Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Journal of Marketing Channels

ISSN: 1046-669X (Print) 1540-7039 (Online) Journal homepage: https://www.tandfonline.com/loi/wjmc20

Omnichannel approach: Factors affecting


consumer acceptance

Susana Costa e Silva, Carla Carvalho Martins & João Martins de Sousa

To cite this article: Susana Costa e Silva, Carla Carvalho Martins & João Martins de Sousa
(2019): Omnichannel approach: Factors affecting consumer acceptance, Journal of Marketing
Channels, DOI: 10.1080/1046669X.2019.1647910

To link to this article: https://doi.org/10.1080/1046669X.2019.1647910

Published online: 20 Aug 2019.

Submit your article to this journal

Article views: 15

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=wjmc20
JOURNAL OF MARKETING CHANNELS
https://doi.org/10.1080/1046669X.2019.1647910

Omnichannel approach: Factors affecting consumer acceptance


Susana Costa e Silva, Carla Carvalho Martins, and Jo~ao Martins de Sousa
Catolica Porto Business School, Universidade Catolica Portuguesa, Porto, Portugal

ABSTRACT KEYWORDS
The omnichannel approach is a new commercial opportunity that aims to offer consumers a Omnichannel; consumer;
unique and satisfying experience through any touch point. This research aims to understand intention to use; brand
the factors that might lead consumers to accept and use this new approach. The conceptual experience; satisfaction
model was tested empirically using primary data collected from 210 Portuguese participants.
The results reported suggest that perceptions of usefulness, ease of use, and compatibility
have a positive effect on brand experience, which strongly affects the behavioral intention
to use. Additionally, a moderation analysis, using user status as a moderator, shows that
while usefulness is important to uniquely explain use intention for low frequency, ease of
use is only a good predictor of use intention for high-frequency users. This research pro-
vides useful insights for academic research by shedding light on this new phenomenon,
and simultaneously for businesses by deriving some implications for defining their omni-
channel strategies.

Introduction for the implementation of an omnichannel strategy


(Piotrowicz & Cuthbertson, 2014). Thus, retailers that
The world of retailing has changed dramatically in
make it as easy as possible for customers to interact
recent decades. The advent of the Internet, along with
related new digital channels, such as mobile and social with the brand seamlessly – without operational or
media, have changed retail business models, the exe- emotional barriers, integrating experiences across
cution of the retail mix, and shopper behavior touchpoints, creating personalized interactions, and
(Verhoef, Kannan, & Inman, 2015). Consumers use delivering timely communications – will be in the best
multiple means to connect with their companies of position to succeed (Frazer & Stiehler, 2014; Fulgoni,
choice and expect a consistent and integrated service 2014; Piotrowicz & Cuthbertson, 2014; Rey-Moreno &
experience across channels. They are willing to move Medina-Molina, 2016; Verhoef et al., 2015).
seamlessly between channels – traditional store, As a recent phenomenon, the emergence of omni-
online, and mobile – depending on their preferences, channel retailing raises opportunities for academics to
their current situation, the time of day, and the prod- produce insights that can help address challenges
uct category. Such channel movement is known as the faced by retailers. Because companies may achieve
omnichannel approach. Consumers expect that con- competitive advantage by adopting an omnichannel
tact that begins on one interaction channel can be approach, consumer intentions to use this kind of
continued on another, with all the relevant contextual commerce approach are of the utmost interest to
data preserved across channels (Piotrowicz & researchers and practitioners. Therefore, this study
Cuthbertson, 2014). This has created significant chal- strives to understand the key factors underlying con-
lenges for retailers (Frazer & Stiehler, 2014; Neslin sumers’ intentions to use an omnichannel approach.
et al., 2014; Verhoef et al., 2015). Even if a retailer More specifically, this article examines the influence
offers multichannel access, online and traditional of compatibility, cost, perceived risk, perceived useful-
channels are often treated separately, with a lack of ness, and perceived ease of use on behavioral inten-
integration in pricing, promotion, marketing, brand tion to use and on actual use. Moreover, it is expected
building, supply chain management, and experience that different user groups (high-frequency vs.
provided across all channels; this creates difficulties low-frequency users) may be influenced by different

CONTACT Susana Costa e Silva ssilva@porto.ucp.pt Cat


olica Porto Business School, Universidade Cat
olica Portuguesa, Rua Diogo Botelho, 1327,
4169-005 Porto, Portugal.
This article was accepted under the editorship of Neil Herndon.
ß 2019 Taylor & Francis Group, LLC
2 S. C. SILVA ET AL.

