Keywords

1 Introduction

Omnichannel retailing is a cross-channel approach of retailers that provides all channels to customers in an integrated way [1]. In the past, brick-and-mortar retail stores were unique in allowing consumers to touch and feel merchandise and provide instant gratification; Internet retailers (online shop providers), meanwhile, tried to woo shoppers with wide product selection, low prices and content such as product reviews and ratings. As the retailing industry evolves toward a seamless “omnichannel retailing” experience, the distinctions between physical and online will vanish, turning the world into a showroom without walls [2]. Therefore, omnichannel retailing includes all offline, online and mobile channels that enable customers to interact with a retailer in multiple ways. It focuses on a seamless experience across channels, places more emphasis on the interplay between channels and provides services that combine offline and online channels to allow for simultaneous channel usage [3].

This behavior is enabled and supported to a great extent by the omnipresence of smartphones, since they can be used to access (third party) online services while shopping in a brick-and-mortar retail store. Additionally, retailers are implementing and rolling out more and more digital in-store services to improve the customer experience. This includes services that customers can access via personal smartphones as well as dedicated in-store technology [4].

These digital technologies enable retailers to provide personalized services. These tailored services may offer customers additional value, but also require collecting and analyzing data about users and their behaviour to identify customer preferences and needs [5]. This may also lead to privacy concerns of customers and reduce their willingness to use the personalized services.

In the context of online shops this is already widely adopted and users have experienced these services. But the deployment of in-store technologies and omnichannel retailing services also bring more technology-based services to retail stores, providing retailers with the opportunity to collect data in retail stores, too. Consequently, personalized services could also be provided in retail stores. But it is unclear whether privacy concerns and the intention to adopt personalized services in retail stores differ compared to online shops. This is what we examined in this study. We base this study on a study by Sheng/Nah/Siau [6] that investigated the impact of personalization and privacy concerns in a ubiquitous commerce context. We replicated the methodology but altered the context, leading to a replication with a context extension [7].

Structure of the paper.

The remaining paper is structured as follows: In Sect. 2 we lay out related work concerning the purchasing context, privacy concerns and the intention to adopt omnichannel retailing services. Section 3 describes the research methodology. Section 4 presents the results of the study. Finally, Sect. 5 discusses the results and draws conclusions.

2 Omnichannel Retailing

In this section we analyze related work of omnichannel retailing aspects that may influence privacy concerns and the intention to adopt omnichannel retailing services to derive hypotheses for the subsequent empirical study.

2.1 Retailing Context

The retailing context defines the conditions that a customer experiences while buying products. For the study two different context categories were chosen:

  • Buying in an online shop

  • Buying in a retail store

These two alternatives are characterized through different external factors that influence customer behavior.

Online shops are commonly used in developed countries. To access them customers use different devices like desktop PCs, laptops, tablets or smartphones. To customers these shops provide various advantages, e.g. higher product variety, more detailed product information, customer reviews, lower prices, higher price transparency and location - as well as time-independency. To interact with the retailer online shops provide multiple technology based communication services like email, contact form, (video) chat or phone, but do not provide face-to-face interaction with shop assistants [8].

Brick-and-mortar retail stores on the other hand provide the opportunity to obtain personal advice from shop assistants and to experience products, their characteristics and their qualities.

Omnichannel retailing tries to combine online and offline channels and enables customers to even use them simultaneously. Consequently they are able to use online channels to access additional product information or use payment services on smartphones while in a retail store.

For this study we consider the two contexts online shop and retail store as mutually exclusive and compare them in an experimental setting.

2.2 Privacy Concerns

Gathering personal data about customers enables retailers to provide personalized goods and services. In addition, studies show that consumers are willing to disclose such data for personalized services if they provide additional value to the consumer [9]. On the other hand privacy concerns of customers rise. Faced with personalized services, consumers worry about the loss of control over the distribution and use of this data. They may feel profiled and tracked [10]. Expectancy theory claims that users are behaving in a way to maximize positive outcomes while minimizing negative outcomes [11]. Consequently customers would like to obtain personalized products or services, by providing as little information as possible [12, 13]. Additionally the personalization-privacy paradox describes the tendency that consumers who value information transparency are also less likely to participate in personalization [14].

