Retail applications : what does it take for a shopping app to be successful?

(1)

Retail applications. What does it take

for a shopping app to be successful?

Master Thesis

Master Business Administration, Digital Business track University of Amsterdam, faculty of Economics and Business

Author: Milo van der Zanden Student ID: 10372288 Supervisor: Jonne Guyt

(2)

2 Statement of Originality

This document is written by Milo van der Zanden who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

Purpose - The purpose of this study is to discover which functionalities are the most important for shopping apps in different retail settings. As shopping applications have become a popular channel for many retailers, a better understanding as to how different functionalities of these shopping apps impacts the success of these shopping apps is required. This is especially important because consumers are very picky about which shopping app they download and use, leaving many shopping apps unsuccessful in their goal to attract and engage more consumers.

Design/methodology/approach – By means of an online distributed survey, 239 consumers were asked to evaluate shopping apps that they have on their mobile device or tablet. In total, 163 consumers correctly filled in the survey. (Multivariate) multiple regressions and the PROCESS tool were used for analysis.

Findings - Informational, transactional and experiential functionalities of shopping apps drive usage, intention to keep using and intention to recommend through perceived usefulness, perceived ease of use, perceived enjoyment and attitude. Almost no differences in importance of functionalities between shopping apps were found when accounting for

product nature and product involvement, which is surprising to say the least. Therefore, future research should try to investigate the different effects between specific categories, instead of the relatively broad distinction of product nature and product involvement. Originality/value - This is the first study to explore the impact of shopping app functionalities on app success. Its results suggest that understanding what it takes to maximize the success of shopping applications will be a key topic for future research.

Key words – retail, shopping app functionalities, technology acceptance, consumer behaviour, product involvement, product nature.

(4)

4

1. Introduction.

“Don’t blink, the customer is changing fast,” said Terry Lundgren, executive chairman of

Macy’s Inc. The recent shifts in consumer mind-set paired with technology improvements are changing the nature of retail. As a result, the online retail industry has been the fastest-growing part of the total retail sector, but especially mobile shopping applications have the potential to transform the world of retail even further (Taylor & Levin, 2014). Although shopping apps are still a relative new phenomenon, there has been an increase in consumers’ use of those apps in the recent years (Solomon, 2015). Retail managers also increasingly recognize the potential for apps to grow sales (Dinner, van Heerde & Neslin, 2015). The “apps culture”, as one retail industry watcher explains, is growing to the point that “if you haven’t embraced it yet, you probably will, since ultimately every smartphone user on the planet is expected to buy into it” (Johnson, 2010, p. 24). However, research by Forrester consulting (2015) indicates that 60% of the consumers who use their mobile in the shopping process have less than two shopping apps on their mobile devices. Shopping apps are more of a ‘pull’ instead of a ‘push’ tool for retailers (Bhave, Jain & Roy, 2013), and consumers decide which shopping apps they download, keep, use or delete. Retailers who manage to get their app on consumers’ mobile devices and stimulate active usage, thus have a strong advantage over their competitors.

Although many academics have explored the online and mobile browser environment of shopping (Wang, Malthouse & Krishnamurthi, 2015), very few have yet published research concerning mobile shopping apps. Academics who have conducted preliminary research in the area of shopping apps investigated for example the effect of shopping app usage on brand attitudes and purchase intentions (Bellman, Potter,

(6)

Treleaven-6 Hassard, Robinson, and Varan, 2011). However, it is also necessary for academics and retail

practitioners alike to also identify the factors influencing consumers’ intention to use shopping apps. This can provide them with important information about when shopping apps are able to become successful or not (Kang, 2014). In this light, Taylor and Levin (2014) proved that mobile phone platform and recency of store visit had a positive effect of consumers` likelihood to use shopping apps and Magrath and McCormick (2014) investigated which design elements should be utilised for mobile shopping applications in the fashion industry. Furthermore, recent business research shows that poor usability of shopping apps is the most important factor in deciding not to download, not to use or to delete an app (Deloitte, 2012; Eshet and Bouwman, 2015; Forrester Research, 2011). One of the reasons for poor usability includes inappropriate functionalities of mobile apps (Tarute, Nikou & Gatautis, 2017). It can thus be concluded that shopping apps have to meet the performance expectations of the consumers or they will not be successful and end up as a waste of investments. However, further research to what it takes for a shopping app to be successful, and especially which functionalities are the most import remains scarce.

To fill this gap, this study investigates which shopping app functionalities are important for the success of these apps. The set of functionalities investigated in this study are informational, location-based, experiential, transactional and loyalty-based functionalities. A model is proposed in which consumers’ perception of these five functionalities of shopping apps predict their perceptions about the usefulness, ease of use and enjoyment of shopping apps. These last three perceptions are part of the technology acceptance model (TAM-model), which was developed to understand the adoption and

usage of new technologies (Davis, 1989) and which has widely been used and recognized for this purpose. According to this model, an increase in the perceived usefulness, perceived

(7)

7 ease of use and perceived enjoyment derived from using the shopping app would positively

influence consumers’ attitude toward the shopping app and consequently stimulate desired consumer behaviour. The desired outcome of the customer behaviour, and therefore the measures of success of shopping apps in this particular study, are usage of the app, intention to recommend the app to others and intention to keep using the app in the future.

