurement of churn at the daily
level without imposing any restriction on how the average
churn rate evolves over time. Furthermore, the nonparamet-
ric baseline hazard makes the model robust to unobserved
heterogeneity in all but extreme cases (e.g., Meyer 1990).
We can represent the model to test H2a as follows:
where i indexes the customer, t indexes the amount of days
since the customer joined the bank, hi(t) is the hazard rate
for the customer’s defection, and a(t) is the log of the non-
parametric baseline hazard common across all customers.
To test whether the difference in churn propensity changes
over time (H2b), we extend Equation 6 with the interaction
between RPiand ln(t). The latter is also a test of whether
the RP effect meets the proportionality assumption (e.g.,
Blossfeld, Hamerle, and Mayer 1989), but we use it here to
test a hypothesis of substantive interest.
To test H3and assess the difference in customer value,
we again use the regression model in Equation 4, but now
with observed customer value as the dependent variable.
Theoretical claims can be subjected to empirical validation
or refutation only by comparing hypothesized effects with
actual data. As a result, the truth content of H3can be validly
tested using the observed customer value as the dependent
variable but not customer lifetime value, which itself is based
on predictions. Still, given the right censoring of our data
and the hypothesized erosion of the margin differential over
time, it is informative also to perform a similar analysis with
the six-year customer lifetime value as the dependent variable.
To calculate the predicted values entering the customer life-
time value metric, we use (1) predicted annual contribution
margins from a fixed-effects model, such as that specified
in Equation 5, but one in which we allow all parameters to
vary between referred and nonreferred customers, and (2)
predicted annual survival rates from a parametric Weibull
hazard model because the nonparametric baseline hazard of
the Cox model does not allow for forecasts.5
( ) ln[ ( )] ( ) ,6 1
h t t RP X
i i k ik
= + +
α β β
2006 2007 2008
.315 .331
Referred Nonreferred
0 200 400 600 800 1000
Notes: Customers were able to leave immediately after joining, but
only a handful did so. The earliest defection took place after
64 days, and only 27 customers left within the first year of
Is the Contribution Margin of Referred Customers
In accordance with H1a, referred customers are, on average,
4.5 cents per day more profitable than other customers
(Mann–Whitney test, p< .001). The difference is even larger
after we control for differences in customer demographics
and time of acquisition, variables on which the two groups
of customers are not perfectly matched. The first column of
Table 2 reports the coefficients of Equation 4, controlling
for age, sex, marital status, and month of acquisition.
Whereas the average contribution margin of nonreferred
customers in our sample is 30 cents per day, customers
acquired through the referral program have a margin that is
7.6 cents per day or 27.74 euros per year higher (p< .001),
an increase of approximately 25%. Among the covariates,
older age and being widowed are associated with a higher
contribution margin, whereas being married is associated
with a lower contribution margin. The pattern in the
monthly coefficients suggests that the bank was more suc-
cessful in acquiring profitable customers in some months
than in others. The R-square is low, indicating that impor-
tant elements other than acquisition method, acquisition
time, and demographics drive customer profitability.
Although the large unexplained variance depresses the
power of statistical tests and thus makes it more difficult to
reject the null hypothesis, H1a is strongly supported.
Does the Contribution Margin of Referred
Customers Remain Higher?
H1b predicts that the difference in contribution margin
erodes the longer a customer stays with the bank. The
results support this. Column 2 of Table 2 reports the coeffi-
cients of the fixed-effects model in Equation 5. There is a
significant, negative interaction between referral program
and the number of days the customer has been with the
bank. The difference in daily contribution margin between
referred and nonreferred customers decreases by 23.1 cents
per 1000 days, or 8.4 cents per year.
The individual-level fixed effects (intercepts) in the
model capture the expected daily contribution margin when
the included covariates equal zero (i.e., on the day of acqui-
sition in 2006). Regression of these 9495 fixed-effects esti-
mates on the referral program and control variables indi-
cates that a referred customer has an expected contribution
margin on the first day of joining the firm that is 19.8 cents
higher than that of a nonreferred customer with similar
demographics and time of acquisition.6This implies that
Referral Programs and Customer Value / 53
the expected contribution margin advantage of a referred
customer disappears after 857 days (.198/.000231), or
approximately 29 months after the customer joined the
Is the Retention of Referred Customers Higher?
To test whether the retention rate is higher for referred than
for nonreferred customers (H2a), we use the Cox propor-
tional hazard model specified in Equation 6. The results
reveal that the association between referral program and
churn (i.e., nonretention) is significantly negative and siz-
able. Use of only referral program as the explanatory
variable shows that at any point in time, customers acquired
through the referral program who are still with the firm are
approximately 13% less likely to defect than nonreferred
customers who are still with the firm. After we control for
differences in demographics and month of acquisition (see
Column 3 of Table 2), the effect size increases to approxi-
mately –18% [exp(–.198) – 1]. This multiplicative effect of
18% is relative to a baseline hazard that is small. As the sur-
vival curves in Figure 2 show, the probability of being an
active customer (i.e., a nonchurning customer) after 33
months is 82.0% for referred customers and 79.2% for non-
referred customers. Age is associated with a higher churn
rate, whereas the opposite holds for being widowed. There
is also a trend in the monthly coefficients, indicating that
customers acquired late in 2006 (especially in September
and later) exhibit more churn than those acquired earlier.
This trend is a cohort effect and not duration dependence,
which is captured in the nonparametric baseline hazard.
Does the Retention of Referred Customers
Remain Higher?
We also assess whether the difference in retention varies
over the customer’s lifetime (H2b). To do so, we extend the
Cox model with an interaction between the referral program
variable and the natural logarithm of the customer’s dura-
tion with the bank (see Column 4 of Table 2). The interac-
tion is not significant, and the model fit does not improve
significantly (p> .05). So, although there is an eroding dif-
ference between referred and nonreferred customers in con-
tribution margin, there is no such erosion for customer
Are Referred Customers More Valuable?
Using the observed customer value, we find that, on aver-
age, referred customers are 18 euros more valuable
(Mann–Whitney test, p< .001). After we control for demo-
graphics and month of acquisition, the difference increases
to 49 euros (Column 5 of Table 2; p< .001). A referred cus-
5In-sample parameter estimates from the Cox and Weibull mod-
els are nearly identical. The reason for using the Cox model in
testing the hypotheses is the absence of a restrictive parametric
assumption on the duration dependence.