Hence, the result of this risk assessment influences the behavior of consumers. Their willingness to provide data to companies in exchange for services that offset the perceived risks of this disclosure is called privacy calculus [15]. Consequently, depending on the purpose and the context of use the customers’ privacy concerns may vary then using certain technologies. Thus, privacy concerns of customers in the context of IS adoption are situation dependent ([16, 17, 18]).

For this study we have to distinguish between the purchasing situations in online shops and retail stores. In both situations customers may have to reveal personal information to a retailer to access personalized services. In the case of retail stores customers also reveal their physical location. Additionally customers are in a public space and using a personalized service may disclose personal preferences to a third person. Consequently they may further lose perceived control over their data which may lead to increased privacy concerns [19].

Thus based on the general tendency of customers to disclose as little information as possible to minimize negative outcomes [11], to create a privacy calculus [20], to abstain from e-commerce transactions when privacy concerns emerge [15] and to prefer situations with high perceived control over personal data [19] we hypothesize that the privacy concerns of personalized services are higher for customers when they are shopping in a retail store than when they are using an online shop.

  • H1: The privacy concerns triggered by personalized services are higher when customers buy in a retail store than in an online shop.

2.3 Intention to Adopt

Studies have shown that trust, perceived usefulness and perceived ease of use together explain a considerable proportion of variance in intended adoption of online shopping services [21]. Additionally it was found that trust is an important factor for adopting personalized services that require gathering and analyzing personal data of customers [22]. Consequently, for this study we discuss these three factors for personalized services in the contexts of online shops and retails stores:

  • Trust: It has been shown that in the context of online shops repeat customers trusted e-vendors more than new users [23]. But currently the majority of people have not experienced personalization in retail stores. Additionally, as discussed in Sect. 2.2, the use of personalized services may reveal personal data or preferences which may represent a higher risk in a public space. Consequently we presume trust in personalized services in online shops may be higher than in retail stores.

  • Perceived Usefulness: Also perceived usefulness was found to be higher in repeat users and the general experience with personalized services in online shops is much higher than in retail stores [23]. The use of technology may also be seen as an additional layer between the customer and the already present products in a retail store. Consequently we presume that perceived usefulness of personalized services in online shops may be higher than in retail stores.

  • Perceived Ease of Use: Since online shops have already been widely used, users have experience using these systems, and more experience corresponds with higher perceived ease of use ratings in online shops [23]. In-store technologies and systems on the other hand are not very common and may represent an additional layer between the customer and the product, which may be associated with further effort or burden to access products. Thus we argue that perceived ease of use may be higher in online shops.

Consequently, based on [6] we hypothesize that in the context of retail stores the acceptance of personalized services is higher and this leads to a higher intention to adopt them than in online shops.

  • H2: The intention to adopt a personalized service is higher when customers buy products in an online shop than in a retail store.

2.4 Relation Between Privacy Concerns and Intention to Adopt

A recent study conducted by [24] indicates that privacy concerns are the most important factor for the intention to adopt innovative electronic shopping concepts. Results show that the intention to adopt was reduced when customers feared a possible misuse of personal data generated during the shopping process. Subsequently, concerns about privacy reduce the intention to adopt services based on their expected positive or negative effects.

In the context of this study negative effects are the violation of perceived privacy due to gathering, analyzing and use of personal data. Positive outcomes may be produced by personalized offers, product presentations, vouchers, etc. that are based on user behavior. The consideration of positive effects of personalization due to loss of privacy has to be perceived positively for the user for the decision to use such a service. Hence, if this calculation is negative, the users will not intend to use the service [15]. We therefore hypothesize that there is a similar negative relation between privacy concerns and the intention to adopt personalized services.

  • H3: Privacy concerns have a negative effect on the intention to adopt services in an omnichannel retailing context.

