Content uploaded by Edward C Malthouse
Author content
All content in this area was uploaded by Edward C Malthouse on Jan 17, 2017
Content may be subject to copyright.
SERVICE SCIENCE
Vol. 00, No. 0, Xxxxx 0000, pp. 000–000
issn 2164-3962 |eissn 2164-3970 |00 |0000 |0001
INFORMS
doi 10.1287/xxxx.0000.0000
c
0000 INFORMS
Authors are encouraged to submit new papers to INFORMS journals by means of
a style file template, which includes the journal title. However, use of a template
does not certify that the paper has been accepted for publication in the named jour-
nal. INFORMS journal templates are for the exclusive purpose of submitting to an
INFORMS journal and should not be used to distribute the papers in print or online
or to submit the papers to another publication.
The Dynamics of Consumer Engagement with Mobile
Technologies
Vijay Vishvanathan
Department of Integrated Marketing Communications, Northwestern University, Evanston, IL 60208,
vijay-viswanathan@northwestern.edu
Linda Hollebeek
?, University of Food Plains, Food Plains, MN 55599, deseo79@gmail.com
Edward Malthouse
Department of Integrated Marketing Communications, Northwestern University, Evanston, IL 60208, ecm@northwestern.edu
Ewa Maslowska
Department of Integrated Marketing Communications, Northwestern University, Evanston, IL 60208,
ewa.maslowska@northwestern.edu
Su Jung Kim
Department of Advertising, Iowa State University, Ames, IA ???, ksio2@northwestern.edu
Wei Xie
Department of Industrial Engineering, RPI University, ???, NY ???, WeiXie2013@northwestern.edu
While significant insights about the customer engagement concept have been gleaned in recent literature,
little remains known regarding the nature and dynamics characterizing customers’ engagement with mobile
apps, particularly from a longitudinal perspective. Therefore, a key objective of this paper is to examine how
customer engagement with branded mobile apps is related to purchase behaviors over time as a dynamic
iterative process. To investigate this issue we deploy a unique dataset addressing customers’ mobile app
engagement and purchases. Surprisingly, the results from a VAR model suggest that customer mobile app dis-
engagement, where consumers abandon the app, has a stronger long-term effect on purchase behaviors than
customers’ engagement with the app. In addition, purchase behavior alleviates customer disengagement with
the app. The study, therefore, provides novel findings pertaining to the dynamic inter-relationship between
customers’ engagement with new digital media and purchase behaviors, from which we draw important
scholarly and managerial implications.
Key words : Engagement, disengagement, mobile apps, purchase, VAR model
1
Vishvanathan et al.: Customer Engagement with Mobile Apps
2Service Science 00(0), pp. 000–000, c
0000 INFORMS
1. Introduction
Approaches and tools for managing customer interactions have evolved rapidly over the last two
decades (Hoffman and Fodor 2010, Kaplan and Haenlein 2010). A significant driver of this rapid
shift resides in the emergence and proliferation of modern information and communication technolo-
gies (ICTs), which serve as platforms facilitating interactions with and among consumers (Brodie
et al. 2013, Hoffman and Novak 1996, Sawhney et al. 2005, Trusov et al. 2009, Wiertz and Ruyter
2007). Examples include personal computing devices, the Internet, social media platforms, mobile
devices, and mobile applications. These technologies have spawned numerous new ways for com-
panies to interact with consumers (Kaplan and Haenlein 2010, Men and Tsai 2013), which are also
subject to continuous innovation and evolution. The technologies also produce a digital record of
many interactions, creating big data sets.
Among these emerging technologies, branded mobile applications (hereafter referred to as ‘mobile
apps’) used on smartphones and tablets are changing the ways in which customers interact with
brands. According to Nielsen (2014), two-thirds of U.S. consumers own smartphones and spend
86% of their phone usage time interacting with apps. Further, the total number of downloaded
mobile apps is expected to reach 268 billion by 2017 (Fox 2013), while the global mobile app market
is anticipated to reach a value of $55 billion in 2016 (Schadler and McCarthy 2012). (ECM: These
numbers look out of date. Can we find more recent numbers than from 2012? Note that when
I see stats from 2011/12 it signals to the reviewer that this has been around at other journals
without being updated.) Moreover, Gruman (2011) predicts that by 2020, mobile devices will
increasingly include wearable components that integrate wirelessly with each other, as well as other
nearby devices. Overall, these trends suggest how deeply mobile apps have penetrated smartphone
users’ daily lives (Report 2011). Their broad accessibility, ease of use and ubiquitous nature makes
apps powerful tools facilitating customer engagement and empowerment (Schadler and McCarthy
2012); thereby, assisting individuals to proactively co-create their own, as well as each other’s,
brand-related experiences (De Valck et al. 2009, Hollebeek et al. 2014, Nambisan and Baron 2007,
Prahalad and Ramaswamy 2004). Correspondingly, Schadler and McCarthy (2012) of Forrester
Research posited “mobile [to be] the new face of engagement.”
There is a growing body of research testing how the use of mobile apps affects subsequent
purchases (Bellman et al. 2011, Wang et al. 2015, Kim et al. 2015). While the number of studies
is currently small, there is a growing consensus that mobile adoption and use positively affect
subsequent purchases, confirming that developing a mobile app can be an effective strategy for firms
seeking to engage, co-create with and, ultimately, retain customers (Schadler and McCarthy 2012).
Most contemporary firms are concerned with the strategic enhancement of customer loyalty and
lifetime value (Rust et al. 2004). Traditionally, the specific types of activities deployed to achieve
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 3
such objectives, including advertising and other promotional tools, tend to be relatively costly in
nominal terms (Kaplan and Haenlein 2010, Mangold and Faulds 2009). Engaging customers by
virtue of branded mobile apps, by contrast, constitutes a potentially cost-saving alternative (after
recovery of the incurred app development costs), which may be used toward the achievement of
broader strategic organizational objectives.
Previous research on mobile apps, however, has focused on a one-way relationship between app
use and outcomes such as purchase. We posit that the relationship is more complicated and mul-
tidimensional. Use of a branded app certainly exposes customers to a brand and thereby increases
the likelihood of purchase and use of the service, but the relationship could go the other way as
well: a customer who uses a service and derives value from it may seek out the app and use it. For
example, someone who flies often with an airline may download and use the airline’s app because
it conveniently provides flight status, allows the customer to select seats, etc. Use of the app may,
in turn, increase the customer’s loyalty (i.e., future ticket purchases) with the airline. Thus, there
is a dynamic, iterative relationship where purchases cause app use, which causes additional pur-
chases, which causes more app use, and so on. Rather than being unidirectional, the relationship is
symbiotic, where app use and purchase mutually sustain and reinforce each other. Moreover, there
is a possibility that the app will disappoint the user by not providing sufficient value, breaking the
symbiosis. When this occurs, it is of interest to know whether purchases are affected. For example,
can a disappointing experiences with an app cause a customer to reduce the purchase of some
service?
The fact that there are few empirical studies that examine the relationship between customers’
mobile app engagement and their purchase behaviors, and none that model the dynamic interplay
between engagement, purchase and consumption, provides the impetus for the present, longitu-
dinal investigation. Extending Van Ittersum et al. (2013), we hypothesize that multiple variables
iteratively affect one another over time. Correspondingly, we adopt a vector autoregressive (VAR)
model to analyze this iterative, longitudinal set of processes (Srinivasan et al. 2004, Trusov et al.
2009). We also examine customer mobile app disengagement, as reflected by individuals’ discon-
tinued usage of a focal mobile app (Hollebeek et al. 2014, Peck and Malthouse 2011), which has
not received any empirical scrutiny.