6This difference in daily contribution margin of 19.8 cents is
close but not identical to the 18 cents difference between the two
groups of customers in 2006, shown in Figure 1. The small dispar-
ity between the two values occurs because the former controls for
differences in demographics and time of acquisition, whereas the
latter does not. Another reason for the disparity is that the former
is the difference on the day of acquisition, whereas the latter is the
difference on an average day in 2006.
7Note that in the model with the interaction term included, the
coefficient of referral program (.917, p> .05) is not the average
difference between referred and nonreferred customers anymore
but rather the difference between the two groups on the day of
acquisition (i.e., when the interaction variable Log(duration)
equals 0; see Irwin and McClelland 2001). Thus, the insignificant
coefficient of referral program in the model including the interac-
tion term does not invalidate the finding of a significant difference
in retention between the two groups posited in H2a.
54 / Journal of Marketing, January 2011
( (02> )
(02> 54:80);:054 /;84 )9,8<,+ ;9:53,8
54:80);:054 (8.04#03, ( #03, ;9:53,8 0-,:03,
(8.04 %(8>04. /;84 %(8>04. %(2;, %(2;,
Referral program .076*** a–.198** .917 49.157*** 39.906***
(.010) (.059) (1.479) (7.096) 7.152
Age .003*** .011** .011*** 1.879*** 1.626***
(.000) (.002) (.002) (.283) (.285)
Female –.009 –.034 –.034 –4.459 –3.376
(.010) (.056) (.056) (6.902) (6.958)
Sarried –.078* –.027 –.028 –52.798* –52.258*
(.033) (.166) (.166) (22.427) (22.563)
Single –.040 –.163 –.163 –27.306 –24.035
(.033) (.167) (.167) (22.573) (22.706)
Divorced –.016 –.176 –.177 –12.278 –7.656
(.037) (.183) (.183) (24.776) (24.933)
Widowed .111* –.470* –.470* 76.085* 87.249**
(.046) (.212) (.212) (31.128) (31.355)
Acquired January 2006 .172*** –1.828** –1.833*** 228.228*** 247.960***
(.039) (.201) (.201) (31.589) (31.666)
Acquired February 2006 .063* –1.365** –1.369*** 127.706*** 133.591***
(.031) (.160) (.159) (24.172) (24.411)
Acquired March 2006 .089** –1.155** –1.157*** 136.393*** 135.755***
(.026) (.126) (.126) (19.103) (19.280)
Acquired April 2006 .084** –1.215** –1.208*** 124.793*** 123.153***
(.027) (.140) (.140) (18.753) (18.895)
Acquired May 2006 .082** –1.529** –1.524*** 114.302*** 119.426***
(.025) (.150) (.150) (16.791) (16.909)
Acquired June 2006 .066** –1.016** –1.013*** 91.090*** 92.643***
(.022) (.122) (.122) (14.326) (14.475)
Acquired July 2006 .062** –1.026** –1.023*** 79.574*** 84.200***
(.021) (.122) (.122) (12.717) (12.839)
Acquired August 2006 .059** –.841** –.838*** 69.213*** 73.167***
(.020) (.119) (.119) (12.111) (12.233)
Acquired September 2006 .077** –.679** –.676*** 72.213*** 76.352***
(.022) (.126) (.126) (13.199) (13.335)
Acquired October 2006 .037 –.434** –.432*** 36.602* 39.391***
(.020) (.108) (.108) (11.133) (11.257)
Acquired November 2006 .021 –.217* –.215* 19.252 20.551
(.019) (.105) (.105) (10.497) (10.632)
Year 2007 (dummy) –1.306
Year 2008 (dummy) –2.259
Cumulative Days (in thousands) 3.513
Cumulative days (in thousands) ¥
referral program –.231**
Log(duration) ¥referral program –.176
Constant .154*** 66.250* 120.949***
(.040) (26.742) (26.937)
Observations 9495 28,353 9495 9495 9495 9495
.025 .350 .040 .040
Log-pseudo-likelihood –11,715.6 –11,715.4
*p< .05.
**p< .01.
***p< .001.
aCaptured by customer-specific fixed effects.
Notes: Robust standard errors are in parentheses.
tomer is approximately 25% more valuable to the bank than
a comparable nonreferred customer, within the observation
period. If we take into account the difference in acquisition
costs of approximately 20 euros, the difference in customer
value is nearly 35%. These results strongly support H3.
Because the margin differential of referred customers
erodes over time even though the loyalty differential does
not, the question arises whether referred customers remain
more valuable beyond the observation period. Repeating the
analysis for the six-year customer lifetime value, we show
that referred customers indeed remain more valuable. The
average customer lifetime value of referred customers is
approximately 6 euros higher than that of other customers
(Mann–Whitney test, p< .001). After we control for differ-
ences in customer demographics and time of acquisition,
the value differential is approximately 40 euros (Column 6
of Table 2; p< .001). Because the average customer lifetime
value of a nonreferred customer is 253 euros, a referred cus-
tomer is approximately 16% more valuable to the bank than
a comparable nonreferred customer over a horizon of six
years. If we take into account the difference in acquisition
costs of approximately 20 euros, the difference in customer
lifetime value is approximately 25%. This value differential
is quite considerable.
We also assess the extent to which the differences in
customer value are robust across various subsets of cus-
tomers. Table 3 reports the regression coefficients for the
referral program in models of customer value, with the
same controls as in the previous analysis in Columns 5 and
6 of Table 2. Row 1 of Table 3 shows that the results for the
retail customer segment are nearly identical to those for the
entire sample. This is not surprising, because retail cus-
tomers make up approximately 90% of our overall sample.
More informative is that the difference in customer value
also exists in the nonretail segment (Row 2 of Table 3).
Rows 3 and 4 of Table 3 show that the positive referral
differential exists among high-margin customers, defined as
those in the top decile based on margin, but not low-margin
customers, defined as those in the bottom decile based on
margin.8The remaining rows in Table 3 show that sizable
value differentials between referred and nonreferred cus-
tomers exist among both men and women and among all
age ranges, except for those over the age of 55. Overall, the
acquisition through a referral program is associated with
higher customer value for the majority of customer types,
but not for all. These results suggest that using referral pro-
grams might not be beneficial in all customer segments, an
idea we develop further in the “Discussion” section.