Furthermore, the retail industry knows a vast number of varying subcategories such as fashion and groceries for example (Berman, Evans & Lowry, 1995). Academics need to account for these differences while doing research in the field of mobile shopping (Childers, Carr, Peck & Carson, 2002). This results in the important notion that the drivers of success for one category of shopping apps, may differ from the others. To account for these possible differences, this study separates retail categories from each other by product nature (utilitarian vs. hedonic) and product involvement (low vs. high involvement). Both distinctions are acknowledged to be key determinants in understanding consumer behaviour (Babin et al., 1994; Celsi & Olson, 1988) and it can thus be expected that the effect of the shopping app functionalities in the proposed model differ for each group. This study thus also investigates whether product nature and product involvement moderate the relationships between the shopping app functionalities, the dimensions TAM-model and the behavioural variables.

While trying to answer this question, this study contributes to both theory and practice. Theoretically, it extends the academic literature regarding mobile shopping apps by explaining parts of the consumer behaviour when it comes to shopping apps. The results are also relevant for retail practitioners concerned with developing a shopping app for their own

business by providing them with insights about which shopping app functionalities are more important than others for them.

(8)

8 This study is structured as follows. The first section reviews findings from earlier

research in the area of mobile commerce and branded apps which are relevant for understanding mobile shopping applications. Here, the concepts of product nature and product involvement are also explained. This section also incorporates predictions and concludes with a visual overview of the conceptual model. The second section describes the methodology of this study and the third section gives an outline of the data analysis and reports the results. In the fourth section, the results are interpreted and discussed. Furthermore, the theoretical and managerial implications and limitations of the study are mentioned and directions for future research are suggested in this section. The fifth and final section summarizes the findings and concludes this study.

2. Theoretical Framework

2.1 M-commerce and the mobile shopping app.

Mobile commerce (M-commerce) allows consumers to search for and purchase products via mobile devices (Yeh & Lin, 2009). It differs from E-commerce in the sense that M-commerce is always accessible as long as consumers have a mobile device and internet connection. M-commerce continues to grow exponentially, but due to the limitations of the mobile web, more retailers are seizing the opportunity by developing their own mobile shopping app (Wong, 2015). In literature, mobile shopping apps are part of the family of branded apps. Based on the definition of branded apps of Bellman et al (2011), mobile shopping apps are defined in this study as: software downloadable to a mobile device, which prominently demonstrates a retailer’s identity, mostly via the name of the app and the presence of the retailer its logo or icon, which can be used for multiple shopping purposes such as e.g.

(9)

9 buying products and searching for product information. A mobile shopping app offers users

a condensed and mobile-optimised retail store which is convenient and user-friendly in its use and which, depending on the retailer, may be transactional or non-transactional oriented (Wong, 2012). Mobile shopping applications are distinct from mobile browser shopping websites in the sense that apps are more of a ‘pull’ instead of a ‘push mechanism (Bhave et al., 2013). Consumers decide whether they download, keep, use or delete an app and only consumers who decided to download the app are exposed to it. Therefore, most users are regular consumers of the retailer. When done right, a mobile shopping app can assist retailers in providing their most important consumer base with the best possible shopping experience. A successful shopping app helps a retailer to improve engagement with regular customers, to look like an innovative brand and increase sales overall and through the app. It enhances consumers’ attitude towards the retailer and increases their purchase intention (Bellman et al., 2011). A shopping app can thus be a powerful tool and an additional channel for a retailer.

Consumers are steadily downloading more shopping apps to their mobile devices and logically, more retail managers recognize the potential of these apps (Dinner et al., 2015). They have welcomed mobile apps as an additional communication channel to attract new customers and increase loyalty among existing ones (Wang et al., 2016). However, although many retailers now have their own shopping app, not all apps are as successful as others. Consumers are selective about the shopping apps they download and use. Research by Forrester consulting (2015) shows that 60% of the consumers who use their mobile devices in the shopping process have less than two shopping apps on their mobile devices.

Consumers are picky, and apps that do not meet the expectations are quickly removed or replaced. The challenge for retailers is thus to build shopping apps that meet the specific

(10)

10 needs of their customers so that it meets their expectations. Only then will their apps be

successful.

2.1 Determining the success of shopping apps.

A shopping app is one of the many channels that retailers can nowadays use to reach consumers. According to Song and Zinkhan (2003) the success of a retail channel is determined by consumers’ behavioural intentions associated with the use of a specific channel. They specifically mention several behavioural intentions associated with online channels such as: repeat visits to the channel, repeat purchases, and positive remarks or comments about the channel. With regard to shopping apps, Wu (2015) states that the success of shopping apps mainly depends on users’ continual usage. Drawing from these intentions, this study focuses on the most commonly referenced online behavioural intentions with regard to M-commerce – usage of the app, intention to keep using the app and intention to recommend the app to others – as measures for success of the shopping app as these relate most directly to successful shopping apps in terms of usage.