factors regarding omnichannel behavior adoption. consumers and talk to the brand using social media
Therefore, the present study also examines whether (Neslin et al., 2006). Moreover, powered by mobile
user status is a moderator of the relationships between and social media, customers not only move between
the abovementioned factors and the adoption of channels but also use these channels at the same time.
omnichannel behavior. For instance, customers may search for information
This research is structured as follows: in the next online while being in the store in order to obtain
section, the theoretical background of the research more information (Verhoef et al., 2015).
is presented, along with the development of specific In this context, omnichannel can be defined as an
research hypotheses. This is followed by a descrip- emerging approach to channel integration that aims
tion of the research methodology and results. After to deliver seamless customer experience through every
presenting the discussion and implications of the possible channel a retailer or brand can connect with
results, the paper concludes with some managerial their end consumers. It entails the ability to offer cus-
implications, limitations, and suggestions for fur- tomers a continuous experience across brands, for-
ther research. mats, and devices (Bell, Gallino, & Moreno, 2014;
Frazer & Stiehler, 2014; Fulgoni, 2014; Verhoef et al.,
Theoretical background and theoretical model 2015). Building a truly omnichannel context is not a
straightforward task, and requires that brands be pre-
The omnichannel approach pared to reorganize their supply chains and their mar-
As recently as 2002, Schoenbachler and Gordon keting processes to meet their customers’ needs
asserted that: “Traditional retail businesses and even (Neslin et al., 2006).
catalog marketers today face a critical decision - to
accept a new, yet unrefined business model including e- Conceptual model: Factors of adoption of the
commerce or to retain their old business model and omnichannel approach
risk becoming obsolete and left behind by new, web-
based competitors” (Schoenbachler & Gordon, 2002, p. A common element underlying most of the multiple
43) Now, the multichannel approach is no longer an models that intend to explain the adoption of innova-
unrefined business model, and choosing to adopt it or tions is the inclusion of perceptions about innovation
not is no longer an issue. We are now in the omni- as key independent variables. For instance, the
channel retailing era. Omnichannel retailing must be Technology Acceptance Model (TAM) (Davis, 1989;
perceived as an evolution of multichannel retailing Davis, Bagozzi, & Warshaw, 1989) and its precursor,
(Verhoef et al., 2015). Multichannel retailing primarily the theory of reasoned action (Ajzen & Fishbein,
arose as a consequence of the growth of the Internet; 1980), both postulate that perceptions or beliefs about
it has been defined as a distribution strategy to serve innovation are fundamental in the development of
customers using more than one selling channel or attitudes that eventually result in system utilization
medium, implying a division between the physical and behavior (Agarwal & Prasad, 1998). According to
online channels (Frazer & Stiehler, 2014). Due mostly TAM, the adoption behavior is determined by the
to mobile and social media growing as important intention to use a particular technology, which is, in
channels of contact between consumers and brands, turn, determined by the perceived usefulness and the
natural borders between physical and online channels perceived ease of use of that technology. Two similar
are now progressively disappearing (Piotrowicz & concepts were already present in Rogers’ (1983) theory
Cuthbertson, 2014). Rigby (2011, p. 75) was the first of diffusion of innovations: relative advantage (close
to use the expression “omnichannel retailing” as an to usefulness), which is the degree to which an innov-
integrated sales experience that melds the advantages of ation is perceived as being better than the idea it
physical stores (such as personal service, the ability to replaces; and complexity (close to ease of use, but in
touch products, and shopping as an experience) with the opposite direction), which is the degree of diffi-
the information-rich experience of online shopping. An culty that the user has in using and understanding an
“omnishopper” moves freely between channels. He or innovation. The variables of this model have been
she may explore the characteristics of a product using widely applied to explain the adoption of systems in a
a mobile app, compare prices on several websites large number of contexts, namely search engines (e.g.
from his or her laptop, and, finally, buy the product Morgan-Thomas & Veloutsou, 2013), e-commerce
at a physical store. Later, he or she may share opin- (e.g. Gefen, Karahanna, & Straub, 2003; Pavlou, 2003;
ions and experiences about the product with other Rose, Clark, Samouel, & Hair, 2012), and mobile
JOURNAL OF MARKETING CHANNELS 3

commerce (m-commerce) (e.g. Kim, Mirusmonov, & Hypothesis 3(b): Perceived compatibility has a
Lee, 2010; Wu & Wang, 2005). Although the omni- positive direct effect on perceived usefulness of the
channel approach is much more than a technology, its omnichannel approach.
foundations lie in the capability of using several tech- Individuals make behavioral choices by weighing
nologies, often during a unique purchase. Thus, it is up the costs and benefits of those behaviors. To adopt
expected that the adoption of an omnichannel an omnichannel approach, consumers must deal with
approach should, at least in part, be explained by the nonnegligible financial costs (hereafter referred to as
variables therein. Based on this, the following hypoth- cost) such as equipment costs, access costs, and trans-
eses are proposed: action fees (Wu & Wang, 2005). Hence, the following
Hypothesis 1: Perceived usefulness has a positive hypothesis is presented:
direct effect on behavioral intention to use an Hypothesis 4: Cost has a negative direct effect on
omnichannel approach. behavioral intention to use.
Hypothesis 2(a): Perceived ease of use has a positive Adopting an omnichannel approach entails engag-
direct effect on behavioral intention to use an
ing in online activities and transactions. Since the use
omnichannel approach.
of a website for transactions involves a certain degree
Original TAM research has shown that usefulness of uncertainty, people are still anxious about diverse
mediates the relationship between ease of use and types of risks. Referring to e-commerce, Biswas and
technology usage (Davis, 1989; Davis et al., 1989). Biswas (2004) stressed the particular importance of
This link has also been found in more recent studies three types of risks: performance risk, which is the
(e.g. Venkatesh & Davis, 2000; Wu & Wang, 2005). uncertainty and the consequence of a product not
Therefore, the following hypothesis is added to the functioning at some expected level; financial risk,
conceptual model. which is the possibility of monetary losses; and trans-
Hypothesis 2(b): Perceived ease of use has a positive action/privacy risk, which is uncertainty associated
direct effect on perceived usefulness of an with providing information (such as credit card num-
omnichannel approach. ber, name, address, etc.) to a seller during the course
Rogers (1983) added to the previously mentioned of a transaction. In this respect, the omnichannel
relative advantage and complexity other factors that approach brings about new elements to this equation,
could explain the adoption of innovations, such as since it allows customer behavior to be tracked in a
perceived compatibility, defined as the degree to which way that has never been possible before. Therefore,
an innovation is perceived as consistent with the exist- customers are increasingly concerned about who
ing values, past experiences, habits, lifestyle, and needs owns, stores, controls, and shares information about
of potential adopters. Previous studies have shown them (Piotrowicz & Cuthbertson, 2014). Perceived
that compatibility is an antecedent of both e-com- risk has already been proven to have a negative
merce and m-commerce use intention (e.g. Hsi-Peng impact on attitudes towards online transactions (e.g.
& Su, 2009; Lin, 2007; Wu & Wang, 2005). Based on Glover & Benbasat, 2010; Lopez-Nicolas & Molina-
this, we propose the following hypothesis: Castillo, 2008; Pavlou, 2003), namely through mobile
applications (Wu & Wang, 2005). Therefore, we pro-
Hypothesis 3(a): Perceived compatibility has a
positive direct effect on behavioral intention to use an pose the following hypothesis:
omnichannel approach. Hypothesis 5: Perceived risk has a negative direct
Compatibility has been frequently addressed as an effect on behavioral intention to use an
omnichannel approach.
independent antecedent of innovation adoption.
However, the assumption that individuals are unlikely Several social-psychological models, such as the
to view a specific system as useful if it is not compat- theory of reasoned action (Ajzen & Fisbbein, 1974),
ible with the manner in which they prefer to work the theory of planned behavior (Ajzen, 1985), and
has led some researchers to hypothesize and demon- Triandis’ (1980) attitude-behavior theory, propose that
strate the positive effects of compatibility on perceived the most important predictor of a person’s behavior is
usefulness (e.g. Karahanna, Agarwal, & Angst, 2006; his or her intention to perform it. Multiple studies on
Wu & Wang, 2005). In this way, it is assumed that technology acceptance and usage have validated this
compatibility has both a direct effect on use intention connection (e.g. Davis et al., 1989; Venkatesh &
and an indirect effect through the mediation of useful- Davis, 2000). Although it is assumed that, regarding
ness. Therefore, we hypothesize that: the omnichannel approach, some consumers with a
4 S. C. SILVA ET AL.