Robustness Checks
Table 4 shows that the hypothesis tests are robust to includ-
ing retail versus nonretail segment membership as an addi-
tional control variable and allowing the effect of the referral
program to vary as a function of age, sex, marital status, and
retail segment membership. Given the results of Table 3, we
Referral Programs and Customer Value / 55
also allowed for a nonlinear effect of age.9We mean-center
all variables that interact with referral program, so its coef-
ficient still reflects the main effect. This coefficient keeps
its sign and significance in each model, so the hypotheses
remain supported. The coefficients are larger than in Table
2, in which we did not control for segment membership and
nonlinear age effects, indicating that our main analysis pro-
vides rather conservative estimates of the referral effects.
As a second robustness check, we repeated the analyses
presented in Table 2 for the sample including all outliers.
The direction and significance of the referral effect remained
the same, but the size of several effects increased. The dif-
ference in daily contribution margin increased from 7.6
cents to 16 cents per day, the margin erosion increased from
23.1 cents to 45.4 cents per thousand days, the churn hazard
reduction remained at 20%, and the difference in customer
lifetime value increased from 40 euros to 66 euros. These
8Low-margin customers and high-margin customers are found
in both the retail and the nonretail segments.
9Because some readers may be interested in how the effect of
referral program is moderated by covariates in the time-varying
contribution model, we estimate the latter using a random coeffi-
cients specification rather than the fixed-effects specification used
in Table 2.
)9,8<,+ ;9:53,8
;9:53,8 0-,:03,
%(2;, %(2;,
!5);9: !5);9: 5-
":(4+(8+ ":(4+(8+ !,-,88,+
88589 88589 ;9:53,89
Retail customers 48.620*** 39.082*** 8384
(6.574) (6.633) (4473)
Nonretail customers 77.309** 69.803* 1111
(29.855) (30.023) (536)
High margin customers 80.421** 69.669* 950
(27.768) (28.004) (533)
Low margin customers –1.146 –13.212*** 962
(1.581) (2.087) (247)
Male customers 51.679*** 42.305*** 4371
(10.600) (10.669) (2150)
Female customers 47.437*** 38.274*** 5124
(9.604) (9.690) (2859)
<25 years of age 35.662** 17.701 1808
(12.914) (12.945) (1242)
26–35 years of age 101.975*** 85.280*** 2170
(14.908) (14.822) (1298)
36–45 years of age 66.148*** 57.401** 1621
(17.534) (17.707) (835)
46–55 years of age 62.763** 56.834** 1437
(19.671) (19.827) (617)
56–65 years of age 9.433 5.122 1153
(21.189) (21.195) (481)
>65 years of age –1.577 –8.589 1306
(21.421) (21.409) (536)
*p< .05.
**p< .01.
***p< .001.
Notes: Each row displays the coefficient of referral program in models
with the same control variables as in Table 2 but estimated
for specific segments.
56 / Journal of Marketing, January 2011
( (02> )
(02> 54:80);:054 /;84 )9,8<,+ ;9:53,8
54:80);:054 (8.04#03, ( #03, ;9:53,8 0-,:03,
(8.04 %(8>04. /;84 %(8>04. %(2;, %(2;,
Referral program .133*** .228*** –.270** .791 88.413*** 79.695***
(.017) (.062) (.084) (1.479) (10.870) (10.949)
Age (mean centered)a.310*** .529** 1.421*** 1.421*** 199.247*** 164.893***
(.060) (.191) (.361) (.362) (41.286) (41.647)
Age2(mean centered)a–.004 –.779 –.014 –.014 –2.276 –2.190
(.003) (.733) (.014) (.014) (1.834) (1.848)
Female (mean centered) –.008 .035 .017 .016 –3.822 –2.828
(.013) (.041) (.076) (.076) (9.393) (9.487)
Married (mean centered) .050 .828*** –.445* –.445* 34.372 38.493
(.039) (.126) (.201) (.202) (27.838) (28.187)
Single (mean centered) .080* .933*** –.472* –.473* 52.010 59.890*
(.040) (.128) (.211) (.211) (27.774) (28.119)
Divorced (mean centered) .111* .916*** –.589* –.590* 77.122* 86.686**
(.044) (.139) (.231) (.231) (31.289) (31.696)
Widowed (mean centered) .340*** 1.104*** –1.011*** –1.012*** 236.446*** 254.709***
(.058) (.154) (.272) (.273) (40.821) (41.259)
Nonretail segment (mean centered) .440*** .551*** –.263* –.263* 310.169*** 311.262***
(.031) (.060) (.124) (.124) (21.972) (22.152)
Age ¥referral programa,b .151 .081 –.531 –.532 99.676 117.447*
(.086) (.255) (.514) (.514) (57.745) (58.183)
Age2¥referral programa, b –.015*** –.311 .022 .022 –10.212*** –10.392***
(.004) (1.018) (.020) (.020) (2.539) (2.556)
Female ¥referral programb–.005 –.019 –.104 –.104 –2.980 –2.982
(.020) (.056) (.112) (.112) (13.210) (13.320)
Married ¥referral programb–.162* –.974*** .953** .951** –108.801* –114.852*
(.068) (.174) (.362) (.362) (45.824) (46.049)
Single ¥referral programb–.097 –.972*** .743* .743* –62.859 –70.679
(.068) (.175) (.366) (.366) (45.988) (46.207)
Divorced ¥referral programb–.147* –.982*** .916* .917* –103.598* –112.706*
(.074) (.190) (.394) (.394) (50.159) (50.458)
Widowed ¥referral programb–.270** –1.181*** 1.246** 1.248** –197.134** –211.207**
(.091) (.222) (.456) (.456) (60.695) (61.046)
Nonretail ¥referral programb–.025 –.084 .014 .013 –49.754 –49.193
(.046) (.084) (.188) (.188) (30.805) (31.001)
Year 2007 (dummy) –1.487***
Year 2008 (dummy) –2.562***
Cumulative days (in thousands) 4.004***
Cumulative days (in thousands) ¥–.220*
referral program (.101)
Log(duration) ¥referral program –.167
Constant .211*** .244*** 100.219*** 147.265***
(.016) (.056) (9.809) (9.921)
Observations 9495 28,353 9495 9495 9495 9495
.107 .099c.123 .122
Log-pseudo-likelihood –11,705.8 –11,705.6
*p< .05.