In many studies, a consumer his/her intention to keep using a channel of a retailer is seen as a result of his/her attitude towards using the specific channel (Koufaris, 2002). Studies concerning the theory of reasoned action (Sheppard, Hartwick & Warshaw, 1988) also showed strong correlation between attitude and consumers’ intention to engage in a specific behaviour. This has been empirically supported in a variety of different context including E-commerce and M-commerce (Yoh, Damhorst, Sapp & laczniak., 2003; Taylor & Levin, 2014). It is likely that this link also exists for of shopping apps. On the one hand, if

consumers do not like an app, i.e. if they have a negative attitude towards the app, chances are that they will delete the app and never download it again. On the other hand, if

(11)

11 consumers do like the app, i.e. if they have a positive attitude towards the app, chances are

that engage more with the app and thus are of more value for the retailer. It is thus proposed that consumers’ attitude towards shopping apps will be a significant predictor of consumers’ behaviour in regard to those apps. In other words, it is expected that if a consumer his/her positive attitude towards a shopping app increases that his/her frequency of app usage, intent to keep using the app in the future and intent to recommend the app to others also increases. This study thus hypothesizes the following.

H1a. Attitude toward a shopping app is positively related to frequency of usage of the shopping app.

H1b. Attitude toward a shopping app is positively related to intent to keep using the shopping app in the future.

H1c. Attitude toward a shopping app is positively related to intent to recommend the shopping app to others.

2.2 Drivers of attitude towards shopping apps

As a start of understanding what it takes for consumers to have a favourable attitude towards shopping apps, this study draws upon the technology acceptance model (TAM-model) from the information systems literature. This model was developed to understand the adoption and usage of new technologies (Davis, 1989). It revolves around the premise that the intention to use a technology is the result of a consumers’ attitude towards the new technology, and that the attitude towards the new technology is influenced by the benefits derived from using the new technology. Bhave et al. (2013) already stated that attitude towards branded apps in general are influenced by benefits derived from using these apps,

(12)

12 which is why the TAM-model is an appropriate model for examining which drivers are the

most important to increase attitude towards shopping apps. Although the model is limited to only three determinants of attitude towards the new technology, it is widely recognized as a proper starting point for investigating the influencers of the success of a new technology.

The first determinant of the TAM-model is the perceived usefulness (PU) of the technology, which refers to the degree to which using the technology will improve users’ performance. The second element is the perceived ease of use (POEU) of the technology, which refers to the degree to which using the technology is effortless and easy to learn. While PU refers to the outcome of the shopping experience, PEOU refers to the process leading to the final outcome (Davis, 1989). It is often confirmed that PU and PEOU can both predict consumers’ attitude towards the new technology (Davis, 1989; Ha & Stoel, 2009). The third dimension of the TAM model, which was later added by Davis, Bagozzi and Warshaw (1989), is perceived enjoyment (PE). PE is referred to as the extent to which the activity of using the technology is perceived to provide value in its own right, besides any performance consequences that may occur (Davis, 1989). In other words, the degree to which the new technology is fun to use.

This characterization of technology adoption is consistent with research on retail shopping behaviour. Furthermore, the TAM-model has also been widely used as a base for understanding consumer behaviour in retail and M-commerce settings (Wu & Wang, 2005), where it has supported the presence of both utilitarian and hedonic motivations to use a new retail technology. Within the TAM-model, PU and PEOU of the shopping app can be

thought of as reflecting the more instrumental (utilitarian) aspects of shopping, while PE embodies the more experiential (hedonic) aspect of shopping (Childers et al., 2002). Thus,

(13)

13 while some consumers may be using the shopping apps primarily for instrumental purposes,

others may be primarily enjoying these shopping apps, and thus both factors can ultimately affect their attitude toward using mobile shopping apps. The TAM-model thus already has a proven record in serving as a valid model to predict the acceptance of both e-commerce and m-commerce innovations (Wu & Wang 2005; Ha & Stoel, 2009). Following the existing literature, the presence of the relationships between the three perceptual dimensions of the TAM-model and attitude towards shopping apps are thus also expected in this study.

However, not many academics also consider the direct effects of PU, PEOU and PE on behaviour such as usage, intent to keep using the app and intent to recommend the app to others and the few who did consider this, found contradicting results. On the one hand, Chong (2013) and Wu (2015) state that consumers engage in mobile shopping activities more frequently when they perceive it as useful and easy to use. On the other hand, Ruiz-Mafé, Sanz-Blas and Aldás-Manzano (2009) found no evidence that PU was directly related to usage and continuance intention of shopping via mobile devices, even though PU was found to affect attitude. One explanation for this might be found in the fact that results concerning consumer behaviour in retail tends to differ between retail categories (Childers et al., 2002). This will be accounted for later in this study, but in general, this study predicts that a positive direct trend does indeed exists between PU, PEOU and PE and usage, intent to keep using the app and intent to recommend the app to others. This study thus hypothesizes the following:

H2a. The PU of a shopping app is positively related to a) usage of the app, b) intention to keep using the app in the future and c) intention to recommend

(14)

14 H2b. The PEOU of a shopping app is positively related to a) usage of the app, b) intention to keep using the app in the future and c) intention to recommend the app to others.