capturing the desired phenomenon and to ensure that


no important aspects had been omitted.
The survey instrument comprised three sections.
The first section was devoted to socio-demographic
variables. The second section assessed the use of dif-
ferent channels across the five phases of the customer
journey – stimulation, search for information, pur-
chase, delivery, and aftersales service – adopting
Delft’s (2013) instrument. The third section included
all the scales used to operationalize each construct
that compose the conceptual model proposed. As
Figure 1. Conceptual model. these were based on previous literature and thereby
built to be used in different research contexts, namely
high intention to use might still not have actually mobile commerce (Wu & Wang, 2005), they had to
tried it, generally, the following is expected: undergo some adaptations (see Appendix). In all
cases, five-point Likert scales, in which 2 indicates
Hypothesis 6: Use intention has a positive direct
effect on the actual use of the omnichannel approach. “completely disagree” and 2 “completely agree,” were
used to measure the variables included in the theoret-
Finally, it is expected that the influence of the dif- ical model.
ferent determinants on the adoption of the omnichan- In order to clarify the concept of the omnichannel
nel approach should vary according to the status of approach we referred, in the survey, to the fact that
the user. Thus, we propose the following hypothesis: there is no consensual definition of the phenomenon,
Hypothesis 7(a–f): User status moderates the but that it is assumed to refer to the hybrid use, with
influence of the determinants – risk, cost, no barriers, of the different available channels of
compatibility, usefulness, and ease of use – on future interaction between companies and consumers, in
use intention, as well as the impact of the future use
intention on the actual use. digital (e-mail, websites, social media networks, apps,
among others, regardless of whether accessed through
On the basis of the hypotheses, we propose the smartphones, tablets, or laptops), physical, or mixed
model of Figure 1. environments (phone, stores, catalogs, flyers, etc.). In
addition, we stated that one of the main assumptions
Research methodology of this concept concerns the possibility of simultan-
eous use of different channels “24 hours per day,
Considering the purpose of the study, a quantitative
7 days a week, everywhere,” with the objective of pre-
cross-sectional research strategy was used (Bryman &
senting a unified image of the firm to its customers.
Bell, 2015). Primary data was collected through a
Before answering the questions concerning percep-
web-based survey administered to a sample of
tions of the omnichannel approach, the respondents
Internet users recruited through a nonprobabilistic
were given a practical example of an omnichannel
sampling process, by convenience, using the social
experience using the customer journey of purchasing
media platform Facebook. Researchers asked their
a scarf.
social network friends to participate and spread the
questionnaire via their contacts. The data collection
occurred mostly in Portugal over a period of 6 weeks, Analysis and results
but some data were obtained from emigrant
Sample characterization and the use of channels
Portuguese nationals living in several countries. The
across the customer journey
questionnaire was subjected to a multi-stage develop-
ment process that included the translation and adapta- The sample comprised 106 females and 84 males, with
tion of various scales to Portuguese reality and to the an average age of 36-years-old. Most of the partici-
specific object of study – the omnichannel approach. pants were aged between 18–25 (31.6%) and 26–35
Once the initial questionnaire had been generated, an (25.2%). About 84% of all participants had completed
interactive personal interview process was conducted higher education and 58.4% were full-time workers.
to refine the instrument. These interviews enabled the In addition, 46.8% said that when they make pur-
researchers to assess whether the instrument was chases, they frequently or very frequently use an
JOURNAL OF MARKETING CHANNELS 5

Table 1. Sample characterization.