**p< .01.
***p< .001.
aWe divide age by 100 for better readability.
bThe first variable in this interaction is mean centered.
cBecause the model is a random coefficients model estimated with residual maximum likelihood, this value is a pseudo-R-square calculated as
the squared correlation between predicted and actual values.
Notes: Robust standard errors are in parentheses. All models include dummies for month of acquisition.
results suggest that our main analysis is rather conservative
with regard to the size of the referral differentials.
Although hazard analysis properly accounts for right
censoring, managers are also interested in simply knowing
who is likely to have remained with the firm within a cer-
tain time frame. Therefore, we also assessed the relation-
ship between referral and the probability of still being with
the bank 21 months after acquisition. This time span is the
longest duration observable without right censoring for
every customer, including those who were acquired last, at
the end of December 2006. Using logistic regression and
controlling for demographics and month of acquisition, we
find that referred customers are approximately 22% less
likely to leave the firm within the first 21 months (p< .01).
Computing the customer lifetime value over three,
rather than six, years after acquisition and repeating the
analysis by controlling for demographics and time of acqui-
sition yields a value differential between referred and non-
referred customers of 52 euros (p< .001) rather than 40
euros. Both the size and the statistical significance of the
latter value are rather robust to reestimating the model on
smaller random samples of 90% (39 euros, p< .001), 75%
(42 euros, p< .001), 50% (48 euros, p< .001), and 25% (36
euros, p< .01). We also computed the expected value differ-
ential if there were no difference in retention between referred
and nonreferred customers. The differential in six-year cus-
tomer lifetime value decreased from 40 euros to 33 euros.
Finally, we extended the model of margin dynamics and
allowed the effect of time and its interaction with referral to
vary as a function of observed customer demographics,
retail versus nonretail status, and normally distributed unob-
served heterogeneity. This extended random coefficients
model did not fit the data better: The squared correlation
between observed and predicted values (pseudo-R2)
increased by only .1%, and the Bayesian information crite-
rion even decreased. Most important, the coefficients of
central interest and the statistical inference were not
affected: Customers acquired through referral had a sizable
initial margin advantage that eroded to zero after approxi-
mately 1000 days.
Key Findings
Evidence of the economic value of stimulated WOM and of
the customers it helps acquire has been sorely lacking. Our
study addresses this gap in the context of referral programs
and documents the attractiveness of customers acquired
through such a program: Contribution margin, retention, and
customer value all were significantly and sizably higher for
referred customers. In short, referred customers are more
valuable in both the short and the long run. Yet we also find
that the effect is not uniform across all types of customers
and that the referral program was less beneficial when used
to acquire older customers or low-margin customers.
In our application, the value of referred customers in the
six years after acquisition was 40 euros (or 16%) higher
than that of nonreferred customers with similar demograph-
Referral Programs and Customer Value / 57
ics and time of acquisition. Considering the initial reward of
25 euros given to the referrer as an investment, this implies
a return on investment of approximately 60% over a six-
year span. This is a conservative estimate because it does not
take into account that the total acquisition costs of referred
customers are approximately 20 euros lower than those of
other customers.
Implications for Practice
Several scholars have expressed cautious skepticism about
the value of “viral-for-hire” and other stimulated WOM
(e.g., Trusov, Bucklin, and Pauwels 2009; Van den Bulte
2010). Doubts about the benefits of stimulated WOM have
long frustrated managers facing demands to increase their
marketing return on investment. Our findings are important
news for practitioners considering deploying customer
referral programs in their own firm. Assuaging prior skepti-
cism, we document a positive value differential, in both the
short run and the long run, between customers acquired
through a referral program and other customers. Impor-
tantly, this value differential is larger than the referral fee.
Thus, referral programs can indeed pay off.
The positive differential indicates that abuse by oppor-
tunistic customers and other harmful side effects of referral
programs are much less important than their benefits. The
referral program we analyzed was especially prone to
exploitation because no conditions, such as minimum stay
or assets, applied to the newly acquired customer. Finding a
positive value differential of referred customers in this set-
ting is especially compelling. Moving beyond referral pro-
grams specifically, our study indicates that a stronger focus
on stimulated rather than organic WOM is worth consider-
ing (Godes and Mayzlin 2009).
However, our results also suggest that firms should
think carefully about what prospects to target with referral
programs and how big of a referral fee to provide. For the
program we analyzed, we found that the customer value dif-
ferential is much larger in some segments than in others.
People under the age of 55 and high-margin customers are
more attractive to acquire through a referral program. It is
not necessarily a coincidence that these customers also tend
to be more profitable for banks (and many other consumer
marketers). To the extent that the value differential stems
from better matching and social enrichment, as sociological
theory suggests and as employee referral programs have
documented, referral programs do not “create” higher-value
customers by transforming unattractive prospects into
attractive customers. Rather, they help firms selectively
acquire more valuable prospects and retain them longer at
lower cost. Thus, instead of the currently practiced “all-in”
approach, firms should design and target referral programs
such that attractive customers are more likely to be enticed.
Managers must also make their customers aware of their
referral programs. Bank of America, for example, commu-
nicates its referral program on all its automated teller
machines throughout the United States. Connecting referral
programs with online activities might help further increase
their reach beyond existing customers’ networks of strong
ties and face-to-face interactions. Managers must also make
it convenient for prospects to actually become a customer. A
possible application is to partner with online communities
and make it easy for people to start a relationship with the
firm online, immediately after they receive a referral from
an existing customer in the same community. Our results
suggest that such awareness and facilitation efforts should
be targeted selectively to customers who offer the highest
value differential.
The referral fee is another issue that requires attention
when designing a referral program. Many programs offer
the same reward to each referrer (Kumar, Petersen, and
Leone 2010). Yet, as we show, the value of referred cus-
tomers can vary widely even for one company. Thus, firms
may benefit from offering rewards based on the value of the
referred customer. However, the question then becomes
how to do this without adding too much complexity to the
program. There may be a simple answer: A standard
homophily argument suggests that valuable referrers are more
likely to generate valuable referrals. Thus, firms may want to
make the referral fee a function of the value of the referrer.
A different approach to take advantage of the referral
effect would be to try to generate conditions in which non-
referred customers become subject to the same mechanisms
that make referred customers more valuable. To the extent
that the differences we have documented stem from better
matching, from social enrichment, or from other mecha-
nisms that firms can actively foster among all customers,
firms may be able to dramatically “scale up” the beneficial
referral effect beyond dyads of referring and referred cus-
tomers. For example, pharmaceutical companies increas-
ingly involve local opinion leaders in their speaker pro-
grams and other medical education efforts. They do so to
capitalize on these physicians’ relevance and credibility
with practicing physicians.