3 Methodology

In this section we lay out the methodology to test the hypotheses derived in Sect. 2. This is done by conducting an empirical study with a scenario-based online survey. This empirical study is based on a research model that incorporates the three hypotheses and illustrates the way the included factors personalization, context, privacy concerns and intention to adopt influence each other (c.f. Fig. 1):

Fig. 1.
figure 1

Research model

Full size image

  • H1: The privacy concerns triggered by personalized services are higher when customers buy in a retail store than in an online shop.

  • H2: The intention to adopt a personalized service is higher when customers buy products in an online shop rather than in a retail store.

  • H3: Privacy concerns have a negative effect on the intention to adopt services in both an online and retail context.

Scenarios.

To analyze the influence of personalization and purchasing context we used a scenario-based approach that covered all alternative combinations. Scenarios are usually used in experimental studies to define contexts of a study, to describe user tasks or specify certain conditions or variables of a study. Using scenarios enables researchers to analyze phenomena independently from the development stage of needed technologies or functionalities. Consequently, scenario-based approaches can be used to evaluate potential and plausible future situations and to gather information on how people assess them from the present perspective [25].

Consequently, the described research model led to 4 scenarios which are shown in Fig. 2:

Fig. 2.
figure 2

Scenarios for research design

Full size image

The different scenarios were built by identifying the relevant characteristics of the analyzed phenomenon [25] and therefore describe potential situations which should be thought of by the subject during the online survey.

Retailing context.

In the case of omnichannel retailing a study in 2015 [26] revealed that 61% of the subjects were familiar with interactive omnichannel retailing technologies, but only 30% have actually used them actively (mainly self-service terminals). Since only a minority of people have experienced some of these technologies and a comprehensive real world implementation of omnichannel services along the customer journey was not available to us, a scenario-based approach seemed appropriate.

Personalization.

The scenarios should also project characteristics of personalization by using technology along multiple stages of a customer journey without being too complex. Additionally, we decided not to include cutting-edge technology to simulate personalization, because the subjects may not be familiar with it and therefore could struggle to assess it. Hence, the scenarios included technology that we expected the subjects to know about and which were easy to describe, such as virtual shelves and NFC payment (Table 1).

Table 1. Personalization levels in the two contexts
Full size table

Experimental setting.

The study is based on an experimental setting using a within subject design. Thus every subject was exposed to every scenario. To measure privacy concerns and the intention to adopt the omnichannel retail context the subjects had to fill out a survey directly after every scenario without a time limit. The order of the scenarios was randomized.

The survey consisted of the items shown in Table 2.

Table 2. Items to measure privacy concerns and intention to adopt
Full size table

The items in Table 2 were based on the approved methodology of Sheng/Nah/Siau [6]. Originally, Sheng/Na/Siau derived the items for privacy concerns (PC1-4) from ([27, 15]) and the items from intention to adopt (INT1-3) ([21, 19]). All items were rated on a 7-point Likert scale from 1 (“strongly disagree”) to 7 (“strongly agree”). The order of the items and also the order of the scenarios were randomized for each subject.

Data collection.

To conduct this study, non-probability sampling technique was chosen [28], because omnichannel retailing services cannot be considered to be widely implemented and therefore already adopted by a wide population. Consequently, only experienced smartphone users and online shoppers were expected to be potential subjects who have the needed experience with the related technologies, services and personalization issues. Additionally, studies showed that younger people are more likely to adopt new information technologies, while older people are more resistant to using new information technologies. Thus, the subjects were recruited by using social network groups of students in Austria. For our study, students represent a widely homogenous group of subjects with frequent experience in e-commerce, m-commerce and generic smartphone usage.

Survey execution.

The online survey was conducted in July 2016 using the survey tool SoSci Survey (www.soscisurvey.de).

4 Results

4.1 Sample and Measurement Model

Overall, 112 questionnaires were finished validly and could be integrated into the sample. All subjects were Austrian students. Demographic details are shown in Table 3.