Total sample Low frequency users (n ¼ 101) High frequency users (n ¼ 89)
Gender
Female 106 55.8% 61 45
Male 84 44.2% 40 44
Age (average ¼ 36.03) (average ¼ 36.19) (average ¼ 35.85)
18–25 60 31,6% 36 24
26–35 48 25,3% 19 29
36–45 26 13,7% 14 12
46–55 34 17,9% 20 14
56–65 20 10,5% 10 10
Over 65 2 1,1% 2 0
Education level
Under high school 6 3.2% 5 1
Senior high school 24 12.6% 16 8
College 6 3.2% 3 3
Graduate 81 42.6% 42 39
Post-graduate 73 38.5% 35 38
Work situation
Full-time worker 114 60.0% 55 59
Part-time worker 9 4.7% 3 6
Unemployed 17 8.9% 12 5
Retired 2 1.1% 2 0
Full-time student 29 15.3% 20 9
Working student 18 9.5% 8 10
Never worked/domestic 1 0.5% 1 0
Omnichannel frequency of use
2-Never/almost never 9 4.7% 9 0
4-Sporadically 34 17.9% 34 0
6-Sometimes 58 30.5% 58 0
8-Frequently 61 32.1% 0 61
5-Very frequently/Always 28 14.7% 0 28

omnichannel approach. More detailed information on that high-frequency users use websites, online stores,
the respondents can be found in Table 1. and mobile applications significantly more often than
Table 2 illustrates the use of different channels do low-frequency users across all phases of the cus-
across the customer journey, considering both the tomer journey. It is, however, worth noting that the
total sample and the sample split into two different groups do not present statistically significant differen-
groups. Participants were assigned to each group ces with respect to the use of social media in any of
according to their (self-reported) frequency of use of these phases. Both seem to make little use of this
the omnichannel approach. Participants (n ¼ 89) that channel during their customer journeys.
answered 4 ¼ frequently or 5 ¼ very frequently/always
to the question “How often do you engage in pur-
Scale reliability and validity
chases using an omnichannel approach?” (the answer
to which was mandatory) were assigned to the group We conducted initial reliability analyses of all the
of high-frequency users. The remaining participants scales used in the third part of the questionnaire using
(n ¼ 101), who answered with lower values, were Cronbach’s alpha in SPSS software. All the scales
assigned to the group of low-frequency users. showed Cronbach’s alpha values above 0.70, indicating
The results of this analysis show that online chan- acceptable internal consistency for the evaluation of
nels have become particularly important in the phase the measured constructs (Hair, 2006). Moreover, in
of information search. Websites and online stores are order to guarantee measurement reliability and valid-
the most frequently used channels for that purpose. ity, and using AMOS 23.0 software, we carried out a
Nevertheless, in the other phases of the customer confirmatory factor analysis, applying structural equa-
journey traditional channels prevail. There is an espe- tion modeling with maximum likelihood estimation
cially strong preponderance of the use of brick-and- method (Byrne, 2013; Hair, 2006). Table 3 provides
mortar stores in the phase of buying. the results of this measurement analysis.
When we compare the profile of high-versus low- The values for composite reliability, computed on
frequency users of the omnichannel approach, differ- the basis of standardized loadings, were all above Hair
ences can be found in every stage of the customer (2006) recommended minimum level of 0.70, corrobo-
journey, but only regarding the utilization of elec- rating the existence of internal consistency. The obser-
tronic channels. Independent samples t-tests showed vation of standardized loadings supports the existence
6 S. C. SILVA ET AL.

Table 2. Utilization of channels across the customer journey.


Channels used across Low freq. users t-test for the difference
customer journey phases Total sample mean (n ¼ 101) mean High freq. users (n ¼ 89) mean between groups
Stimulation
Websites or online stores 2.96 2.52 3.45 7.21
Mobile Apps 2.16 2.01 2.33 2.17
Social networking sites 2.37 2.26 2.49 1.50
Bricks and mortar shop 3.89 3.89 3.89 0.03
Mass media 2.70 2.68 2.72 0.26
Catalogs 1.48 2.67 2.85 1.41
Family and friends 3.19 3.21 3.18 0.23
Search for information
Websites or online stores 4.00 3.77 4.26 3.49
Mobile Apps 2.57 2.32 4.26 3,15
Social networking sites 2.48 2.48 2.48 0.05
Bricks and mortar shop 3.69 3.67 3.72 0.33
Mass media 2.38 2.47 2.28 1.28
Catalogs 2.71 2.74 2.66 0.54
Family and friends 3.14 3.18 1.09 0.77
Buy
Websites or online stores 2.72 2.25 3.26 7.49
Mobile Apps 1.86 1.62 2.13 3.84
Social networking sites 1.64 1.67 1.61 0.53
Bricks and mortar shop 3.98 3.97 3.99 0.13
Catalogs 1.98 1.99 1.97 0.16
Delivery
Buy in a shop and take 4.28 4.34 4.22 1.11
home directly
Buy in a shop and receive 1.96 1.93 1.99 1.11
at home or elsewhere
Buy online and receive at 2.84 2.36 3.38 7.17
home or elesewhwere
Buy online and reveive at a 1.79 1.59 2.01 3.16
pick up point
Service
Websites or online stores 2.60 2.29 2.94 4.24
Mobile Apps 1.45 1.35 1.57 1.99
Social networking sites 1.60 1.59 1.61 0,10
E-mail 2.87 3.02 3.34 3.13
Bricks and mortar shop 3.94 4.03 3.83 1.43
Telephone 3.17 3.02 3.34 1.99
Differences between low frequency and high frequency users, with t-tests. Scale 1 to 5, where 1 is never or almost never and 5 is very
frequently or always. p < .01; p < .05.

Table 3. Results of confirmatory factor analysis.


Construct Standardized loading Cronbach’s alpha Composite reliability Average variance extracted
Risk 0.831 0.756 0.771 0.464
0.723
0.600
0.533
Cost 0.822 0.698 0.714 0.466
0.702
0.478
Compatibility 0.903 0.839 0.852 0.664
0.888
0.622
Usefulness 0.863 0.894 0.897 0.636
0.818
0.803
0.785
0.711
Perceived ease of use 0.901 0.891 0.895 0.740
0.896
0.779
Intention to use 0.889 0.894 0.893 0.736
0.853
0.830

of convergent validity since only one construct – cost theoretical relevance), which is generally the min-
– contained an item whose weight was slightly below imum recommended threshold (Hair, 2006). In terms
0.50 (though this was not excluded given its of variance extracted, the majority of constructs also
JOURNAL OF MARKETING CHANNELS 7

Table 4. Squared correlations between constructs.