Firms in the same industry often reward referrers with
the same amount. For example, most German banks offer
25 euros for a referral, as does the bank we studied. Our
results indicate that managers set the referral fee rather low,
allowing the firm to reap attractive returns from its pro-
gram. Offering higher rewards might lead to even more cus-
tomer acquisitions while still providing positive returns on
investment. Firms should calculate the reward considering
their specific program and the customers it attracts instead
of merely following their competitors.
Further Research
Our study focuses on referred and nonreferred customers of
one particular bank. Although its customer base and referral
program have no unusual characteristics, replications would
nonetheless be welcome. Such studies require rich, firm-
specific data on a large set of customers, with individual
profitability observed over a long period. Therefore, we
expect replications and extensions to come from other
single-firm studies such as ours and those of Godes and
Mayzlin (2009), Haenlein (2010), Iyengar, Van den Bulte, and
Valente (2011), and Nitzan and Libai (2010). Because the
mechanisms of better matching and social enrichment are
likely to be more important for complex products with impor-
tant experience attributes, rather than simple products with
58 / Journal of Marketing, January 2011
mostly search attributes (e.g., Coverdill 1998; Kornish and
Li 2010; Rees 1966), studies of multiple products with vary-
ing levels of complexity would be especially informative.
It is likely that the quality of the matches with the firm
deteriorates as existing customers refer more new cus-
tomers. It would be of practical interest to know at what rate
the quality of referrals decreases and at what point it tends
not to justify the cost of acquisition anymore. It may also be
useful to know if the motivation of the referrer changes
depending on the reward and whether the size of the reward
affects the quality of the referred customer.
Several of the implications for practice point to the
benefits of better understanding the drivers of the value dif-
ferential we documented. Although our results are consis-
tent with the better matching and social enrichment mecha-
nisms we used to develop our hypotheses, our analysis
focused on the consequences for contribution margin, reten-
tion, and customer value rather than on the intervening
mechanisms. Research aimed at more directly parsing out
the mechanisms is likely to require information about actual
dyads of referring and referred customers. This would
enable researchers to test, for example, the social enrich-
ment argument by matching the referred customer with the
respective referrer and analyzing the dependence of their
retention. Additional survey data may help document differ-
ences in product knowledge over time and shed light on the
existence of an informational advantage eroding over time.
Having matched dyad-level data on both referring and
referred customers would also make it possible to check
whether referral dyads exhibit homophily and whether the
value of referred customers varies systematically with that
of their referrer (Haenlein 2010; Nitzan and Libai 2010).
This would yield valuable insights for the design of individ-
ual rewards instead of the currently practiced “one-size-fits-
all” approach.
This study provides the first assessment of economically rele-
vant differences between customers acquired through a refer-
ral program and customers acquired through other methods.
It documents sizable differences in contribution margin,
retention, and customer value; analyzes whether these differ-
ences erode or persist over time; and investigates differences
across customer segments. The finding that, on average,
referred customers are more valuable than other customers
provides the first direct evidence of the financial attractive-
ness of referral programs and also offers much-needed evi-
dence of the financial appeal of stimulated WOM in general.
Improvements in the targeting, design, and implementa-
tion of such programs will require a better understanding
of the drivers of the value differential. The dyadic inter-
dependence in the behavior of the referrer and the referred
customer deserves special attention in further research
because it is likely to prove highly relevant to both better
theoretical understanding and more effective program
Referral Programs and Customer Value / 59
Lewis, Michael (2006), “Customer Acquisition Promotions and
Customer Asset Value,Journal of Marketing Research, 43
(May), 195–203.
Meyer, Bruce D. (1990), “Unemployment Insurance and Unem-
ployment Spells,” Econometrica, 58 (July), 757–82.
Montgomery, James D. (1991). “Social Networks and Labor-Market
Outcomes: Toward an Economic Analysis,” American Eco-
nomic Review, 81 (December), 1408–1418.
Mummert, Hallie (2000), “The Year’s Best Bells & Whistles,Tar-
get Marketing, 23 (11), 3–5.
Neckerman, Kathryn M. and Roberto M. Fernandez (2003),
“Keeping a Job: Network Hiring and Turnover in a Retail
Bank,” Research in the Sociology of Organizations, 20,
Nitzan, Irit and Barak Libai (2010), “Social Effects on Customer
Retention,” Report No. 10-107. Cambridge, MA: Marketing
Science Institute.
Parasuraman, A., Valarie A. Zeithaml, and Leonard L. Berry
(1985), “A Conceptual Model of Service Quality and Its Impli-
cations for Future Research, Journal of Marketing, 49 (Fall),
Rees, Albert (1966), “Information Networks in Labor Markets,”
American Economic Review, 56 (March), 559–66.
Ryu, Gangseog and Lawrence Feick (2007), “A Penny for Your
Thoughts: Referral Reward Programs and Referral Likeli-
hood,” Journal of Marketing, 71 (January), 84–94.
Svensson, Lars E. (1994), “Estimating and Interpreting Forward
Interest Rates: Sweden 1992-1994,” NBER Working Paper
No. 4871. Cambridge, MA: National Bureau of Economic
Thaler, Richard H. and Cass R. Sunstein (2008), Nudge: Improv-
ing Decisions About Health, Wealth, and Happiness. New
Haven, CT: Yale University Press.
Trusov, Michael, Randolph E. Bucklin, and Koen Pauwels (2009),
“Effects of Word-of-Mouth Versus Traditional Marketing:
Findings from an Internet Social Networking Site,” Journal of
Marketing, 73 (September), 90–102.
Van den Bulte, Christophe (2010), “Opportunities and Challenges
in Studying Customer Networks,” in The Connected Customer:
The Changing Nature of Consumer and Business Markets, S.
Wuyts, M.G. Dekimpe, E. Gijsbrechts, and R. Pieters, eds.
London: Routledge, 7–35.
——— and Stefan Wuyts (2007), Social Networks and Marketing.
Cambridge, MA: Marketing Science Institute.
Villanueva, Julian, Shijin Yoo, and Dominique M. Hanssens
(2008), “The Impact of Marketing-Induced Versus Word-of-
Mouth Customer Acquisition on Customer Equity Growth,”
Journal of Marketing Research, 45 (February), 48–59.