This large-scale, longitudinal investigation into customers’ mobile app engagement directly
responds to the calls in the literature (Brodie et al. 2011, Hollebeek 2011,b, 2012, Leeflang 2011, MSI
2014) and provides novel insights into this emerging area. Further, by investigating the dynamic
inter-relationship between customers’ app engagement, purchase and consumption behaviors, this
study also directly responds to the calls from Van Doorn et al. (2010) and Hollebeek et al. (2014) for
further exploration of customer engagement-based causal relationships within broader nomological
Vishvanathan et al.: Customer Engagement with Mobile Apps
4Service Science 00(0), pp. 000–000, c
0000 INFORMS
networks. In addition, investigation into customer engagement fits within the broader theoretical
strands of literature addressing service-dominant (S-D) logic and relationship marketing (Palmatier
et al. 2009, Vargo and Lusch 2004, 2008, Brodie et al. 2011).
2. Literature Review & Conceptual Development
2.1. The Emerging Use of Mobile Apps as an Engagement Tool
Branded apps represent a relatively new way of communicating with customers (Kim et al. 2013,
2015), and may be viewed as an interactive advertising sub-form (Bellman et al. 2011). Wang
et al. (2016) discuss how mobile apps and platforms can serve as a new and innovative advertising
channel. Mobile media are characterized as being portable, personal, multi-modal, and converged
(Larivi`ere et al. 2013). With their portable and personal nature, mobile apps have a great potential
to be a pull promotional tool, i.e., the consumer chooses to download an app considered relevant and
subsequently decides when, where, and how often to use it (Bellman et al. 2011). The interactive,
multi-modal, and converged nature of mobile apps allows consumers to have longer attention and
deeper cognitive processing of relevant information. In addition, sticky apps that provide value
relevant to customers help them build a habit of using the apps on a regular basis (Wang et al. 2015).
In sum, the unique characteristics of mobile apps facilitate a high level of customer engagement
with the mobile app and the brand that provides the app (Calder and Malthouse 2008, Kim et al.
2013), which ultimately impacts purchase behavior (Kim et al. 2015).
2.2. Conceptualizing Customer Mobile App Engagement Behavior
Brodie et al. (2011, p. 260) suggest that engagement occurs by virtue of interactive, co-creative
customer experiences with a focal agent/object (e.g., a brand) in specific service relationships and
that varying context-dependent conditions generate differing customer engagement levels. Further-
more, they suggest that it exists as a dynamic, iterative process within service relationships that
co-creates value. Other authors have highlighted the particular importance of the directly observ-
able, behavioral dimension underlying the engagement concept. Specifically, this has been referred
to as customer engagement behaviors (CEBs) (Jaakkola and Alexander 2014, Van Doorn et al.
2010, Verleye et al. 2013). Van Doorn et al. (2010, p. 253) define CEBs as “customers’ behavioral
manifestation toward a brand or firm, beyond purchase, resulting from motivational drivers.” The
conceptual ambit of CEBs includes a broad array of behaviors, including word-of-mouth activity,
helping other customers, writing reviews, and customers’ focal branded mobile app-related behav-
iors (e.g., log-ins and check-ins, which represent the particular CEBs examined in this study). Based
on these conceptualizations, we define mobile app engagement behavior as customers’ interactive
experience with the focal branded mobile app. CEBs are expected to affect particular customer- and
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 5
firm-based variables, including the development of specific attitudes, and the attainment of focal
financial outcomes (Van Doorn et al. 2010).
Additionally, extending the work of Van Doorn et al. (2010), Verleye et al. (2013) and Jaakkola
and Alexander (2014), we study customer disengagement, which is indicated by when a customer
who was formerly engaged stops performing CEBs. O’Brien and Toms (2008) states, “Disengage-
ment occurred when participants made an internal decision to stop the activity, or when factors in
the participants external environment caused them to cease being engaged.” (ECM: Ewa, these are
great. Can you give a page numbers for both?) Likewise Brodie et al. (2013) state that “‘Termina-
tion’ represents a state of more permanent disengagement, and as such, refers to the conclusion of
a consumer’s engagement with a particular brand community.” In the case of mobile apps, disen-
gagement is indicated by discontinued use of check-ins and logins. Peck and Malthouse (2011, pp.
230-232) discuss psychological motivations for disengaging with a media product such as a mobile
app and identify several experiences including information overload, poor or annoying design, and
poor-quality content. Our focus is on the observable manifestation, discontinued use.
2.3. The Dynamic Inter-Relationship between Customer Mobile App Engagement,
Purchase and Consumption Behaviors
While the Brodie et al. (2011) definition emphasizes that engagement is a “dynamic, iterative pro-
cess,” there has not been any research that models it as such. Figure 1 gives our conceptual model
showing how the different variables interact. We begin our discussion with the purchase and con-
sumption boxes. Purchase behaviors end in a transaction where (usually) money is exchanged for
the service or product, while consumption involves using whatever was purchased for its intended
purpose. Sometimes purchase and consumption occur more or less concurrently, such as paying
for a bus fare (purchase) and then riding to the destination (consumption), or eating dinner at a
restaurant (consumption) and paying the bill (purchase). In other situations the purchase experi-
ence happens prior to consumption, e.g., one could purchase season theater tickets, but not attend
shows (consumption) for several months. It is not uncommon for the purchase of an airline ticket to
happen weeks or even months before the actual flight (consumption). One could purchase a month
of Netflix or health club service today but watch movies or workout (consumption) throughout the
month. The purchaser might also forget or become too busy to watch/workout, in which case a
transaction has occurred without any consumption. One might also read news stories on websites
or listen to music on a streaming service, both examples of consumption, without paying. They
may eventually purchase a subscription or buy the album at a later time, illustrating how con-
sumption can happen before purchase. These examples illustrate how purchase and consumption
can be different behaviors in various service settings, how one can occur without the other, and
also how one may be antecedent to the other.
Vishvanathan et al.: Customer Engagement with Mobile Apps
6Service Science 00(0), pp. 000–000, c
0000 INFORMS
Figure 1 Conceptual Model of Customer Engagement/Disengagement
Put conceptual framework here.
Now consider the role of engagement behaviors, including those with a mobile app. As the airline
app example in the introduction illustrated, it is easy to construct examples showing how CEBs
can cause, or be caused by, purchase and consumption. One may purchase skis from REI, and
then want to download the REI Snow Report app for information about snow conditions. Using
the app could enhance the “consumption” of the skis, and lead to purchases of other accessories
at REI, such as winter clothing. Reading a favorable consumer review of a restaurant, which is
considered an engagement behavior by Muntinga et al. (2011), may cause someone to visit it. The
consumption experience may cause the diner to write her/her own review, another CEB.
Now consider the role of app disengagement. When a mobile app fails to provide value for
customers, they will stop using it. More generally, a customer could become engaged with any
environment designed to facilitate CEBs, such as a company-sponsored discussion forum focused
on discussion of a software product (e.g., the Microsoft Office discussion forum). A poor purchase
or consumption experience could cause a customer to cut all ties with the brand, including an app
or discussion forum. Likewise a poorly designed, bug-prone or worthless app in terms of fulfilling
a customer’s goals may cause the customer to discontinue using it. The more important questions
are whether (1) abandoning an app could affect subsequent purchase or consumption of the service
itself, and (2) whether favorable purchase and consumption episodes can cause a customer to
re-engage with the app and other CEBs (i.e., decrease disengagement).
(There’s still debate about whether this is necessary. I dropped the previous paragraph.