Table 3. Demographic information of subjects.
Full size table

Before confronting the subjects with the different scenarios their experience with mobile technologies in general and with omnichannel retailing was gathered. All subjects already used smartphones with internet access in general. Explicit experience with retailing services on smartphones was gathered and analyzed based on the number of years smartphones were used (i) with apps from retailers, (ii) if mobile shopping was performed on smartphones, and (iii) if smartphones were used in retail stores to access services (e.g. research for product details, compare prices, mobile payment, location-based services, etc.). The results show that all subjects have experience with using smartphones and a majority of the subjects also have extensive experience using them in a shopping contexts.

Furthermore, the frequency of the usage was collected by using a 6-point Likert scale from never (1) to often (6). For easier readability, the frequency values were cumulated in the three categories rarely, sometimes and frequently in Table 4. The results show that the majority of the subjects not only have extensive experience using smartphones for retailing services, but also use them at least sometimes or frequently. Smartphones are used most intensely not only for mobile shopping activities, but also with retailer apps.

Table 4. Experience and frequency of smartphone usage with retailing services
Full size table

Therefore, we can conclude that the subjects were sufficiently experienced in using smartphones for e-commerce and m-commerce services, which suggests that they are potential omnichannel retailing service users. Additionally, the subjects were relatively young and had a high level of formal education, suggesting they are more likely to adopt new information technology. Although this may limit the generalizability of the study, these subjects are potential customers targeted by omnichannel retailing services and form a homogenous group [6].

Factor Analysis.

In order to assess the reliability and validity of the constructs Privacy Concern (PC) and Intention to Adopt (INT) a factor analysis was conducted. All items were loaded on the constructs they were intended to measure, with non-significant loadings on the other construct, which corresponds to the results of the original study from Sheng/Nah/Siau [6] (Table 5).

Table 5. Factor analysis; rotation method: Oblimin with Kaiser normalization.
Full size table

Results show items PC1, PC2, PC3 and PC4 load very high on privacy concerns. Items INT1, INT2 and INT3 load very high on intention to adopt.

Table 6. Variance explained
Full size table

Cronbach’s alpha coefficients for all constructs far exceed the threshold of 0.7 ([29]) and therefore the measurements for privacy concerns and intention to adopt are highly reliable in terms of internal consistency (Table 7).

Table 7. Cronbach’s Alpha Coefficients
Full size table

4.2 Hypothesis Testing

Hypothesis H1 and H2 claim different effects of personalized services and the purchasing context on privacy concerns and the intention to adopt these services. We used repeated measure ANOVA to analyze this interaction, which can be considered to be an appropriate statistical method in this context.

For examining the relationships between privacy concerns and intention to adopt personalized services (H3) a regression analysis was conducted. Based on the fact that the study is a within-subject design, we tested the relationship between privacy concerns and intention to adopt for each scenario separately in order to guarantee the independence assumption of the regression.

Table 8. Mean and standard deviations for privacy concerns
Full size table
Table 9. Results for repeated-measure ANOVA for Privacy Concerns
Full size table

Privacy Concerns.

Privacy concers were analyzed using a repeated-measure ANOVA test using the within-subject factors personalization and retailing context as independent variables. Table 8 reports the mean values and standard deviations, Table 9 reports the results of the ANOVA test.

Figure 3 depicts the interaction effect of personalization and context on privacy concerns. As presented, personalization triggers higher privacy concerns in both online (t = 8.002 p < 0.000) and retail (t = 8.225, p < 0.000) contexts. The difference between customers’ privacy concerns for personalized services versus non-personalized services is greater in the retail context (mean difference = 1.29) than in the online context (mean difference = 1.17). For non-personalized services, there is a significant difference in privacy concerns between online and retail contexts (t = 3.167, p = 0.002); for personalized services, the difference between online shop and retail store is nearly the same (t = 5.428, p = 0.000) and the difference is also significant. Hence, H1 is supported.

Fig. 3.
figure 3

Estimated marginal means of privacy concerns

Full size image

Intention to Adopt.