Wiesel, Thorsten, Bernd Skiera, and Julian Villanueva (2008),
“Customer Equity: An Integral Part of Financial Reporting,”
Journal of Marketing, 72 (March), 1–14.
Wirtz, Jochen and Patricia Chew (2002), “The Effects of Incen-
tives, Deal Proneness, Satisfaction and Tie Strength on Word-
of-Mouth Behaviour,International Journal of Service Indus-
try Management, 13 (2), 141–62.
Wooldridge, Jeffrey M. (2002), Introductory Econometrics: A
Modern Approach, 2d ed. Cincinnati: South-Western.
Yim, Chi Kin, David K. Tse, and Kimmy Wa Chan (2008),
“Strengthening Customer Loyalty Through Intimacy and Pas-
sion: Roles of Customer–Firm Affection and Customer–Staff
Relationships in Services,” Journal of Marketing Research, 45
(December), 741–56. (2008), “Confirmed: TiVo Rewards Killed,”
(accessed June 28, 2010), [available at http://www.zatznotfunny.
Belsley, David A., Edwin Kuh, and Roy E. Welsch (1980), Regres-
sion Diagnostics: Identifying Influential Data and Sources of
Collinearity. New York: John Wiley & Sons.
Biggart, Nicole W. (1989), Charismatic Capitalism: Direct Selling
Organizations in America. Chicago: University of Chicago Press.
Biyalogorsky, Eyal, Eitan Gerstner, and Barak Libai (2001), “Cus-
tomer Referral Management: Optimal Reward Programs,”
Marketing Science, 20 (1), 82–95.
Blossfeld, Hans-Peter, Alfred Hamerle, and Karl Ulrich Mayer
(1989), Event History Analysis: Statistical Theory and Applica-
tion in the Social Sciences. Hillsdale, NJ: Lawrence Erlbaum
Bolton, Patrick, Xavier Freixas, and Joel Shapiro (2007), “Con-
flicts of Interest, Information Provision, and Competition in the
Financial Services Industry,” Journal of Financial Economics,
85 (2), 297–330.
Castilla, Emilio J. (2005), “Social Networks and Employee Perfor-
mance in a Call Center,American Journal of Sociology, 110
(5), 1243–83.
Choi, Jeonghye (2009), “Social Influence from Existing to New
Customers,” working paper, the Wharton School, University of
Coverdill, James E. (1998), “Personal Contacts and Post-Hire Out-
comes: Theoretical and Empirical Notes on the Significance of
Matching Methods,” Research in Social Stratification and
Mobility, 16, 247–69.
Edwards, Helen and Derek Day (2005), Creating Passion Brands:
Why It’s Time to Lead from the Heart. London: Kogan Page.
Fernandez, Roberto M., Emilio J. Castilla, and Paul Moore (2000),
“Social Capital at Work: Networks and Employment at a Phone
Center,American Journal of Sociology, 105 (5), 1288–1356.
Fleming, John H., Curt Coffman, and James K. Harter (2005),
“Manage Your Human Sigma,Harvard Business Review, 83
(7–8), 106–114.
Godes, David and Dina Mayzlin (2009), “Firm-Created Word-of-
Mouth Communication: Evidence from a Field Test, Market-
ing Science, 28 (4), 721–39.
Gouldner, Alvin W. (1960), “The Norm of Reciprocity: A Preliminary
Statement,” American Sociological Review, 25 (April), 161–78.
Gupta, Sunil and Valarie Zeithaml (2006), “Customer Metrics and
Their Impact on Financial Performance,Marketing Science,
25 (6), 718–39.
Guseva, Alya (2008), Into the Red: The Birth of the Credit Card
Market in Postcommunist Russia. Palo Alto, CA: Stanford Uni-
versity Press.
Haenlein, Michael (2010), “A Social Network Analysis of Customer-
Level Revenue Distribution,Marketing Letters, forthcoming.
Helm, Sabrina (2003), “Calculating the Value of Customers’
Referrals,” Managing Service Quality, 13 (2), 124–33.
Irwin, Julie R. and Gary H. McClelland (2001), “Misleading
Heuristics and Moderated Multiple Regression Models,” Jour-
nal of Marketing Research, 38 (February), 100–109.
Iyengar, Raghuram, Christophe Van den Bulte, and Thomas W.
Valente (2011), “Opinion Leadership and Social Contagion in
New Product Diffusion,” Marketing Science, forthcoming.
Kornish, Laura J. and Quiping Li (2010), “Optimal Referral
Bonuses with Asymmetric Information: Firm-Offered and
Interpersonal Incentives,Marketing Science, 29 (1), 108–121.
Kumar, V., J. Andrew Petersen, and Robert P. Leone (2007), “How
Valuable Is Word of Mouth?” Harvard Business Review, 85
(10), 139–46.
———, ———, and ——— (2010), “Driving Profitability by
Encouraging Customer Referrals: Who, When, and How,
Journal of Marketing, 74 (September), 1–17.
——— and Denish Shah (2009), “Expanding the Role of Market-
ing: From Customer Equity to Market Capitalization,” Journal
of Marketing, 73 (November), 119–36.
... Marketing researchers denote this term as "word of mouth," We have studied the impact of referrals for decade which are interpersonal communication with far greater influence over consumer attitudes and behavior than either conventional advertising or neutral print sources (Buttle, 1998), and the value of personal referral constitute a significant portion of a customer's value (Kumar et al., 2007). It is studiedin10,000 accounts in a large German bank over a period of three years, and found that customers obtained through referrals are both more loyal and more valuable than other customers (Schmitt et al., 2011). ...
... Customer referral programs (CRPs)-defined as deliberately initiated, actively managed, continuously controlled firm activities aimed to stimulate positive word of mouth among existing customer bases have received increasing attention from marketing researchers and practitioners (Schmitt et al., 2011) . Their objectives are to use the social connections between existing customers and noncustomers to attract the latter to the firm. ...
Full-text available
This study sought to identify the relative roles of Customer Relationship Management (CRM), Integrated Marketing Communication (IMC), Customer Referral Campaign (CRC) and Perceived Experience (PE) in the formation of the Purchase Intentions (PI) of Bangladeshi consumers. A total of 111respondents were literally categorized into apartment owners and prospective buyers, were eventually surveyed with the structured and self-administrated questionnaires followed by few in-depth interviews. The results of the study suggest that the three variables; CRM, IMC, and PE appear to have a significant direct impact on the formation of the Purchase Intentions (PI) of the apartment in Dhaka city. Conversely, CRC found to have no effect on the Purchase Intention (PI).