This provides some further lit review discussing the dynamic process. I could go either way,
keep/drop.) Although increased purchases have been mostly discussed as a consequence of engage-
ment (Van Doorn et al. 2010), the effect of engagement on purchase behaviors does not necessarily
fully describe the relationship between the two concepts. For example, Gummerus et al. (2012)
note that customer engagement with a brand community may develop as a result of pre-existing
brand loyalty, which is further strengthened by the positive experience customers have taking part
in the community activities. Analogously, Brodie et al. (2013) suggest that a number of desig-
nated engagement consequences (including brand loyalty), may act as antecedents to subsequent
engagement behaviors; thus reflecting the dynamic, iterative nature of the engagement process.
3. Research Design
3.1. Data
The main objective of the study is to examine how CEBs with a mobile app and pur-
chase/consumption behaviors influence each other over time. To achieve this objective, we utilize
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 7
a dataset sourced from the Canadian ‘Air Miles Reward Program’ (AMRP), a loyalty program
run under the aegis of LoyaltyOne. The Air Miles brand is one of the most recognized brands in
Canada (ECM: figure out how to make this “Loyalty One.”) (One 2015). It is important to note
that AMRP is unrelated to any airline. It is a coalition loyalty program providing a membership
card that can be swiped at various sponsors across a number of categories, including grocery stores,
petrol stations, drug stores, home improvement stores, and financial services; thus providing exam-
ples of specific CEBs. Customers are able to earn loyalty points on their AMRP accounts when
they use their particular credit cards when making purchases. Within each category, only a single
sponsor is used (e.g., for gasoline, Shell represents the only supplier of points, etc.).
When customers swipe their AMRP card at a sponsor location, they receive points approximately
proportional to their purchase amount and AMRP receives a payment from the sponsor in return
for the points issued. Thus, AMRP earns revenue when the consumer goes to a sponsor and
accumulates points by swiping the card. AMRP intends for members to purchase from sponsors,
rather than non-sponsoring firms. Accumulating points is the purchase behavior of interest to
AMRP, since it is directly linked to its revenues. Consumers are entitled to redeem their points
for a variety of rewards ranging from merchandise (e.g., televisions, blenders, mixers, etc.), travel,
gift certificates, or instant cash. Such rewards are the tangible value the company provides to the
consumer; and therefore, are expected to represent consumers’ main reason to enroll in the AMRP
loyalty program; that is, with customers being motivated to accumulate points with a view to
receiving a future reward (Blau 1964). Redemption is how members “consume” the AMRP service.
We are drawing an analogy with our earlier examples. Just as purchasing a airline or theater
ticket, or Netflix subscription involves a monetary transaction where money is exchanged for the
promise of a service, accumulating points generates revenue for AMRP, with the consumer expec-
tation that the points will eventually provide tangible value. Just as watching a movie is how
consumers derive value from a Netflix subscription, redeeming points for some award is how con-
sumers derive value from AMRP. (ECM: Not sure this is necessary, but I worry that reviewers will
not understand our purchase-consumption distinction and how it applies to AMRP.)
AMRP developed the AIR MILES branded mobile app and launched it in February 2012, which
allows customers (i.e., loyalty program members) to undertake specific CEBs, including logging-
in to check their point balances, browse potential reward items, keep track of their purchase
histories and progress toward the attainment of particular rewards (including by emailing details
regarding particular rewards to a specified address), find sponsors nearby, and check-in at sponsors.
Consumers can share their check-in information on specific social networking sites (e.g., Facebook)
helping to inform their family and friends where they shop, or to keep a record of specific locations
they have visited. It is important to note that consumers are unable to make purchases by using
Vishvanathan et al.: Customer Engagement with Mobile Apps
8Service Science 00(0), pp. 000–000, c
0000 INFORMS
this app. The app was introduced as an additional brand touch point primarily to stimulate the
undertaking of CEBs such as log-ins and check-ins. This is consistent with the definition of CEBs
as customers’ interactions with the AIR MILES app do not directly relate to purchase (Van Doorn
et al. 2010, MSI 2014).
AMRP initially provided daily transactional information reflecting individuals’ usage with the
mobile app, point accumulation and redemption for 548,569 customers from February 7, 2012 to
February 12, 2013. We selected a simple random sample of 10% (i.e., 54,858) of the customers for
the analysis. We then aggregated the available information on CEBs i.e., recency and frequency of
app usage, as well as their point accumulation and redemption at the weekly level for 54 weeks,
thus resulting in a total of 2,962,332 customer-week observations for analysis.
3.2. Measures
We compute two measures of CEBs, namely the frequency and recency of customers’ usage of
the app. Frequency is computed using the total number of log-ins and check-ins carried out by a
customer every week and is a direct measure of a consumer’s cumulative behavioral engagement
with the app. Higher frequency values and increased recency suggest greater CEBs with the app,
while lower frequency values and decreased recency indicate greater disengagement behaviors.
Recency is measured using the number of weeks elapsed since a customer’s most recent app activity
and directly measures whether a customer has become disengaegd with the app. Large values
indicate that the customer has abandoned the app.
We calculate the total number of points accumulated by a customer as a measure of purchase.
In addition, because AMRP is a loyalty program, we also calculate the total number of points
redeemed each week in order to measure redemption behavior and control for it in the analysis.
3.3. Descriptive Statistics
Summary descriptive statistics for the final sample are shown in Table 1. Overall, the statistics
suggest that, on average, customers use the app less than once a week. The average time between
two consecutive uses was 17.7 weeks, while the number of weekly points accumulated equals, on
average, 22.7 points; with average weekly redemption equaling 19.4 points. However, the standard
deviation and range (i.e., max −min) for these variables are positive and substantial. Additionally,
we included individuals’ age and tenure (defined as the number of years in the loyalty program)
as demographic variables. The average age reported in our sample was 37.4 years, and the average
loyalty program membership tenure was 9.5 years.
For the analysis, we log-transformed values of these measures based on the observed right skewed
counts and outliers. Specifically, the log transformation symmetrizes the distributions, reduces the
influence of outliers and stabilizes their variance. Since the minimum value for each of the four focal
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 9
Table 1 Descriptive Statistics
Recency of Frequency of Points Points
App Use App Use Accumulated Redeemed Age Tenure
Mean 17.41 0.0088 22.70 19.40 37.40 9.58
SD 14.55 0.1112 134.56 342.61 12.22 6.29
Min 0.00 0.0000 0.00 0.00 12.10 0.00
Max 54.00 6.9027 42,894.00 122,933.00 78.17 20.91
variables is zero and the logarithm of zero is not defined, we increased the value of each variable by
one prior to the transformation. The correlations between the log-transformed variables are also
reported in Table 2.
Table 2 Bivariate Correlations of Log-Transformed Variables
Recency of Frequency of Points Points Age Tenure
App Use App Use Accumulated Redeemed Age Tenure
Recency of App Use 1.000 −0.547 −0.076 −0.011 0.068 0.065
Frequency of App Use 1.000 0.081 0.017 −0.031 −0.023
Points Accumulated 1.000 0.089 0.199 0.204
Points Redeemed 1.000 0.040 0.046
Age 1.000 0.578
Tenure 1.000
Note: All correlations are significant with p < 0.01; Number of subjects 54,858; Total number of
observations 2,962,332
3.4. Model Specification
Based on the conceptual model shown in Figure 1, we expect frequency, recency, point accumula-
tion and redemption to represent a dynamic, iterative process (Brodie et al. 2011, Jaakkola and
Alexander 2014, Van Doorn et al. 2010). Correspondingly, we employ a vector autoregressive (VAR)
model to account for the dynamic nature of interactions between the four variables studied. Before
constructing a VAR model we conduct various pre-tests.