H2c. The PE of a shopping app is positively related to a) usage of the app, b) intention to keep using the app in the future and c) intention to recommend the app to others.

2.3 Five core functionalities of shopping apps.

Customers use a variety of app features to perform diverse tasks such as searching, retrieving, and sharing information, passing time with entertainment content, paying bills, and navigating maps (Wang, Kim, and Malthouse 2016). However, although the above already offers valuable information for retailers, research aimed at discovering which functionalities and features of shopping apps are the most important for a shopping app to be perceived as useful, easy to use and enjoyable remains scarce. This is important to investigate, because recent studies show that poor usability of shopping apps is the most important factor in deciding not to download, not to use or to delete an app (Deloitte, 2012; Eshet and Bouwman, 2015; Forrester Research, 2011) and that one of the reasons for poor usability includes inappropriate selection of functionalities of mobile shopping apps (Tarute et al., 2017). In this study, a functionality feature of a mobile app is defined as “an action that can be performed by the user”, which represents consumer’s perception of different functions within a mobile app (Adukaite et al., 2013).

Big differences in types of apps and functionalities may exist between different apps from retailers. However, earlier research often only investigated the effectiveness of

(15)

15 shopping apps, and thus distinguish between functionalities, is made by Bellman et al.

(2011), who indeed argue that consumers’ attitude towards a retailer overall and their purchase behaviour may be explained by the different type of apps. While questioning to what extend apps have an effect on brand attitude and purchase intention, they made a distinction between two general types of branded apps: apps with a focus on delivering informational benefits through incorporating informational functionalities in the app and apps with a focus on experiential benefits. The former focusses on providing informational content to consumers whereas the latter is more focussed on providing enjoyment to customers (Bellman et al., 2011).

Bellman’s (2011) distinction of shopping apps is somewhat related to an evaluation of different mobile shopping activities by Chong (2013). Chong identified four key activities that consumers engage in on their mobile device, which are content delivery, entertainment, location-based and transactions. Although Chong’s activities correspond to shopping on mobile devices in general, they form the base for the five core functionalities of shopping apps in this study. Content delivery is described by Chong (2013) as using a mobile device to search for information, such as product descriptions. Entertainment activities refer to using a mobile device for hedonic purposes such as watching fun videos provided by a brand. Location-based activities include receiving information on the basis of the user’s location such as searching for the nearest stores. Finally, transactional activities refer to activities such as paying for products through a mobile device. Chong’s content delivery relates to Bellman’s typology of informational apps and Chong’s entertainment relates to Bellman’s experiential apps. Adding Chong’s other two shopping activities implies that apps can also

(16)

16 To get a better understanding of the different functionalities incorporated in

shopping apps this study thus combines the findings of Bellman et al. (2011) and Chong (2013) to form the first four core functionalities of shopping apps. However, when it comes to shopping apps there is one seemingly important functionality that Bellman et al. and Chong did not take into consideration because they did not focus solemnly on shopping apps, namely loyalty-based functionalities. One of the reasons why consumers download shopping apps is because they want to be rewarded for their loyalty toward the retailer (Chung, Chun & Choi, 2016). Those loyalty-based functionalities involve receiving loyalty points of the retailer in their mobile app and receiving additional and/or personalized discounts as a reward for the fact that they are a loyal customer.

Based on previous research with a focus on identification of shopping app functionalities and their potential effect on consumer behaviour, this study thus addresses the five main functionalities of mobile shopping apps: informational-, location-based-, experiential-, transactional- and loyalty-based functionalities. The main idea for the explicit choice to look at these five functionalities lies in the fact that successful implementation of these functionalities into mobile shopping apps might increase and foster a positive attitude towards the app, which is an important determinant of the app its success (Childers et al. 2002). In the following subsection, these five functionalities of shopping apps are further elaborated and research hypotheses are formulated accordingly. Since the TAM-model is taken as a base, these hypotheses will relate to the functionalities of shopping apps and its corresponding effect on the app its PU, PEOU and PE.

(17)

17 2.3. The effect of 5 core functionalities of shopping apps on their PU, PEOU and PE.