... . Programas de referência(Referral Programs) São formas de WOM incentivadas pela oferta de incentivos e recompensas aos clientes existentes que indicam e trazem novos clientes, sendo uma forma atrativa de adquirir novos(Schmitt et al., 2011).3. Recomendações sociais (Social Recommendations)Referem-se à utilização das redes sociais como meio de aceder a recomendações ou criar novas sobre o que comprar, ler, comer e fazer(Shadkam & O'Hara, 2013). ...
Full-text available
In a context where social networks and online shopping are part of the everyday life of the new consumer, a new digital marketing trend is emerging, leveraged very much in part by the COVID-19 pandemic, which has accelerated and completely changed the e-commerce landscape, perhaps more than at any other time in history: social commerce. Sales are no longer made solely and exclusively through online stores or e-commerce platforms, but are now possible directly from social networks, which have increasingly integrated and native shopping features and tools that leverage social commerce. In order to positively impact the new digital consumer and increase online sales through its social networks, MO, a textile fashion brand that belongs to Sonae Group, set the challenge of understanding how it would be possible to start using social commerce in its business. To do so, it was necessary to better understand this phenomenon, the emerging trends in markets where it is already a reality, as well as the features that the big social commerce players contain, which this study sought to answer. Additionally, good practices that could be applicable to the Portuguese reality and to their own business were identified. The present study, which used a qualitative methodology by using a narrative literature review, content analysis of articles, web pages, reports and social networks of the inspirational international competition and direct observation of MO and its competitors (Zippy, H&M and Lefties) content, resulted in a proposal to boost the sales of MO with social commerce trends and best practices analyzed.
... Customers who are active in carrying out this strategy are known as referred customers. Referral strategy has three main characteristics, which are intentionally initiated, actively managed, and continuously controlled by the company [9]. Referral strategy is different from multilevel marketing. ...
Full-text available
A marketing strategy is an attempt to campaign for a product so that information about the product becomes viral. Advertising strategy began to be combined by involving customers to campaign for the company’s products. This marketing strategy is known as the referral marketing strategy. Viral information in a community can be analogous to the pandemic condition of a disease. This study aims to examine the relationship between the epidemiological model and the marketing model with a referral strategy. This research is a literature study with the results of a marketing model design with a referral strategy. Based on the research, there are 4 groups in marketing modeling with a referral strategy, namely Unaware, Potential Broadcaster, Broadcaster, and Inert. The epidemiological model that is suitable for this condition is the SEIR model. Adjustments to the SEIR Model need to be made because the dynamics of the Unaware can go directly to the Inert because of the trust factor in marketing information. The simulation results show that the parameters that have an important role in making information viral are the customer network and the incubation time of the information becoming viral.
... Conceptualizing the three aspects of CEB as customers' indirect contributions to firms, Pansari and Kumar, (2017) identify customer referrals as incentivized referrals that help attract customers. Because referred customers are more profitable (Schmitt et al., 2011), referrals contribute indirectly to firm performance. Customer influence refers to their impact, especially through social media. ...
Full-text available
This study explores how customer-based corporate reputation (CBR) influences customer engagement behaviors (CEB). It incorporates customer identification and brand love as mediators between CBR and CEB, and the industry type as a moderator, to investigate the direct and indirect effects of CBR on CEB through a moderated mediation analysis. The hypotheses were tested through PLS-SEM and PROCESS. Results confirmed the mediation effects of customer identification and brand love between CBR and CEB, and the positive direct impact of CBR on CEB. Furthermore, this study found that five aspects of CBR have different impacts on four dimensions of CEB. Lastly, the CBR–customer identification–CEB mechanism is stronger in the service than in the product industry, whereas there is no difference in the mediation mechanism of brand love between the product and service.
... In the user-SVA relationship, we link these gifts with the role that users play when they refer the SVA brand. Users that refer the brand may not make the most purchases, but they are more profitable than other, similarly-profiled users (Schmitt et al., 2011). In this way, when users perceive relational cohesion with their SVA, they make positive referrals of the SVA brand and help it to reach other users who would not be attracted by traditional marketing channels, thus contributing to overall user engagement : ...
Full-text available
Purpose This study aims to examine users’ affective relationships with smart voice assistants (SVAs) and aims to analyze how these relationships explain user engagement behaviors toward the brands of SVAs. Drawing on relational cohesion theory, it proposes that cohesion between users and SVAs influences brand engagement behaviors, that is, continuing purchasing other products of the brand, providing knowledge to the brand and referring the brand. Design/methodology/approach Data from a survey of 717 US regular SVA users confirm the validity of the measurement scales and provide the input for the covariance-based structural equation modeling. Findings The results demonstrate that frequent user-SVA interactions evoke positive emotions, which encourage cohesive relationships. Pleasured-satisfaction and interest emerge as strong emotions. Moreover, relational cohesion between users and SVAs promotes engagement with the brand of the assistant. Originality/value This paper applies an interpersonal approach in a context that, to date, has been examined from a predominantly technological perspective. It shows that users develop positive emotions toward smart technologies through their interactions, and establishes the importance of building affective relationships. To the best of the authors’ knowledge, this is the first study to analyze cohesion between users and smart technologies and to examine the effect of this cohesion on user engagement with the brand.
... The popularity and reach of social media sites have inspired many businesses to use social networking sites to promote their product and services. Social media has evolved as an effective tool to promote brands and connecting with the target market (Schmitt et al., 2011). The contemporary marketing approach has changed as social media sites provides tools to have close interaction with consumers and creates new opportunities for the marketer to promote their brands (Holzner, 2008). ...
... Receiving rewards also empower consumers to share their positive experiences online about the brand (e.g. Ba et al., 2001;Lou et al., 2013), encourages them to recommend this brand to others (Schmitt et al., 2011), and increase consumers' attitudes and behaviors, especially in social media (e.g. Gerrath and Usrey, 2020; Rohm et al., 2013). ...