We first test for simultaneity between the focal variables by using the pairwise Granger causality
test (Granger 1969, Hanssens and Schultz 2001). The null hypothesis assumes that adding lagged
values of Xdoes not improve Y’s prediction. For each pair, we conducted the test varying the
number of lags of the independent variable from two to ten, attaining consistent results irrespective
of the number of lags introduced. The pairwise Granger causality test results shown in Table 3
reveal the presence of simultaneity between the variables in our framework (p < .01), with the
exception of the effect of customers’ points redeemed on the recency of their usage of the mobile app.
Therefore, we construct a full VAR to quantify the dynamic interaction among points accumulated,
points redeemed, frequency and recency of app use.
Vishvanathan et al.: Customer Engagement with Mobile Apps
10 Service Science 00(0), pp. 000–000, c
0000 INFORMS
Table 3 Results (p-Values) from Pairwise Granger Causality Test using 10 Lags of Log-Transformed Variables
Explanatory Response Variable
Variable Recency Frequency Points Accumulated Points Redeemed
Recency — 0.00 0.00 0.00
Frequency 0.00 — 0.00 0.00
Recency 0.00 0.00 — 0.00
Points Redeemed 0.39 0.00 0.00 —
An evolving variable has an infinite variance and suggests the presence of permanent effects,
while a stationary variable has a finite variance and suggests the existence of temporary effects.
Testing for stationarity, therefore, is critical since inferential statistics can be computed only for
variables with a finite variance (Granger and Newbold 1974). The null hypothesis of a unit root
test is that a variable has a unit root (i.e., the variable is ‘evolving’). In the event that a variable
has a unit root, a common solution is to use the first differenced value of the variable in the model.
Alternatively, the variable can be used in its current form; that is, in levels, for the estimation. We
conducted the unit root test for panel data as suggested by Im et al. (2003), and reject the null
hypothesis for each of the four focal variables (p < .01). The results, therefore, suggest that we are
able to utilize the values of the variables in levels for the estimation.
Based on the results outlined in the previous sub-section, we specify a VAR model where the
log-transformed variables of customers’ recency of the mobile app use (lnR), frequency of the
mobile app use (lnF), point accumulation though purchases (lnP), and point redemption (lnC) for
individual iin week tis influenced in a dynamic fashion through j= 1,...,p lags of the endogenous
variables. For the nth (n= 1,2,3,4) response variable, parameters bj
nm captures the effect of the
jth lag of the mth (m= 1,2,3,4) endogenous variable and parameters γnh capture the effects of
the h= 1,2 exogenous variables of age and tenure. We also include a vector of intercepts Anand
assume the error terms nt are distributed as MVN(0,Σ). The VAR system can be written as:
ln Rit
ln Fit
ln Mit
ln Cit
=
A1
A2
A3
A4
+
p
X
j=1
bj
11 bj
12 bj
13 bj
14
bj
21 bj
22 bj
23 bj
24
bj
31 bj
32 bj
33 bj
34
bj
41 bj
42 bj
43 bj
44
ln Ri,t−j
ln Fi,t−j
ln Mi,t−j
ln Ci,t−j
+
γ11 γ12
γ21 γ22
γ31 γ32
γ41 γ42
ln ageit
ln tenureit +
1t
2t
3t
4t
.
(1)
4. Results
The number of lags pto be included in the model was decided based on several criteria (Gredenhoff
and Karlsson 1999), including the Schwarz Information Criterion (BIC), Akaike Information Crite-
rion (AIC), and Final Prediction Error (FPE). We consistently found across each of these criteria
that, while their values decrease significantly when one lag of the endogenous variables is included,
there is little change with additional lags. Consequently, we estimate a VARX(1) model; that is,
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 11
a VAR model containing age and tenure as exogenous variables, and one lag of the endogenous
variables.
Given the dynamic nature of our model, the estimated parameters are not easy to interpret
independently (Sims 1980). Consequently, following Sims and Zha (1999) we calculate Impulse
Response Functions (IRFs). The IRFs capture both the direct and indirect effects of a shock to
one of the variables on the other variables in the dynamic system. However, it is important to
note that the focal variables pertaining to CEBs, and point accumulation and redemption are
log-transformed; hence the IRFs obtained are also log-transformed. However, it is desirable to
obtain the IRFs for each variable in the customary units of measurement; that is, without the
log-transformation, thus facilitating interpretation and the development of actionable managerial
recommendations. A common approach to remove the log transformation is to simply take the
exponential. However, Ari˜no and Franses (2000) demonstrate that this approach leads to biased
forecasts, and hence propose an alternative method to obtain the forecasts of the response variables
without log-transformations under the assumption of normally distributed error terms. Wieringa
and Horvth (2005) employ this approach and explain, using a marketing application, how to obtain
the level-impulse response. We follow their methodology and refer the reader to these studies for
further detail.
An IRF for each level is defined as the difference of the level forecasts of the shocked system, and
the level forecasts of the non-shocked system (Wieringa and Horvth 2005). In order to understand
the dynamic effects, we applied a shock by increasing the average value of one of the focal variables
(e.g., recency) by 20% (we have tried other stock levels and found similar conclusions); followed by
the computation of the IRF for the other three variables as the difference of their level forecasts
in the shocked system, and the level forecasts of a non-shocked system. In this case, the IRF of
points accumulated kweeks after the shock was applied to recency is calculated as IRF(k|Xt) =
PXi+k|Xk−PXi+k|Xt(check this)where PXt+k|Xtis the level forecast of points accumulated
at period t+k, and PXt+k|Xkis the level forecast of points accumulated assuming recency is
shocked at time t. The response functions for points redeemed to a shock in recency, too, can be
calculated in a similar manner. Further, we shocked the other variables in the model one at a time
and computed the differences in the level forecasts of the shocked system and the level forecasts
of the non-shocked system for the other three variables, thus obtaining a complete set of IRFs. In
order to obtain the confidence intervals for the IRFs, we performed a bootstrap where we resampled
the errors 1000 times, and then re-estimated the level responses.
A 20% shock generates an increase over the average value of each endogenous variable by the
following amounts (Is this the right table?) (cf. Table 1): an increase of 0.002 in the frequency
of customers’ mobile app use, an additional 4.5 points accumulated, approximately 4 additional
Vishvanathan et al.: Customer Engagement with Mobile Apps
12 Service Science 00(0), pp. 000–000, c
0000 INFORMS
points redeemed, and approximately 3 additional weeks of lapsed use for recency. We summed the
IRFs obtained over time to capture the ‘net effects’ of a shock on other variables in the system.
Since the variables pertaining to frequency and recency have different units of measurement from
those pertaining to point accumulation and redemption, we cannot use the IRFs or net effects
to compare their relative effects. We, therefore, also calculated a measure without units, namely
elasticities at time period k, using the formula (Check equation)
∂PK
k=1 Yt+k
PK
k=1 Yt+k
·Xt
∂Xt
,
where Yis the response variable and Xis the shocked variable. In the interest of brevity, we report
the net effects (cf. Table 4) and elasticities (cf. Table 5) for three particular time periods. We term
the net effects and elasticities one week from the shock (i.e., t+1) as ‘immediate effects,’ 20 weeks
from the week of the shock as ‘medium-term effects,’ and 40 weeks from the week of the shock as
‘long-term effects.’
(The table does not quite fit. One fix would be to use a smaller font. Another is to give 4 accurate
digits. I kind of prefer the latter.)