Several studies indicate the importance of the five proposed functionalities for shopping apps. It is suggested that providing consumers with relevant information about a company its products is one of the most important reasons for consumers to use mobile applications in general (Kennedy-Eden & Gretzel, 2012; Nikou & Mezei, 2013; Tarute et al., 2017). Rewards are also important for consumers to download an app. (Chung et al., 2016). Transactions are considered important because people want to purchase products, also on the go. Furthermore, consumers want a good experience because shopping is fun, which is why experiential functionalities are probably important. Lastly, location-based functionalities offer benefits that only an app can offer, which is why apps offer additional value (Zhao and Balagué, 2015). In the end, consumers’ perception of the benefits that they derive from a shopping app, i.e. PU, PEOU and PE, are dependent of the variety of app functionalities that they use to perform different tasks (Wang et al., 2016). It is thus assumed that if a retailer incorporates one or more of the five functionalities proposed in this study in their app that, when done right, it will increase the PU, PEOU and PE of their app. The according hypotheses of this section are therefore as follows.

H3a. Good informational functionalities of shopping apps are positively related to a) the PU of the app, b) the PEOU of the app and c) the PE of the app.

H3b. Good location-based functionalities of shopping apps are positively related to a) the PU of the app, b) the PEOU of the app and c) the PE of the app.

H3c. Good experiential functionalities of shopping apps are positively related to a) the PU of the app, b) the PEOU of the app and c) the PE of the app.

(18)

18 H3d. Good transactional functionalities of shopping apps are positively related to a) the PU of the app, b) the PEOU of the app and c) the PE of the app.

H3e. Good loyalty-based functionalities of shopping apps are positively related to a) the PU of the app, b) the PEOU of the app and c) the PE of the app.

However, this does not say anything about which functionalities have a bigger effect on PU, PEOU and PU than others. Therefore, a more thorough look at the benefits from each functionality is required. On the one hand of the set of functionalities are the informational, location-based, transactional and loyalty-based functionalities, which all serve the user some sort of utilitarian benefit. That is, they provide the user with functional benefits. On the other hand of the set are the experiential functionalities, which mainly offer the user aesthetic benefits. Since the PE dimension of the TAM-model is the extent to which the activity of using shopping app is perceived to provide value in its own right, besides any performance consequences that may occur (Davis, 1989), it is expected that compared the other functionalities, the experiential functionalities will have the biggest effect on PE of the app. The informational, location-based, transactional and loyalty based functionalities will most likely flow through PU the most.

The following subsections will take a closer look at the possible differences of the effect of these functionalities between different categories where retailers can operate in. More specifically, a moderating role of the nature of the product category (utilitarian vs. hedonic) and the level of consumers’ involvement (low vs. high) with the product category on the effect of the different functionalities on PU, PEOU and PE is predicted. Both distinctions are acknowledged to be key determinants in understanding consumer behaviour

(19)

19 between the functionalities, perceptual dimensions of the TAM-model and attitude differ

between groups.

2.4. The conditional effect of product nature (utilitarian / hedonic).

After a short explanation of product nature and utilitarian and hedonic shopping

motivations, this section will first look at the moderating effect of the nature of the products sold by a retailer on the relationship between the perceptual dimensions of the TAM-model and attitude, and then at its effect on the relationship between the five functionalities and the perceptual dimensions of the TAM-model.

The two main motivations for consumers to buy products are utilitarian and hedonic motivations (Hirschman & Holbrook, 1982). This dichotomy is considered to involve the two principle motivations required to understand consumer behaviour (Babin, Darden & Griffin, 1994) and is often considered as an important moderator (Kleijnen, de Ruyter & Wetzels, 2007). Hirschman and Holbrook (1982) describe consumers as either “problem solvers” or in terms of consumers seeking “fun, fantasy, arousal, sensory stimulation, and enjoyment.” In the former case, consumers’ shopping behaviour is driven by utilitarian motivations. So, they then buy products that are able to fulfil functional needs. In the latter case, consumers’ shopping behaviour is driven by hedonic motivations. They then consume products to fulfil their hedonic needs (Hirschman & Holbrook, 1982). Products can thus be separated in to two product categories: utilitarian products and hedonic products (Dhar & Wertenbroch, 2000)

The dichotomy between utilitarian and hedonic motivation is also applicable to motivations to use a new technology, and thus to the TAM-model (Childers et al., 2002). The first two dimensions of the TAM-model, PU and PEOU assimilate with consumers’ utilitarian

(20)

20 motivations. PU and PEOU both serve the functional needs of consumers. The third element

of the TAM-model, PE, assimilates more with consumer’s hedonic motivation as it derived from the perceived fun or playfulness of the shopping experience (Childers et al., 2002; Lynch, 2014). Concerning this study, the nature of the products sold by a retailer is expected to have an impact on the relationship between three perceptual dimensions of the TAM-model and consumers’ attitude towards the retailers’ app. Because PU and PEOU refer to consumers’ functional needs of a shopping app, it is expected that the effect of PU and PEOU on attitude towards the shopping app is more positive for apps from retailers which sell mostly utilitarian products. The reasoning for this is that when consumers are shopping for utilitarian products, they are looking for convenience and efficiency (Hirschman & Holbrook, 1982; Lynch, 2014) and the shopping app should aid them in achieving this goal. Accordingly, because PE refers to consumers’ hedonic needs of a shopping app, it is expected that the effect of PE on attitude towards the shopping app is more positive for apps from retailers which sell mostly hedonic products. Thus, the nature of the products sold by retailers is expected to serve as a moderator in determining the relationship between the three perceptual dimensions of the TAM-model and the attitude towards the shopping app. The corresponding hypotheses are therefore as follows.