The data associated with the intention to adopt was also analyzed using a repeated-measure ANOVA test with the two within-subject factors personalization and purchasing context as independent variables. Table 10 reports the mean values and standard deviations, while Table 11 reports the results of the ANOVA test.

Table 10. Mean and standard deviations for intention to adopt
Full size table
Table 11. Results for repeated-measure ANOVA for Intention to Adopt
Full size table

Figure 4 depicts the interaction effect of personalization and context on intention to adopt services. It shows that customers are more willing to adopt personalization services online. Customers’ intention to adopt services (personalized or non-personalized) is significantly higher in the online shop than in the retail store context (t = 5.176, p = 0.000 for personalized services; and t = 2.629, p = 0.010 for non-personalized services). In online shops, customers tend to be more willing to adopt non-personalized services than personalized services (t = −1.901, p = 0.060). Also, in the retail store context, customers are more willing to adopt non-personalized services than personalized services (t = −3.410, p = 0.010). Thus, H2 is supported.

Fig. 4.
figure 4

Estimated marginal means of intention to adopt

Full size image

Privacy Concerns and Intention to Adopt.

We analyzed the relationships between privacy concerns and intention to adopt for each scenario separately. As mentioned earlier, this is needed to satisfy the independence assumption.

In the personalization and online shop scenario, privacy concerns negatively influence intention to adopt (B = −0.286, p = 0.000), as presented in Table 12. This means that if the subjects have privacy concerns, the intention to adopt is lower.

Table 12. Results of regression in personalization in online shop scenario
Full size table

In the non-personalization and online shop context, the effect of privacy concerns on intention to adopt is not significant (B = −0.089, P = 0.324), as shown in Table 13.

Table 13. Results of regression in no personalization in online shop scenario
Full size table

In the personalization and retail store context, privacy concerns negatively influence intention to adopt (B = 0.416, p = 0.000), as presented in Table 14. This means that privacy concerns also lead to lower intention to adopt in the retail store context.

Table 14. Results of regression in personalization in retail store scenario
Full size table

In the non-personalization and retail store context, the effect of privacy concerns on intention to adopt is also significant (B = −0.380, p = 0.000), as shown in Table 15. The privacy concerns of the subjects also have a negative influence on the intention to adopt services in the retail store context, if the services are not personalized.

Table 15. Results of regression in no personalization in retail store scenario
Full size table

As presented above, H3 is also supported. The results suggest that privacy concerns have a negative impact on intention to adopt the services.

5 Discussion and Conclusion

This study evaluates the influence of personalization services in an omnichannel retailing context. The conducted study design was derived from an experimental study on the impact of personalization and privacy concerns in a ubiquitous commerce context from Sheng/Nah/Siau [6]. Hence, the context was adapted to two different omnichannel retailing situations in an online shop and in a retail store and four different scenarios for the study were be derived.

Hypothesis H1 and H2 claim that privacy concerns are higher and the intention to adopt personalized services are lower when customers buy in a brick-and-mortar retail store than in an online shop. The results show that both hypotheses could be supported. In addition, personalization of services triggers higher privacy concerns in both, the online shop and retail store context (H1). Data show further that the intention to adopt personalized services is significantly higher in an online shop than in a retail store (H2). Furthermore, it can be stated that in both retailing contexts customers tend to be more willing to adopt non-personalized services than personalized services.

The third hypothesis (H3) analyzed the relationship between privacy concerns and the intention to adopt for each scenario. Results of the regression analysis show that privacy concerns have significant negative effects on the intention to adopt personalized services in online shop and retail store contexts. In the retail store context privacy concerns of the subjects have a significant negative influence on the intention to adopt services even when the services are not personalized. These results indicate that consumers are decidedly more concerned about the influential authority of retailers in a brick-and-mortar in general than in online shops.

Future research could extend this study by examining the moderating effects of demographic features such as age, gender and technology experience. Also the personal trust level and personal attitude to data privacy could be an interesting mediating factor for further studies.