Before proceeding to the results, we briefly summarize the steps in the estimation process:
1. Estimate equation 1 using the least squares approach;
2. Use the estimated parameters and draw residuals from MVN(0,Σ) to estimate the level fore-
casts of the non-shocked system 1000 times for the next forty weeks following Wieringa and Horvth
(2005);
3. Shock each focal variable by 20% of its average value at time tand compute the level forecasts
of the shocked system for the next 40 time periods 1000 times using the same bootstrapped residuals
as in step 2;
4. Compute the mean IRF as the difference of the level forecast of the shocked system and the
level forecast of the non-shocked system and corresponding error bands;
5. Calculate the mean ‘net effect’ for time period t+kas the sum of the IRF until time period
kand also corresponding error brands; and
6. Calculate the elasticities using the formula in the previous paragraph.
(The plots show three lines and it is not clear what they mean, i.e., red vs. blue vs. green.)
Figure 2 plots the IRFs for the four focal variables employed in the study. The top row displays the
response of frequency, points accumulated and points redeemed to a shock in recency; the second
row displays reveals the response of points accumulated, points redeemed and recency to a shock in
frequency; the third row shows the response of frequency, points redeemed and recency to a shock
in points accumulated; and the final row shows the response of frequency, points accumulated and
recency to a shock in points redeemed.
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 13
Table 4 Accumulated IRF: Net Effects of a 20% Increase in the Shocked Variable
Shocked Response Immediate Medium-term Long-term
Variable variable (t+ 1) (t+ 20) (t+ 40)
Recency Frequency −0.00192 −0.07191 −0.09154
[−0.00194,−0.00189] [−0.07503,−0.06878] [−0.09562,−0.08746]
Points −0.98336 −23.86389 −29.46730
Accumulated [−1.18410,−0.78263] [−35.25626,−12.47151] [−43.67357,−15.26102]
Points −0.00030 −0.01229 −0.01528
Redeemed [−0.00030,−0.00029] [−0.01260,−0.01198] [−0.01566,−0.01489]
Frequency Points 0.02229 0.12877 0.12723
Accumulated [0.01774,0.02685] [0.07490,0.18264] [0.07413,0.18032]
Points 0.00006 0.00021 0.00021
Redeemed [0.00006,0.00006] [0.00020,0.00021] [0.00020,0.00021]
Recency 0.00029 0.00841 0.01096
[0.00029,0.00029] [0.00827,0.00854] [0.01075,0.01117]
Points Frequency 0.00092 0.00501 0.00520
Accumulated [0.00091,0.00094] [0.00485,0.00518] [0.00503,0.00537]
Points 0.00806 0.01178 0.01181
Redeemed [0.00786,0.00825] [0.01150,0.01207] [0.01153,0.01210]
Recency −0.01922 −0.32144 −0.40570
[−0.01926,−0.01917] [−0.32618,−0.31671] [−0.41294,−0.39847]
Points Frequency 0.00006 0.00063 0.00065
Redeemed [0.00006,0.00006] [0.00061,0.00065] [0.00063,0.00067]
Points 1.13434 1.93023 1.93562
Accumulated [0.90279,1.36590] [1.36083,2.49963] [1.36355,2.50770]
Recency −0.00032 −0.03022 −0.03876
[−0.00032,−0.00032] [−0.03069,−0.02974] [−0.03948,−0.03803]
Note 95% Confidence Intervals in brackets
Figure 2 IRF Plots (With 95% Confidence Intervals)
Put IRFs here.
The first row of the IRF plots shows that an increase in the value of recency results in less
frequent use of the app in the next period, fewer points accumulated, as well as fewer points
redeemed. These plots also reveal that the effects of a shock to recency on all other three variables
endure for a considerable amount of time (and therefore what? I think this means that the longer
a cusotmer has abandoned an app, the harder it will be to get then to use the app again, and the
accumulation and redemption behavior decline. You might want to split this into four paragraphs
or make it a bullet list?). The second row of IRF plots reveals that more frequent use of the app
has a positive effect on the number of points accumulated (i.e., revenue to the company), as well
as a small positive effect on the number of points redeemed and recency of the customer’s last use
of the app (This sounds like a small positive effect on recency, meaning they are lapsed longer?).
While the effect of frequency on points accumulated and points redeemed seems to diminish after
around 20 weeks, the effect on recency persists for a longer period of time (So what? This means
that use begets more use. Or maybe use in this period inoculates the company from disengagement
Vishvanathan et al.: Customer Engagement with Mobile Apps
14 Service Science 00(0), pp. 000–000, c
0000 INFORMS
Table 5 Elasticities: Immediate, Medium-Term and Long-Term
Shocked Response Immediate Medium-term Long-term
Variable variable (t+ 1) (t+ 20) (t+ 40)
Recency Frequency −0.05439 0.71964 −0.12332
[−0.36226,0.25347] [−0.71622,2.15551] [−1.48204,1.23539]
Points −0.08269 −1.25574 −2.71208
Accumulated [−0.10442,0.06096] [−1.50697,−1.00450] [−4.67793,−0.74623]
Points −0.00512 −9.41374 −0.33110
Redeemed [−0.01720,0.00696] [−27.53426,8.70678] [−0.61446,−0.04775]
Frequency Points 0.00187 0.00998 0.01923
Accumulated [0.00138,0.00237] [0.00744,0.01253] [0.00350,0.03496]
Points 0.00098 0.15715 0.00448
Redeemed [−0.00133,0.00328] [−0.14529,0.45959] [0.00059,0.00837]
Recency 0.00008 0.00193 0.00240
[0.00008,0.00008] [0.00191,0.00195] [0.00236,0.00243]
Points Frequency 0.02619 −0.05953 0.00285
Accumulated [−0.12203,0.17441] [−0.16911,0.05005] [−0.08191,0.08760]
Points 0.13936 8.65454 0.25399
Redeemed [−0.18963,0.46834] [−7.99482,25.30390] [0.02941,0.47856]
Recency -0.00546 −0.07441 −0.08966
[−0.00546,−0.00546] [−0.07539,−0.07344] [−0.09113,−0.08819]
Points Frequency 0.00163 −0.00723 0.00047
Redeemed [−0.00757,0.01083] [−0.02076,0.00631] [−0.00994,0.01088]
Points 0.09538 0.24481 0.54954
Accumulated [0.07031,0.12045] [0.16731,0.32231] [0.05518,1.04391]
Recency -0.00009 −0.00695 −0.00850
[−0.00009,−0.00009] [−0.00704,−0.00687] [−0.00863,−0.00837]
Note: 95% Confidence Intervals in brackets
in the next?). The third row of IRF plots indicates that accumulating additional points results
in customers’ more frequent use of the app and more points redeemed. It also results in reduced
values of recency, i.e., higher level of engagement as fewer weeks elapse between two consecutive
uses of the app by a customer. In the last row of plots, we observe that redeeming more points
results in more frequent use of the app and more points accumulated. Redeeming more points also
results in lower values of recency. (I’m not sure we want to make a big deal about this, but this
finding is contrary to current practice. Loyalty programs usually do not encourage members to
redeem points, because this costs the program money. This says that customers become better,
more engaged customers after a redemption.) The IRF plots for points accumulated and points
redeemed are remarkably similar in that their effects on frequency and recency appear to last for
a longer time than on each other. To summarize, the results are broadly consistent with what one
would expect, and validate the thesis proposing a dynamic, iterative model of CEBs (Brodie et al.
2011, Van Doorn et al. 2010); thus representing the first known empirical investigation.