H4a. The effect of PU and PEOU on attitude towards the shopping app is more positive for utilitarian products compared to hedonic products.

H4b. The effect of PE on attitude towards the shopping app is more positive for hedonic products compared to utilitarian products.

This study also predicts that product nature moderates the relationship between the

(21)

21 their study about utilitarian and hedonic shopping motivations and repeat purchase

intention describe product information and convenience as important utilitarian benefits and enjoyment and value as important hedonic benefits. Regarding this study, convenience can be linked to location-based functionalities and transactional functionalities. Value can be linked to the rewards and discounts received, i.e. loyalty-based functionalities, because consumers may feel joyful due to their success in getting discounts (Chiu et al, 2014). Thus, on the one hand, providing consumers with high quality information about the products offered, saving time and effort by using the shopping app because of the possibility to search for the nearest location and purchasing products are thus linked to utilitarian benefits (Chiu et al., 2014). It is therefore likely that these functionalities are more important for retailers selling utilitarian products. On the other hand, because enjoyment and loyalty-based functionalities are linked to more hedonic benefits of shopping, it is more likely that they are more important for retailers selling hedonic products. This study thus hypothesizes the following.

H5a. The effect of informational, location-based and transactional functionalities on PU, PEOU and PE is more positive for utilitarian products compared to hedonic products.

H5a. The effect of experiential and loyalty-based functionalities on PU, PEOU and PE is more positive for hedonic products compared to utilitarian products.

2.5 The conditional effect of consumers’ product involvement.

After a short explanation of consumers’ product involvement this section will first look at the

(22)

22 dimensions of the TAM-model and attitude, and then at its effect on the relationship

between the five functionalities of shopping apps and the perceptual dimensions of the TAM-model.

A common definition of product involvement is ‘the degree of personal relevance, interest and/or subjective feeling of importance of the product category or purchase decision’ (Zaichkowsky, 1985; Petty, Cacioppo & Schumann, 1983). Consumers might be highly involved with one type of product, but very little involved with another. The difference in the consumer decision making process between buying high and low involvement products is explained by Kotler and Keller (2011). In general, high involvement products are often more expensive, not bought frequently, tricky to buy and self-expressive. High involvement products require consumers to first gather information about the product and develop an image about it, before they actually make a carefully considered decision. Contradictory, low involvement products are typically characterized by the fact that consumers easily switch between brands and products and do little evaluation before purchases.

Product involvement is often regarded as an important moderator in predicting purchase decisions (Celsi & Olson, 1988; Suh & Youjae, 2006) and is also considered as an influencer of consumers’ acceptance of online shopping (Lian & Lin, 2008). It is thus assumable that product involvement will also act as a moderator regarding shopping apps. According to the Elaboration Likelihood Model (ELM) (Petty & Cacioppo, 1986), high product involvement results in consumers being more likely to generate more thoughts about the products and the retailer selling the product. If consumers form more positive thoughts

about the usefulness of the channel (shopping app) which they are buying the product from, involved consumers should be more likely to form a positive attitude about the products

(23)

23 offered and about the channel itself. On the other hand, low product involvement results in

consumers that are less likely to attend product information and for those consumers, convenience and entertainment are more important (Elliot & Speck, 2005). Thus, if consumers form more positive thoughts about the ease of use and enjoyment of the shopping app, less involved consumers should be more likely to form positive attitudes about the products offered and the shopping app itself. Accordingly, the corresponding hypotheses are as follows.

H6a. The effect of PU and on attitude towards the shopping app is more positive for consumers buying high involvement products compared to consumers buying low involvement products

H6b. The effect of PEOU and PE on attitude towards the shopping app is more positive for consumers buying low involvement products compared to consumers buying high involvement products

Since more involved consumers are more likely to attach value to information, do more extensive research and thus attend to peripheral content (Petty & Cacioppo, 1986; Kottler & Keller, 2011), it is expected that informational functionalities of shopping apps are more important for retailers selling high involvement products compared to retailers selling low involvement products. However, since high involvement products are often more expensive and tricky to buy (Kottler & Keller, 2011), consumers often prefer to purchase these products at brick-and-mortar stores or on their computer. It is therefore expected that for high involvement products, transactional functionalities of shopping apps are less important than for low involvement products. In the ELM, entertainment elements are more

(24)

24 functionalities in shopping apps should be more important for low involvement products

compared to high involvement products. Rewards and discounts are more useful for products that are bought frequently and are less effective for products that need much consideration (Chandrashekaran & Grewal, 2003) and the same is suggested for location based services. Therefore, this study predicts that loyalty-based functionalities and location-based functionalities in shopping apps should be more important for low involvement products compared to high involvement products. This study thus hypothesizes the following.