Full-text available
Purpose This study aims to examine the relationships between the online communities’ characteristics and resilience to negative information (RNI) mediated by both brand knowledge and brand involvement. Design/methodology/approach According to stimulus-organism-response (S-O-R) framework, this research postulates that information quality, rewards and virtual interactivity drive RNI directly and indirectly through brand knowledge and/ or brand involvement. A survey of 326 Facebook pages followers was conducted, representing followers of fashion clothing brands in social media platforms in Egypt. We have used AMOS to check the constructs’ validity and reliability, as well as the Hayes’s PROCESS macro to test the mediation. Findings The findings show that information quality, rewards and virtual interactivity are the respective drivers of brand knowledge and brand involvement; the brand knowledge and brand involvement help explain why consumers are resilient to negative information of specific brands; and the drivers of brand knowledge and/ or brand involvement differ in consumers who tend to ignore negative information they receive about the brand. Practical implications Outcomes of the research recommend that executives should identify the outstanding determinants for improving resilient consumers to negative information through creating the highest possible brand knowledge and involvement between the consumers and brands. Originality/value Little attention has been paid to examine the RNI and linking it with brand knowledge and brand involvement in online communities’ context, thus, the current research is conducted.
Full-text available
The study focused on customer value and business strategy of manufacturing firms in Enugu and Anambra States, Nigeria. The study was conducted in ten purposely selected manufacturing firms in Enugu and Anambra states. The specific objectives were to: examine the effect of average purchase value on product prices of manufacturing firms and ascertain the effect of purchase frequency on new competitor entrants into the market of manufacturing firms. The study adopted the survey research design. A sample size of 365 was got using Freund and Williams'(1966) sample size determination statistical formula and same number was administered to members of the board, management and senior staff of the selected firms using Bowley's (1997) proportional allocation formula. 311 copies of the questionnaire were returned. Data were collected from primary and secondary sources. The questionnaire was structured on a five point Likert scale. The face and content validity of the research instrument was evaluated by experts in the industry. The hypotheses were tested with Z-statistical tool. Findings revealed that average purchase value had positive effect on product prices of manufacturing firms in Anambra and Enugu States, Nigeria Z (97, n = 311) =5.1238, p > 0.05, purchase frequency had positive effect on new competitor entrants into the market of manufacturing firms in Anambra and Enugu States, Nigeria Z (97, n = 311) =5.0478, p > 0.05. The study concluded that manufacturing firms should broaden their own strategy to cover the same target and prevent new companies from getting a foothold.
Full-text available
Seeding is the introduction of a new product to a portion of consumers (seeds) before it is widely introduced to the market, and is one of the most important drivers in the design of marketing campaigns and effective word-of-mouth advertising. This research aims to identify different dimensions of seeding through meta-synthesis. For this purpose, 68 articles that had studied the subject of seeding were selected and coded on the concepts extracted from them. In the analysis of chosen researches, from 189 primary codes, 90 indicators (selective code), 18 components (sub-dimension), and five dimensions (main component) were identified. For the first time in this paper, different dimensions of seeding were investigated by the meta-synthesis method. According to the reviewed articles, the concepts extracted from the literature are: 1) Factors affecting seeding with 33 indicators and seven components that are target community characteristics, target community homogeneity, message nature, environmental conditions, internal conditions, marketing tools, and implementation strategies; 2) Characteristics of seeding target market with 21 indicators and three components that are consumer characteristics, customer nature and characteristics of individuals; 3) Communication and networking of seeding with 15 indicators and three components that are a big picture of seeding, position of seeding and nature of social network; 4) Seeding and growth and development with 12 indicators and three components that are value creation, market growth and development and product growth and development; And finally 5) Seeding and marketing with nine indicators and two components that are marketing tools strengthen and customer experience strengthen.
Purpose The present study intends to shed light on behaviour of customers towards usage of social media for purchasing decisions. The study proposes an extension to technology acceptance model (TAM) to analyse the significance of monetary benefits and information reliability on customers' intention to use social media. Design/methodology/approach The sample was drawn from social media users of north-western region of India ( n = 622). The proposed model was tested using exploratory factor analysis and structural equation modelling. Findings Results indicate that monetary benefits and perceived ease of use have significant influence on customers' intention to use social media, while information reliability and monetary benefits significantly influence only through perceived usefulness. Practical implications The findings are valuable to marketers to acknowledge the potential of social media to reach to masses by providing timely and reliable information. The study also reveals that website/app developers should implement a user-friendly interface and reliable content to influence customers' usage behaviour. Originality/value The study is a unique attempt to examine the effect of monetary benefits and information reliability with TAM's key constructs in the context of social media adoption. Studies undertaken aforementioned dimensions are mainly concerned with examining direct contribution of social media and its effect on purchase decisions.
Full-text available
The attainment of quality in products and services has become a pivotal concern of the 1980s. While quality in tangible goods has been described and measured by marketers, quality in services is largely undefined and unresearched. The authors attempt to rectify this situation by reporting the insights obtained in an extensive exploratory investigation of quality in four service businesses and by developing a model of service quality. Propositions and recommendations to stimulate future research about service quality are offered.
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Much research in sociology and labor economics studies proxies for productivity; consequently, little is known about the relationship between personal contacts and worker performance. This study addresses, for the first time, the role of referral contacts on workers' performance. Using employees' hiring and performance data in a call center, the author examines the performance implications over time of hiring new workers via employee referrals. When assessing whether referrals are more productive than nonreferrals, the author also considers the relationship between employee productivity and turnover. This study finds that referrals are initially more productive than nonreferrals, but longitudinal analyses emphasize posthire social processes among socially connected employees. This article demonstrates that the effect of referral ties continues beyond the hiring process, having long-term effects on employee attachment to the firm and on performance.
The manner in which the concept of reciprocity is implicated in functional theory is explored, enabling a reanalysis of the concepts of "survival" and "exploitation." The need to distinguish between the concepts of complementarity and reciprocity is stressed. Distinctions are also drawn between (1) reciprocity as a pattern of mutually contingent exchange of gratifications, (2) the existential or folk belief in reciprocity, and (3) the generalized moral norm of reciprocity. Reciprocity as a moral norm is analyzed; it is hypothesized that it is one of the universal "principal components" of moral codes. As Westermarck states, "To requite a benefit, or to be grateful to him who bestows it, is probably everywhere, at least under certain circumstances, regarded as a duty. This is a subject which in the present connection calls for special consideration." Ways in which the norm of reciprocity is implicated in the maintenance of stable social systems are examined.