In order to better understand the long-term effects and compare the magnitude of the effects
of a shock to different variables over time, we examine the net effects and elasticities reported in
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 15
Tables 3 and 4, respectively. We first inspect how recency of app use affects points accumulated
and points redeemed. The medium- and long-term net effects (cf. Table 4) of a shock to recency on
points accumulated are significantly greater, relative to the immediate effect. In fact, the effect of
recency on the number of points accumulated in the medium- (||= 1.25) and long-term (||= 2.71)
are ‘elastic’ (cf. Table 5). (Maybe I’m confused, but was the error in equation 1. Does it mean
something different here?) Higher values of recency i.e., greater levels of disengagement, also has
(subject is “values” so “has” should be “have”) a negative effect on the number of points redeemed.
The net effects shown in Table 4 also indicate that a decrease in the value of recency elicits a much
smaller response from points redeemed than from points accumulated. The long-term elasticity
(see Table 5) reveals that the effect of recency on points redeemed in in fact ‘inelastic’ (||= 0.33).
The net effects of frequency on points accumulated and points redeemed are positive (cf. Table 4),
but the magnitude of both these effects appears to be relatively small. Further, the elasticities
in Table 5 suggest that the effect of frequency on the number of points accumulated, as well as
points redeemed, across all time periods is quite small, and can hence be deemed ‘highly inelastic’.
To summarize, CEB measured in terms of recency of app use has a stronger, long-term effect on
ensuing point accumulation and redemption, relative to CEB measured in terms of frequency of
app use.
We now examine the net effects of point accumulation and redemption behaviors on the CEB
measures of recency and frequency of the app use respectively, and the corresponding elasticities.
We observe from Table 4 that the long-term effect of points accumulated on recency is greater than
the medium- and short-term effects thus again suggesting the importance of accounting for the
dynamic nature of the relationship between these two variables in the model. Further, the effect of
points redeemed on recency continues to build over time, though to a much smaller extent than the
effect of points accumulated. Comparing the elasticities shown in Table 5, we observe that the effect
of points accumulated on recency is greater, relative to the effect of points redeemed. Similarly,
the net effects of points accumulated and points redeemed on frequency of the app use continue
to build over time, though to a much smaller extent than their effect on recency. Furthermore, the
95% confidence intervals for the corresponding elasticities include zero, suggesting that the effects
are ‘highly inelastic’.
Finally, from a loyalty program perspective, it is important to understand how the number of
points accumulated and points redeemed affect one another over time. We find that the medium-
and long-term net effects (cf. Table 4) of the number of points accumulated on points redeemed are
greater, relative to the immediate effect. Similarly, we find that that the medium- and long-term
net effects of points redeemed on points accumulated are greater, relative to the corresponding
immediate effect. Interestingly, a comparison of the elasticities (cf. Table 5) suggests that the effect
Vishvanathan et al.: Customer Engagement with Mobile Apps
16 Service Science 00(0), pp. 000–000, c
0000 INFORMS
of the number of points redeemed on points accumulated is greater, as opposed to vice versa. This
represents an important finding for loyalty programs, which are based on the premise that the
offering of rewards incentivizes customers to remain loyal to a focal brand.
To summarize, the results obtained from our VAR model broadly reveal the following insights.
First, the findings validate the existence of a dynamic, iterative customer engagement process
comprising a number of focal CEBs; thus representing the first known empirical investigation
into these dynamics. Second, the effects of recency on point accumulation and redemption are
significant and perhaps much larger than advertising/sales elasticities reported in prior research
(i.e., Hanssens and Schultz (2001) observe that advertising elasticities are typically around 0.1).
(This is huge!) Further, greater levels of disengagement as measured by higher levels of recency were
found to generate fewer points redeemed, thus in turn, generating fewer points accumulated. Third,
increased point accumulation and redemption results in lower values of recency i.e., greater level
of engagement, of the app use. In other words, the findings suggest that greater levels of purchase
and redemption behaviors results (I think the subject is “levels” so “results” should be “result”)
in greater engagement with the mobile app, (which in turn further increases accumulation). Since
point redemption was found to favorably affect recency of customers’ usage of the mobile app, one
way to get customers engaged with mobile apps is to stimulate their perceived value extracted
from their interactions with the brand.
Overall, our findings suggest that the emerging brand touch point of mobile apps provides a
powerful tool for fostering customer engagement behaviors that influence marketplace outcomes
such as purchase behavior. Specifically, based on our findings, engaging customers with mobile apps
represents not only a viable revenue generating opportunity, but also a way to reduce advertising
and promotion costs. Savings accrue because the firm does not buy advertising media and instead
utilizes its branded app to control the medium, grow its audience, and co-create and distribute
content. Nor must the firm reduce its margins through promotional tactics. Hence despite the
upfront investment required for the development of the mobile app, our findings show that the
introduction of branded mobile apps is an effective tactic that fosters the development of desirable
CEBs and other ensuing behaviors, including point accumulation and redemption, thus contributing
to the development of customer brand loyalty and lifetime value. As such, the introduction of
branded mobile apps represents a ‘game changer’ for marketing communications.
5. Discussion
5.1. Contributions and Implications
While many firms engage their customers with specific digital touch points and tools, including
mobile apps, little is known about the effectiveness of such approaches. The main contribution
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 17
of this study is that it responds to this observed challenge and provides initial insights into the
nature and dynamics characterizing the inter-relationship between customers’ specific mobile app-
related CEBs and purchase behaviors over time, while controlling for point accumulation. While the
finding that customers’ point accumulation and redemption behaviors drive their subsequent use
of the mobile app may seem intuitive, our study is the first to rigorously investigate, and quantify,
this particular association; thus providing a methodological contribution in customer engagement
research (Brodie et al. 2011, Hollebeek et al. 2014, Leeflang 2011).
The second contribution is that we conceptualize, and empirically investigate, the novel term
of customer disengagement. Specifically, while our findings indicated greater mobile app-related
CEBs, typically, to be beneficial to firms, specific form of app-related customer disengagement,
by contrast, was shown to have the potential to engender harmful effects for organizations and
their marketing programs. To illustrate, while customers may feel engaged with a focal app in
some respects (e.g., high perceived informational value of the app), individuals may simultaneously
feel disengaged with the app in others (e.g., low user friendliness; app fails to provide the desired
features, does not work on consumer’s phone, or did not perform as expected, etc.). When an
app’s perceived disengaging features debilitate the user from achieving her goals, we expect the
emergence of specific app-related customer disengagement, which ultimately lead to the customer’s
discontinued use of the app. Furthermore, the study finds that customer disengagement has a
sizeable long-term effect on purchase behaviors. This is an interesting and important result for
the marketing community. Given the absence of empirically-based insights into these dynamics to
date, our work provides a seminal contribution.
In addition to the stated scholarly contributions and implications, this research also offers a
number of managerial implications. First, our investigation is important because if the association
between specific mobile app-related CEBs, and their ensuing point accumulation and redemption
would not be significant, firms should cease, or at least reduce, investing their marketing resources in
the development of mobile apps designed for the purpose of stimulating specific CEBs. Conversely,
if a significant relationship between the stated dynamics does exist, CEBs not only provide a
powerful tool for the strategic development of customer loyalty, but can also reduce spending on
traditional promotional forms, including advertising (given mobile apps, typically, are relatively
inexpensive to run subsequent to their initial development cost).