Risk Cost Compatibility Usefulness Ease of use Future use intention
Alpha (0.464)
Alpha 0.359 (0.466)
Alpha 0.304 0.071 (0.664)
Alpha 0.235 0.104 0.618 (0.636)
Alpha 0.107 0.063 0.347 0.320 (0.740)
Alpha 0.305 0.076 0.746 0.729 0.430 (0.736)
Average variance extracted on diagonal. p<.01.

Table 5. Model fit indices and recommended values.


Fit indices Measurement model Recommended values
v2 (209 df) 373.359 (p ¼ .00)
CMIN/DF 1.786 <3.00
Goodness of fit index (GFI) 0.856 >0.90
Standardized root mean square residual (SRMR) 0.051 <0.09
Root mean square error of approximation (RMSEA) 0.065 <0.08
Non-normed fit index (NNFI) or (TLI) 0.932 >0.92
Comparative fit index (CFI) 0.944 >0.92

Figure 2. Results of structural model (standardized loadings and R2s) and fit measures.

complied with the recommended minimum of 0.50 that with large samples v2 is not an adequate measure
(Hair, 2006) and those that did not – perceived risk of fit. From the alternative indices used, only GFI was
(average variance extracted [AVE] ¼ 0.46) and cost below the recommended value. In our favor, the lit-
(AVE ¼ 0.47) – were very close to it. In order to com- erature has indicated that GFI (as well as other abso-
plement the validity analysis of the abovementioned lute fit indices that are simple transformations of v2),
constructs, we analyzed relationships between con- have similar problems to those of v2, which led to the
structs as suggested by the literature. Correlations decline of its use (Hair, 2006).
between the constructs of the measurement model
were all below 0.8 (Bagozzi & Yi, 1988), suggesting
Test of structural relationships
discriminant validity. Additionally, the shared variance
among each of the two factors was generally lower The structural model proposed exhibited a satisfactory
than their AVEs, satisfying Fornell and Larcker (1981) level of fit (v2 ¼ 311.204, df ¼ 196; RMSEA ¼ 0.056;
discriminant validity criterion, as depicted in Table 4. TLI/NNFI ¼ 0.948; CFI ¼ 0.956). The analysis of the
The overall measurement model fit was judged to structural relationships (see Figure 2) confirmed all
be satisfactory based on the comparison of its fit indi- hypotheses proposed, except Hypothesis 4 and 5; thus
ces and recommended values (Hair, 2006), both of the negative influence of perceived cost (c ¼ 0.08;
which are presented in Table 5. The analysis showed a p > .05) and risk (c ¼ 0.04; p > .05) on the intention
statistically significant v2, which theoretically means to use the omnichannel approach were rejected. As
the model should be rejected. However, it is known expected, perceived compatibility, usefulness, and ease
8 S. C. SILVA ET AL.

Table 6. Averages of (composite) measures of model constructs, for the whole sample and for low vs. high frequency users, with
t-tests.
All Low freq. users (n ¼ 101) High freq. users (n ¼ 89)
Mean SD Mean SD Mean SD t
Risk 0.05 0.80 0.16 0.83 –0.30 0.70 4.07
Cost –0.28 0.86 –0.14 0.87 –0.43 0.82 2.40
Compatibility 0.23 0.91 –0.27 0.74 0.80 0.75 –9.84
Usefulness 0.51 0.82 0.06 0.66 1.02 0.67 –9.90
Ease of use 0.66 0.84 0.33 0.80 1.03 0.73 –6.33
Future use intention 1.49 0.91 0.89 0.72 2.16 0.58 –13.23
Actual use 0.34 1.08 –0.51 0.66 1.31 0.47 –22.30
Scale: 2 to 2, where 2 is totally disagree and 2 is totally agree. p.01; p.05.

Table 7. Results of group analysis.


Std. coefficients in unconstrained model v2 test for the difference
v2 for equal (constrained) between unconstrained
Model path Low freq. users High freq. users paths model and constrained models

Hypothesis 1(a) EU ! UI 0.07 0.63 616.70 (397gl); p ¼ .000 10.24 (1gl); p ¼ .001
Hypothesis 1(b) EU ! USE 0.07 0.23 606.01 (397gl); p ¼ .000 1.54 (1gl); p ¼ .215
Hypothesis 2 USE ! UI 0.47 0.07 615.89 (397gl); p ¼ .000 9.43 (1gl); p ¼ .002
Hypothesis 3(a) CPT ! UI 0.42 0.34 607.80 (397gl); p ¼ .000 1.35 (1gl); p ¼ 0.248
Hypothesis 3(b) CPT ! USE 0.48 0.63 608.66 (397gl); p ¼ .000 2.20(1gl); p ¼ .138
Hypothesis 6 UI ! AU 0.68 0.52 607.30 (397gl); p ¼ .000 0.83 (1gl); p ¼ .215
Unconstrained model v2 ¼ 606.47 (396 gl); p ¼ .000 Fully constrained model 31.65 (6gl); p ¼ .000
v2 ¼ 638.11
(402gl); p ¼ .000
EU: ease of use; USE: usefulness; CPT: compatibility; UI: use intention; AU: actual use. p < .01; p < .05.