H7a. The effect of informational functionalities on PU, PEOU and PE is more positive for high involvement products compared to low involvement products.

H7b. The effect of transactional, experiential, loyalty-based and location-based functionalities on PU, PEOU and PE is more positive for low involvement products compared to high involvement products.

2.7 Conceptual Model.

The predictions in the theoretical framework are visualized in the conceptual model below (Model 1). Variables are grouped in order to provide a clear overview. However, relationships between the individual variables in both groups are expected and will be tested accordingly.

(25)

25 Model 1

Conceptual model.

(26)

26

3. Methodology.

In this chapter, the methodology and procedures that are used before empirically testing the model are described. This section is build up as follows. First, the research design and procedure of data collection will be described. Subsequent will be the data characteristics of

the sample, which is followed by a description of the survey design. A description of the measurements of all the variables will be given before finally explaining how the data will be analysed in order to test the hypotheses.

3.1 Research Design and procedure.

To find out which shopping app functionalities are the most important for consumers to keep using shopping apps and to find out whether and how this differs between different retailers, a cross-sectional survey will be used. This method is the most appropriate, because of the time and budget limit in which this study has to be completed. Using a survey to gather data is valid for this study because the use of a survey enables the possibility to explain how factors are related. A survey makes it easy to compare the answers of a large number of people and is therefore a good strategy for the collection of the data needed to answer the hypotheses in this study (Saunders, Saunders, Lewis & Thornhill, 2012, p. 177). The internet-mediated method is used to develop the survey. An online survey will be used because it is easy to use for respondents and because there will be no need to manually enter the gathered data afterwards, it will save much time (Sauders et al. 2012, p. 422). A survey also makes it easy to quickly reach different people (Saunders et al., 2012, p. 177). The survey will be self-administered because a survey is completed faster when respondents fill in the questions themselves. The survey used in this study will allow for this because most of the questions will be scaling questions.

(27)

27 Because of the time pressure and limited budget of this study, a combination of

convenience sampling and snowball sampling will be used. The main focus will be on requesting acquaintances to participate in this survey. Subjects will be approached by means of social media and e-mail. Secondly, the researcher will also invite visitors of the Kalverstraat, the biggest shopping district in Amsterdam, to participate in the survey if needed. This increases the external validity of this study, because by doing so not all participants will be (in)directly acquainted to the researcher, which leads to a larger diversity in participants. As a result, the average age of the participants will be relatively low compared to the population. This could be seen as a limitation. On the other hand, the percentage of retail app users will be higher among younger people compared to older ones. In the end, this sampling method results in participants from a restricted sample, which limits the generalization of this study. This point will also be highlighted in the limitations section of this study.

3.2 Sample.

In total, 239 respondents finished the survey, but 39 respondents were excluded for further analysis because they either did not have a shopping app on their mobile phone or tablet or

because they failed to successfully answer the attention checks. Furthermore, 37 respondents evaluated the app of an online retailer, which is not the focus of this study. Those responses were therefore also removed from the dataset. The remaining 163 responses will be used for the analysis.

(28)

28 3.3 Survey Design.

The online survey will be established by using the online survey tool Qualtrics. To increase the quality, the internal validity and the response rate of the final survey, it was pilot tested with a small group (N=20) of respondents. Some adjustments in formulations of questions were applied after their feedback. There will be a total of 27 questions in the final survey. An overview of the survey, its questions and the items used for each construct can be found in Appendix A. The survey is structured as follows.

After respondents click on the link of the survey, they are first provided with information about the nature of the study including the time it takes to complete the survey, the anonymity of the survey and the option to win a prize after completion of the survey. After they agree with this, they can continue with the survey. First of all, respondents are introduced to the concept of shopping apps. Examples of shopping apps are given so that it is clear for respondents what shopping apps are. To make sure respondents can qualify for this study, they are asked whether they do or do not have a shopping app on their smartphone or tablet. If they answer this question with ‘yes’, they can continue with the survey. If they answer this question with ‘no’, they are directed out of the survey since they do not qualify for the sample. Then, respondents are asked to mention all the shopping apps that they have on their mobile devices. Afterwards, one of these apps is randomly selected by Qualtrics. Respondents have to indicate how often they use the app per month, how likely they are to recommend it to others and how likely they are to keep using the app in the future. The survey continues with questions concerning how the respondents perceive the app that they have to evaluate. First attitude towards the app is measured, followed by

the perceptual dimensions from the TAM-model. Then, consumers are asked how they perceive the functionalities of the shopping app. In the final two questions, respondents are

(29)

29 asked about their age and gender. The survey concludes with a thank you note and the

possibility to leave an e-mail address to have a chance at one of the rewards.