Drawing on dynamic models deploying actual customer engagement, purchase and redemption
behaviors, we obtain evidence for the latter; that is, for the existence of a positive relationship
between mobile app-related CEBs and point accumulation and redemption; thus indicating a high
strategic importance of firms’ development and adoption of branded mobile apps. Specifically, our
findings provide evidence for a potential ‘game changing’ nature regarding the specific ways for
Vishvanathan et al.: Customer Engagement with Mobile Apps
18 Service Science 00(0), pp. 000–000, c
0000 INFORMS
firms to engage their customers, thus directly addressing firms’ core strategic decision-making on
their app-related, and broader mobile, strategies. We expect this research to provide important
insights facilitating managers’ decision-making regarding the development of specific new media
strategies and tactics. Specifically, our research may help practitioners answer questions such as
‘given the heightened number of brand touch points resulting from customers’ adoption of focal
branded mobile apps, to what extent do their specific app-related CEBs affect ensuing purchase
behaviors over time?’ and ‘how do customer disengagement with a branded mobile app affect
ensuing purchase behaviors?’
Our findings, broadly, provide a cautionary note for managers undertaking traditional adver-
tising investments into their brands, which not only typically represent a relatively costly form
of marketing communications lacking personalization, but also, are predominantly based on an
assumed one-way transmission of the promotional message from the firm to (prospective) cus-
tomers culminating, ultimately, in the attainment of particular market outcome metrics. However,
in the emerging digital era, (prospective) customers and firms are increasingly intertwined in sets
of dynamic, interactive two-way exchanges and the undertaking of specific CEBs and customer dis-
engagement (Hoffman and Novak 2012). As such, the ways to optimally execute specific marketing
communications is changing. Hence by providing managers with new strategic insights regarding
key dynamics characterizing particular CEBs and customer disengagement, our findings provide an
impetus for practitioners to re-assess the effectiveness of their specific strategic brand touch points
in driving important organizational objectives, including customer acquisition and retention.
Further, we also find customers’ point accumulation and redemption to affect their subsequent
use of the branded mobile app; thus providing a further managerial contribution of our work. Based
on this finding, we highlight the importance of firms’ strategic adoption of integrated marketing
communications (IMC) exhibiting synergy or consistency across organizational marketing commu-
nications, including those employed at point of purchase, as well as those related to the adoption
of specific mobile apps. Finally, we find that mobile app-related customer disengagement exert a
negative effect on customers’ future purchases with the firm. A major implication arising from this
finding is that firms need have a clear understanding of specific customers’ (or customer segments’)
perceived engaging and disengaging attributes of specific mobile apps prior to launching the app,
as well as undertake careful monitoring regarding the performance of particular apps.
5.2. Limitations and Future Research
While the study makes important contributions, we acknowledge it is also subject to certain lim-
itations. First, our measures of CEBs are limited to addressing customers’ specific behavioral
expressions of engagement (I don’t think these cites are needed.) (Jaakkola and Alexander 2014,
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 19
Van Doorn et al. 2010, Verleye et al. 2013), thus rendering customer manifestations of their specific
cognitive and emotional engagement with our studied mobile app largely implicit (Brodie et al.
2011). While the study of customers’ engagement behaviors has provided highly valuable findings,
future research may wish to more explicitly incorporate specific cognitive and emotional engage-
ment facets in their research designs (Hollebeek et al. 2014). (I don’t think this cite is needed.)
Second, while we investigate engagement and disengagement with our selected app, it is impor-
tant to understand and isolate their drivers (Brodie et al. 2011, Malthouse and Calder 2011). To
illustrate, the cognitive and emotional drivers of engagement may differ, conceptually, from those
that drive disengagement. For example, while engagement may occur based on a high perceived
entertainment value of a focal app, disengagement may result from customers’ perceived difficulty
in operating the app (as stated). It is therefore necessary to better understand the specific drivers
of both to design more effective apps in the future.
Third, the focal firm of this study did not carry out any significant marketing campaigns during
the time period that we had data for. However, future studies could include variables pertaining to
firms’ specific advertising and communications campaigns at different points in time and examine
how these variables affect engagement with the mobile app.
Fourth, our investigation was limited to a single branded mobile app. More research is required
that investigates mobile app-related behaiors in different contexts, including within or across partic-
ular product categories, brands or different countries or cultures. Specifically, based on consumers’
distinct cross-cultural preferences for interacting with particular promotional content (Nakata
2009), their mobile app-related preferences are also expected to differ across cultures, although
insights into this area remain nebulous in the literature to date. Although we provide an initial
investigation into disengagement, future studies should extend and validate our present findings
across other contexts.
Fifth, while we explored specific CEBs with a focal branded mobile app, insights regarding CEBs
with other emerging digital touch points and tools, including social media, QR codes and specific
wearable components (e.g., Google Glass), as well as the integration of focal mobile apps and other
touch points, remain limited to date. Future research may therefore wish to investigate the nature
and dynamics characterizing these developments.
Despite these limitations, this study helps us further our understanding of the relationship
between customer (dis)engagement with emerging digital platforms such as mobile apps and behav-
iors that occur in the marketplace. The results from a longitudinal analysis of a unique dataset
unearth interesting and important insights on (dis)engagement and its relationship with other
brand-and purchase-related behaviors. This study is therefore an ideal stepping stone for future
Vishvanathan et al.: Customer Engagement with Mobile Apps
20 Service Science 00(0), pp. 000–000, c
0000 INFORMS
work on (dis)engagement and its role in influencing the relationship between firms and their con-
sumers.
Acknowledgments
The authors gratefully acknowledge the existence of the Journal of Irreproducible Results and the support
of the Society for the Preservation of Inane Research.
References
Ari˜no MA, Franses PH (2000) Forecasting the levels of vector autoregressive log-transformed time series.
International Journal of Forecasting 16(1):111–116.
Bellman S, Potter RF, Treleaven-Hassard S, Robinson JA, Varan D (2011) The effectiveness of branded
mobile phone apps. Journal of Interactive Marketing 25(4):191–200.
Blau P (1964) Exchange and Power in Social Life (New York: Wiley).
Brodie RJ, Hollebeek LD, Juri´c B, Ili´c A (2011) Customer engagement: Conceptual domain, fundamental
propositions, and implications for research. Journal of Service Research 14(3):252–271.
Brodie RJ, Ili´c A, Juri´c B, Hollebeek LD (2013) Consumer engagement in a virtual brand community: An
exploratory analysis. Journal of Business Research 66(1):105–114.
Calder BJ, Malthouse EC (2008) Media engagement and advertising effectiveness. Kellogg on advertising
and media, 1–36 (New York: Wiley).
De Valck K, Bruggen GHV, Wierenga B (2009) Virtual communities: A marketing perspective. Decision
Support Systems 47(3):185–203.
Fox Z (2013) Global app downloads to pass 100 billion this year, http://mashable.com/2013/09/19/app-
downloads-100-billion/?utm cid=mash-com-fb-main-photo (Accessed June 1, 2014).
Granger C (1969) Investigating causal relations by econometric models and cross-spectral methods. Econo-
metrica: Journal of the Econometric Society 37(3):424–438.
Granger C, Newbold P (1974) Spurious regressions in econometrics. Journal of Econometrics 2(2):111–120.
Gredenhoff M, Karlsson S (1999) Lag-length selection in var-models using equal and unequal lag-length
procedures. Computational Statistics 14(2):171–187.
Gruman G (2011) Where mobile technology is heading, from 2011 to 2020. InfoWorld
Http://www.infoworld.com/d/mobilize/where-mobile-technology-heading-2011-2020-402 (Accessed
May 31, 2014).
Gummerus J, Liljander V, Weman E, Pihlstr¨om M (2012) Customer engagement in a facebook brand com-
munity. Management Research Review 35(9):857–877.