of use proved to be good predictors of intention to However, this result did not allow us to individually
use the omnichannel approach. Moreover, as hypothe- identify which paths are significantly different among
sized, the influence of both perceived compatibility groups. Therefore, further analysis was conducted by
and perceived ease of use on use intention is exerted constraining one path at a time and comparing each
directly, but also indirectly, through the mediation of of those partially constrained models with the uncon-
usefulness. Finally, use intention is a good predictor strained model. The results (see Table 7) show signifi-
of the actual use of the omnichannel approach. The cant differences among the two groups with regard
model explains 65% of the intention to use an omni- only to the path from ease of use to use intention –
channel approach and 45% of its actual use. meaning that ease of use is a very good predictor of
To examine potential differences in path estimates intention to use in the case of high-frequency users
between the two samples – high-frequency versus but not in the case of low-frequency users – and the
low-frequency users – a multiple group analysis path from usefulness and use intention – indicating
(Byrne, 2013) was conducted. The results of this can that usefulness is a good predictor of use intention for
be seen in Table 6. low-frequency users, but not high-frequency users.
First, the model was trimmed, taking into account
nonsignificant p-values of loadings on both groups.
Discussion and implications
Thus, the paths between cost and use intention, and
risk and use intention were cut from the model since Although most hypotheses of the proposed conceptual
the path was nonsignificant for both high-frequency model were confirmed, contrary to our hypotheses,
and low-frequency users. After trimming the model, and to previous findings in e-commerce and m-com-
we tested it with all the path coefficients set as equal merce, our results indicate that neither perceived risks
across the two groups (fully constrained model). Next, nor perceived costs associated with the adoption of an
we tested an unconstrained model in which the path omnichannel approach have a significant influence on
coefficients were allowed to vary between the groups. the intention to adopt it. A possible explanation for
The unconstrained model showed a significantly better this relates to the fact that we are studying a popula-
fit compared to the constrained model (Dv2(6df) ¼ tion of Internet users that are, in most cases, also
1.65; p < .01), thereby supporting that consumer smartphone users. Even for those with more basic pat-
experience with the omnichannel approach affects terns of usage, adopting an omnichannel approach
relationships between the constructs of the model. would not greatly increase the risks and costs that
JOURNAL OF MARKETING CHANNELS 9

already arise from Internet usage in general and, for high-frequency users of the omnichannel
thereby, to which they are already exposed, regardless approach, usefulness is fundamental to explain use
of whether they are omnichannel consumers. intention for low-frequency users alone. This suggests
All the remaining hypothesized relationships were that the major hindrance for low-frequency users is
confirmed. Our results showed that intention to use not the difficulty in using the necessary means to
an omnichannel approach is explained by variables adopt the omnichannel approach, but finding that this
that have been used in the past to explain the adop- approach does not significantly improve their shop-
tion of innovations, namely TAM traditional variables ping experiences. Hence, companies should focus on
– usefulness and ease of use – as well as compatibility increasing the potential of their omnichannel presence
with the consumer lifestyle and values. Moreover, to solve customers’ problems. Low-frequency users
these three variables proved to be closely related. will only adopt more omnichannel shopping behaviors
Usefulness, in the case of the omnichannel approach, if they perceive them as advantageous when compared
seems to be, to an important extent, explained by the to traditional processes of purchase. On the other
ease of use and compatibility. This means that useful- hand, usability should also be a concern of companies
ness entails, at least in part, being easier to use and that want to take the most advantage of this approach.
more compatible with consumers’ lives, compared to Despite the results pointing out that use complexity is
other means of making purchases. In other words, the not a significant obstacle to the adoption of omni-
usefulness of the omnichannel approach has to do channel behaviors, among frequent users, those who
with convenience – that is, the possibility of accessing are more willing to continue using it are those who
and of buying anything, anywhere, anytime perceive it as easier to use. Therefore, ease of use may
(Piotrowicz & Cuthbertson, 2014; Verhoef et al., not be considered that relevant to improving the
2015), which is not possible using traditional purchase usage of less frequent users of this approach but
approaches. These results confirm the results obtained should be considered important to increase the chan-
in the literature for e-commerce, as well as m-com- ces of using the alternative approach among those
merce (e.g. Rose et al., 2012; Wu & Wang, 2005). that are already using it.
Finally, as expected, use intention is a satisfactory pre-
dictor of the actual use of the omnichannel approach.
Limitations and further research
This paper also offers some interesting insights for
managers and entrepreneurs, who need to learn more As with any study, this research has a number of limi-
about implementing an omnichannel approach. tations that should be taken into consideration. One
Preventing perceived risk and financial costs should such limitation is related to the convenience sampling
not be the main concern of companies because they process: because a nonprobabilistic method, it does
are not obstacles to omnichannel adoption. The main not allow us to generalize these results to the entire
efforts must be spent on making omnichannel usage Portuguese population. In addition, the data collection
easy and compatible with consumers’ lives. In fact, was performed only via online channels, since we only
some managers may think that one of the most rele- assessed participants already using these channels.
vant hindrances for customers’ use of this approach This limitation leads us to suggest that future studies
could be the burden of the risk involved. However, use randomized stratified samples through various
our results suggest differently. They seem to associate online and offline collection methods. We also high-
any costs that may result from the use of this lighted the importance of variables that can contribute
approach to the company. If the company is propos- to better explain the importance of omnichannel
ing this alternative, it is because they believe it may approaches to consumers; in particular, the import-
pay off. Companies do it in the belief that additional ance for processes of social influence, quality of ser-
channels provide additional value. Thus, if the com- vice, trust, loyalty, motivation, and time, among
pany cannot guarantee that this approach will work, it others. We suggest that these be covered in future
would be better not to communicate it at all. studies on the subject.
Otherwise, they could be exposed to the risk of poor We acknowledge that there is a tendency to sup-
opinions of the omnichannel alternative, which could port that the perceived ease of use increases in the
in turn negatively impact current options. The above presence of experienced users. This means that in
thus tends to work as a guarantee for the customer. order to better understand which factors limit the use
Another key finding of this study was that while of this approach, mainly nonusers or infrequent users
ease of use is only a good predictor of use intention should have been selected. As our sampling method
10 S. C. SILVA ET AL.