Hanssens LJP Dominique M, Schultz RL (2001) Market Response Models: Econometric and Time Series
Analysis, volume 12 (New York: Springer).
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 21
Hoffman DL, Fodor M (2010) Can you measure the roi of your social media marketing? MIT Sloan Manage-
ment Review 52(1):41–49.
Hoffman DL, Novak TP (1996) Marketing in hypermedia computer-mediated environments: Conceptual
foundations. Journal of Marketing 60:50–68.
Hoffman DL, Novak TP (2012) Toward a deeper understanding of social media. Journal of Interactive
Marketing 26(2):69–70.
Hollebeek LD (2011) Demystifying customer brand engagement: Exploring the loyalty nexus. Journal of
Marketing Management 27(7–8):785–807.
Hollebeek LD (2011b) Exploring customer brand engagement: Definition & themes. Journal of Strategic
Marketing 19(7):555–573.
Hollebeek LD (2012) The customer engagement/value interface: An exploratory investigation. Australasian
Marketing Journal 21(1):17–24.
Hollebeek LD, Glynn MS, Brodie RJ (2014) Consumer brand engagement in social media: Conceptualization,
scale development and validation. Journal of Interactive Marketing 28(2):149–165.
Im KS, Pesaran MH, Shin Y (2003) Testing for unit roots in heterogeneous panels. Journal of Econometrics
115(1):53–74.
Jaakkola E, Alexander M (2014) The role of customer engagement behavior in value cocreation: A service
system perspective. Journal of Service Research In Press:DOI: 10.1177/1094670514529187.
Kaplan AM, Haenlein M (2010) Users of the world, unite! the challenges and opportunities of social media.
Business Horizons 53(1):59–68.
Kim E, Lin JS, Sung Y (2013) To app or not to app: Engaging consumers via branded mobile apps. Journal
of Interactive Advertising 13(1):53–65.
Kim SJ, Wang RJH, Malthouse EC (2015) The effects of adopting and using a brand’s mobile application
on customers’ subsequent purchase behavior. Journal of Interactive Marketing 31:28–41.
Larivi`ere B, Joosten H, Malthouse EC, van Birgelen M, Aksoy P, Kunz WH, Huang MH (2013) Value fusion:
the blending of consumer and firm value in the distinct context of mobile technologies and social media.
Journal of Service Management 24(3):268–293.
Leeflang P (2011) Paving the way for distinguished marketing. International Journal of Research in Marketing
28(2):76–88.
Malthouse EC, Calder BJ (2011) Engagement and experiences: Comment on brodie, hollebeek, juric, and
ilic (2011). Journal of Service Research 14(3):277–279.
Mangold WG, Faulds DJ (2009) Social media: The new hybrid elements of the promotion mix. Business
Horizons 52(4):357–365.
Vishvanathan et al.: Customer Engagement with Mobile Apps
22 Service Science 00(0), pp. 000–000, c
0000 INFORMS
Men LR, Tsai WHS (2013) Beyond liking or following: Understanding public engagement on social networking
sites in china. Public Relations Review 39:13–22.
MSI (2014) 2014–2016 marketing science institute research priorities, http://www.msi.org/research/2014-
2016-research-priorities/ (Accessed May 9, 2014).
Muntinga DG, Moorman M, Smit EG (2011) Introducing cobras: Exploring motivations for brand-related
social media use. International Journal of Advertising 30(1):13–46.
Nakata C (2009) Beyond Hofstede: Culture Frameworks for Global Marketing and Management (London:
Palgrave Macmillan).
Nambisan S, Baron RA (2007) Interactions in virtual customer environments: Implications for product
support and customer relationship management. Journal of Interactive Marketing 21(2):42–62.
Nielsen (2014) The digital consumer, http://www.nielsen.com/content/corporate/us/en/insights/reports/2014/the-
us-digital-consumer-report.html (Accessed July 7, 2014).
O’Brien HL, Toms EG (2008) What is user engagement? a conceptual framework for defining user engagement
with technology. Journal of the American Society for Information Science and Technology 59(6):938–
955.
One L (2015) Our history, from https://www.loyalty.com/about-us/our-history.html.
Palmatier RW, Jarvis CB, Bechkoff JR, Kardes FR (2009) The role of customer gratitude in relationship
marketing. Journal of Marketing 73(5):1–18.
Peck A, Malthouse EC (2011) Medil l on Media Engagement (Creskill, NJ: Hampton Press).
Prahalad CK, Ramaswamy V (2004) Co-creation experiences: The next practice in value creation. Journal
of Interactive Marketing 18(3):5–14.
Report UF (2011) Mobile traffic forecasts 2010-2020 report, http://www.umts-
forum.org/component/option,com docman/task,cat view/gid,485/Itemid,213/ (Accessed June 3,
2014).
Rust RT, Lemon KN, Zeithaml VA (2004) Return on marketing: Using customer equity to focus marketing
strategy. Journal of Marketing 68(1):109–127.
Sawhney M, Verona G, Prandelli E (2005) Collaborating to create: The internet as a platform for customer
engagement in product innovation. Journal of Interactive Marketing 19(4):4–17.
Schadler T, McCarthy JC (2012) Mobile is the new face of engagement, http://cdn.blog-
sap.com/innovation/files/2012/08/SAP Mobile Is The New Face Of Engagement.pdf (Accessed June
9, 2014).
Sims CA (1980) Maeroeconomics and reality. Econometrica 48:1–48.
Sims CA, Zha T (1999) Error bands for impulse responses. Econometrica 67(5):1113–1155.
Vishvanathan et al.: Customer Engagement with Mobile Apps
Service Science 00(0), pp. 000–000, c
0000 INFORMS 23
Srinivasan S, Pauwels K, MHanssens D, Dekimpe MG (2004) Do promotions benefit manufacturers, retailers,
or both? Management Science 50(5):617–629.
Trusov M, Bucklin RE, Pauwels K (2009) Effects of word-of-mouth versus traditional marketing: Findings
from an internet social networking site. Journal of Marketing 73:90–102.
Van Doorn J, Lemon KN, Mittal V, Naß S, Pick D, Pirner P, Verhoef PC (2010) Customer engagement
behavior: Theoretical foundations and research directions. Journal of Service Research 13(3):253–266.
Van Ittersum K, Wansink B, Pennings JM, Sheehan D (2013) Smart shopping carts: How real-time feedback
influences spending. Journal of Marketing 77(6):21–36.
Vargo SL, Lusch RF (2004) Evolving to a new dominant logic for marketing. Journal of Marketing 68(1):1–17.
Vargo SL, Lusch RF (2008) Service-dominant logic: Continuing the evolution. Journal of the Academy of
Marketing Science 36(1):1–10.
Verleye K, Gemmel P, Rangarajan D (2013) Managing engagement behaviors in a network of customers and
stakeholders: Evidence for the nursing home sector. Journal of Service Research 17(1):68–84.
Wang B, Kim S, Malthouse EC (2016) Branded apps and mobile platforms as new tools for advertising. The
New Advertising: Branding, Content, and Consumer Relationships in the Data-driven Social Media Era
(ABC-CLIO Santa Barbara, CA).
Wang RJH, Malthouse EC, Krishnamurthi L (2015) On the go: How mobile shopping affects customer
purchase behavior. Journal of Retailing 91(2):217–234.
Wieringa JE, Horvth C (2005) Computing level-impulse responses of log-specified var systems. International
Journal of Forecasting 21(2):279–289.
Wiertz C, Ruyter KD (2007) Beyond the call of duty: Why customers contribute to firm-hosted commercial
online communities. Organization Studies 28(3):347–376.