did not anticipate this framing, we conducted a mod- Ajzen, I., & Fishbein, M. (1980). Understanding attitudes
eration analysis so that we could compare the effects and predicting social behavior. Englewood Cliffs, NJ:
of independent variables in the intention of use Prentice-Hall.
Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of struc-
according to user status. Future studies are advised to tural equation models. Journal of the Academy of
focus mainly on nonusers of the omnichannel Marketing Science, 16(1), 74–94.
approach in order to better understand the reasons Bell, D. R., Gallino, S., & Moreno, A. (2014). How to win in
for their nonuse. In this case, a more qualitative an omnichannel world. MIT Sloan Management Review,
method could be used. 56, 45–53.
Another limitation arises from the fact that the sur- Biswas, D., & Biswas, A. (2004). The diagnostic role of sig-
nals in the context of perceived risks in online shopping:
vey was conducted online, which means that the sam- Do signals matter more on the web? Journal of
ple may be biased towards heavier Internet users. In Interactive Marketing, 18(3), 30–45. doi:10.1002/dir.20010
the future a different sampling method may be used, Bryman, A., & Bell, E. (2015). Business research methods.
in accordance with our suggestion in the previous Oxford, UK: Oxford University Press.
paragraph, to ensure that the whole population is rep- Byrne, B. M. (2013). Structural equation modeling with
resented. Similarly, bias may arise depending on the AMOS: Basic concepts, applications, and programming.
New York, NY: Routledge.
type of consumers or product purchased; thus, our Davis, F. D. (1989). Perceived usefulness, perceived ease of
results could vary for other types of products. For use, and user acceptance of information technology. MIS
instance, for products with high need-for-touch (e.g. Quarterly, 13(3), 319–340. doi:10.2307/249008
clothing and footwear) (Peck & Childers, 2003), the Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). User
odds of buying online, and therefore experiencing an acceptance of computer technology: A comparison of two
omnichannel approach, are slimmer than when buy- theoretical models. Management Science, 35(8), 982–1003.
doi:10.1287/mnsc.35.8.982
ing products such as books or CDs, which display a Delft, W. V. (2013). Omni channel shopping behaviour dur-
lower need-for-touch. However, we believe that once ing the customer journey: An empirical study into the con-
the omnichannel approach becomes more common tribution of omni channel shopping characteristics during
among Web users, it will also reduce uncertainty in the customer journey by consumer segments (Unpublished
purchasing high need-for-touch products: consumers graduation thesis). Technische Universiteit Eindhoven,
may increase their chances of touching the products Eindhoven, the Netherlands.
Fornell, C., & Larcker, D. F. (1981). Evaluations structural
through the several touchpoints of the customer jour- equation models with unobservable variables and meas-
ney, and in the same vein can return items easily. urement error. Journal of Marketing Research, 18(1),
Nevertheless, this should be tested in future studies. 39–50. doi:10.2307/3151312
At a national level, we suggest that studies be con- Frazer, M., & Stiehler, B. E. (2014). Omnichannel retailing:
ducted to understand the impacts of using multiple The merging of the online and off-line environment. Paper
channels over the consumer journey, but across the presented at the Global Conference on Business and
Finance, San Jose, Costa Rica
various business sectors. It is also critical to conduct Fulgoni, G. M. (2014). “Omni-channel” retail insights and
comparative studies of companies that have already the consumer’s path-to-purchase. Journal of Advertising
implemented an omnichannel strategy. Research, 54(4), 377–380. doi:10.2501/JAR-54-4-377-380
Finally, it would be appropriate to carry out a lon- Gefen, D., Karahanna, E., & Straub, D. W. (2003). Trust
gitudinal study to verify the evolution of variables and TAM in online shopping: An integrated model. MIS
such as confidence, satisfaction, and loyalty in compa- Quarterly, 27, 51–90.
Glover, S., & Benbasat, I. (2010). A comprehensive model
nies that adopt an omnichannel approach. of perceived risk of e-commerce transactions.
International Journal of Electronic Commerce, 15(2),
47–78. doi:10.2753/JEC1086-4415150202
References
Hair, J. F. (2006). Multivariate data analysis. Upper Saddle
Ajzen, I. (1985). From intentions to actions: A theory of River, NJ: Pearson Prentice Hall.
planned behavior. In: J. Kuhl & J. Beckmann (Eds.), Hsi-Peng, L., & Su, P. Y.-J. (2009). Factors affecting pur-
Action control: From cognition to behavior (pp. 11–39). chase intention on mobile shopping web sites. Internet
Berlin, Germany: Springer-Verlag. Research, 19(4), 442–458.
Agarwal, R., & Prasad, J. (1998). A conceptual and oper- Karahanna, E., Agarwal, R., & Angst, C. M. (2006).
ational definition of personal innovativeness in the Reconceptualizing compatibility beliefs in technology
domain of information technology. Information Systems acceptance research. MIS Quarterly, 30(4), 781–804.
Research, 9(2), 204–215. doi:10.1287/isre.9.2.204 Kim, C., Mirusmonov, M., & Lee, I. (2010). An empirical
Ajzen, I., & Fisbbein, M. (1974). Factors influencing inten- examination of factors influencing the intention to use
tions and the intention-behavior relation. Human mobile payment. Computers in Human Behavior, 26(3),
Relations, 27(1), 1–15. doi:10.1177/001872677402700101 310–322. doi:10.1016/j.chb.2009.10.013
JOURNAL OF MARKETING CHANNELS 11

Lin, H.-F. (2007). Predicting consumer intentions to shop Rey-Moreno, M., & Medina-Molina, C. (2016).
online: An empirical test of competing theories. Omnichannel strategy and the distribution of public serv-
Electronic Commerce Research and Applications, 6(4), ices in Spain. Journal of Innovation & Knowledge, 1(1),
433–442. doi:10.1016/j.elerap.2007.02.002 36–43. doi:10.1016/j.jik.2016.01.009
Lopez-Nicolas, C., & Molina-Castillo, F. J. (2008). Customer Rigby, D. (2011). The future of shopping. Harvard Business
knowledge management and e-commerce: The role of Review, 89(12), 65–76.
customer perceived risk. International Journal of Rogers, E. M. (1983). Diffusion of Innovations (3rd ed).
Information Management, 28(2), 102–113. doi:10.1016/j. New York, NY: Free Press of Glencoe.
ijinfomgt.2007.09.001 Rose, S., Clark, M., Samouel, P., & Hair, N. (2012). Online
Morgan-Thomas, A., & Veloutsou, C. (2013). Beyond tech- customer experience in e-retailing: An empirical model
nology acceptance: Brand relationships and online brand of antecedents and outcomes. Journal of Retailing, 88(2),
experience. Journal of Business Research, 66(1), 21–27. 308–322. doi:10.1016/j.jretai.2012.03.001
doi:10.1016/j.jbusres.2011.07.019 Schoenbachler, D. D., & Gordon, G. L. (2002). Multi-chan-
Neslin, S. A., Grewal, D., Leghorn, R., Shankar, V., Teerling, nel shopping: Understanding what drives channel choice.
M. L., Thomas, J. S., & Verhoef, P. C. (2006). Challenges Journal of Consumer Marketing, 19(1), 42–53. doi:10.
and opportunities in multichannel customer manage-
1108/07363760210414943
ment. Journal of Service Research, 9(2), 95–112. doi:10. Triandis, H. C. (1980). Values, attitudes, and interpersonal
1177/1094670506293559
behavior. In: H. Howe & M. Page (Eds.), Nebraska
Neslin, S. A., Jerath, K., Bodapati, A., Bradlow, E. T., Deighton,
Symposium on Motivation. (Vol. 27) (pp. 195–259).
J., Gensler, S., … Venkatesan, R. (2014). The interrelation-
Lincoln, NB: University of Nebraska Press.
ships between brand and channel choice. Marketing Letters,
Venkatesh, V., & Davis, F. D. (2000). A theoretical exten-
25(3), 319–330. doi:10.1007/s11002-014-9305-2
Pavlou, P. A. (2003). Consumer acceptance of electronic sion of the technology acceptance model: Four longitu-
commerce: Integrating trust and risk with the technology dinal field studies. Management Science, 46(2), 186–204.
acceptance model. International Journal of Electronic doi:10.1287/mnsc.46.2.186.11926
Commerce, 7(3), 101–134. Verhoef, P. C., Kannan, P., & Inman, J. J. (2015). From
Peck, J., & Childers, T. L. (2003). Individual differences in multi-channel retailing to omni-channel retailing:
haptic information processing: The “need for touch” Introduction to the special issue on multi-channel retail-
scale. Journal of Consumer Research, 30(3), 430–442. doi: ing. Journal of Retailing, 91(2), 174–181. doi:10.1016/j.jre-
10.1086/378619 tai.2015.02.005
Piotrowicz, W., & Cuthbertson, R. (2014). Introduction to Wu, J. H., & Wang, S. C. (2005). What drives mobile com-
the special issue information technology in retail: Toward merce?: An empirical evaluation of the revised technology
omnichannel retailing. International Journal of Electronic acceptance model. Information & Management, 42(5),
Commerce, 18(4), 5–16. doi:10.2753/JEC1086-4415180400 719–729. doi:10.1016/j.im.2004.07.001
12 S. C. SILVA ET AL.

Appendix: Constructs under study

Constructs evaluated Items


Perceived risk (Wu & Wang, 2005)  I think monetary transactions (e.g. payments over the internet) in this type of approach have a
high risk
 I agree that purchases of goods and services by using this approach Omni channel have a high risk
(e.g. product received does not match the product you ordered)
 I believe that obtaining information and promotional campaigns for products through this approach
have a high risk (e.g. information through websites, promotional campaigns through social
networks, etc.)
 I think that the use of a omnicanal approach puts my privacy at risk (e.g. supply of personal data for
online payments, exposure in social networks, etc.)
Cost (Wu & Wang, 2005)  I think the cost of equipment for the adoption of a omnicanal approach is quite high (e.g.
Smartphones / tablets)
 I believe that the cost of access to multiple channels simultaneously is quite high (e.g. Spending on
online access, visits to shops, phone, etc.)
 I consider that the transaction fees inherent to a omnicanal approach are too high (e.g. Online
payment rates for certain services, telephone call rates, etc.)
Compatibility (Wu & Wang, 2005)  The omnicanal approach is compatible with most aspects of my transactions (e.g. desire for
personalized service, rapid procedures, the means of payment flexibility, etc.)
 The omnicanal approach fits in my lifestyle
 Adopting a omnicanal approach fits well with the way I like to get involved in the buying process
Perceived utility (Wu & Wang, 2005)  The use of a omnicanal approach will improve my performance during the buying process (e.g. I have
more alternatives for my shopping desires; I will get faster information about the product I want, etc.)
 The use of a omnicanal approach will increase my productivity during the purchase process (e.g.
getting faster and more complete price information and alternative products, allowing me to
save time)
 The use of a omnicanal approach will increase my effectiveness during the purchase process (e.g. my
needs will be met faster)
 I think the use of a omnicanal approach is very useful in my involvement in the purchasing process.
 I recognize that it is easier to find what I’m looking through a omnicanal approach.
Perceived ease of use (Wu & Wang, 2005)  I find it easy to learn to use a omnicanal approach.
 I consider easy that to become competent in the use of a omnicanal approach.
 I think the use of a omnicanal approach is easy.
Intention to use (Wu & Wang, 2005)  Assuming I have access to a omnicanal approach I intend to use it.
 Given that I have access to a omnicanal approach, I would like to use it.
 The use of a omnicanal approach is appealing to me
Current use (Wu & Wang, 2005)  How often do you engage in purchases using an omnicanal approach

You might also like