3.4 Measurements.

Usage, intention to recommend, intention to keep using a shopping app, attitude towards the app and the perceptual dimensions from the extended TAM-model (PU, PEOU and PE) are the dependent variables in this study. These variables will be measured by using validated questions which are adopted from earlier studies. The perceptual dimensions of the TAM-model will be tested in relation with the functionalities of shopping apps. These functionalities are divided under five dimensions: experiential, informational, location-based, loyalty-based and transactional functionalities. The nature of the products (utilitarian vs. hedonic) sold by the retailer and consumers’ level of involvement (low vs. high) with these products are expected to act as a moderating variable in this model. A precise description of how the above-mentioned variables are measured is given in the following subsections.

3.4.1 Usage, intention to recommend and intention to keep using the app.

Usage of the shopping app is measured on a per month basis. Consumers are asked to give an estimation of how many times they had used the app in a month. Behavioural intention to recommend the app to others is measured with a 10 point Likert item ranging from 0 (very unlikely) to 10 (very likely). A high score indicates high recommendation intention. The question that is asked is: “please indicate how likely you are to recommend the app of (retailer) to family or friends”. This item has been adopted from Reichheld (2003). Behavioural intention to keep using the app is also measured with a 10 point Likert item

(30)

30 ranging from 0 (very unlikely) to 10 (very likely). A high score indicates high intention to keep

using the app. The question that is asked is: “please indicate how likely you are to keep using the app of (retailer) in the future”. This item has been adopted from Quix and van der Kind (2012).

3.4.2. Attitude towards the shopping app of the retailer.

Attitude is measured by asking respondents the question: “to what extend do you agree with the following statement”. The construct consists of three elements, which have been adopted from Davis (1989). The items are measured on a 7 point Likert scale ranging from 1 (totally disagree) to 7 (totally agree). A high score indicates a favourable attitude towards the shopping app of the retailer. An Example is: I find using this app positive.

3.4.3. TAM-variables.

The extended TAM-model used in this study consists of the following three elements: Perceived usefulness, Perceived ease of use and perceived enjoyment. These elements are measured by using the scale that Chong (2013) also used in his research. This scale is based on the original TAM scale from Davis (1989) and UTAUT2 scale from Venkatesh et al. (2012),

but is adjusted slightly so that it fits with m-commerce. Respondents have to indicate to what extent they agree with statements regarding to the shopping app they use. Examples of items are: “This app improves my performance while shopping” (perceived usefulness), “Learning how to use this app is easy for me” (perceived ease of use) and “Using this app is fun” (perceived enjoyment). Each element is measured with four items on a 7-point Likert scale ranging from 1 (totally disagree) to 7 (totally agree). A high score indicates that the app is perceived as either very useful, very easy to use or very enjoyable to use.

(31)

31 3.4.1 Functionalities of the shopping app.

The functionalities of shopping apps are the independent variables. This construct is divided in five dimensions. Those five dimensions are: experiential, informational, location-based, loyalty based and transactional features. This division has already been verified by (Deen, 2015), who based them on the findings regarding m-commerce activities of studies from Bellman et al. (2011) and Chong (2013). The quality of the functionalities was measured with aa 7 point Likert item ranging from 1 (totally disagree) to 7 (totally agree). Answers for these functionalities are gathered by asking the respondents the following question: “Please indicate to what extent you agree with the following statements with regard to the app.” An example of such a statement is: “This app provides good informative features. For example: searching for products, searching for product information, reading reviews, comparing product prices.”

3.4.5. Moderating variables – product nature and product involvement.

The product nature and product involvement are considered as the moderating variables in this study. Both variables are measured using a dichotomous scale. The nature of a product category is divided in utilitarian products and hedonic products and thus measures whether a product category is purchased because of consumers’ utilitarian versus hedonic motivation. Products involvement is divided in low involvement products and high involvement product and thus measures to what extend consumers are involved with the product they are buying based on the importance of these products for consumers and the impact of these products on their lives. The level of product involvement and the product

(32)

32 gave each app a 0 (indicating low involvement / utilitarian nature), or a 1 (indicating high

involvement / hedonic nature).

3.4.6. Demographics.

Last but not least, simple demographics are also accounted for to be able to say something about the generalization of this study. Derived from Saunders et al. (2012), gender is measured on a binary scale in which “1” is male and “0” is female. Age is measured on a nominal interval scale, as it is measured by date of birth.

3.4.7. Data analysis.

Various methods of analysis will be used to test the relationship between the variables. Multivariate (multiple) regressions will be used to explain the relationship between 1) the

shopping app functionalities and the perceptual dimensions of the TAM-model and 2) the relationship between the perceptual dimensions of the TAM-model and attitude, usage, intention to recommend and intention to keep using. Multivariate multiple regressions will be more efficient in this study compared to performing regressions for each dependent variable separately because a correlation structure among the dependent variables is present (Hartung & Knapp, 2014). To check for the moderating effects of product nature and product involvement, Hayes (2012) PROCESS tool will be used. This is an appropriate analytical tool to measure the conditional effects of product nature and product involvement with the help of bootstrapping method (n=5000). Because product nature and product involvement are dichotomous variables, the independent variables in each model will be standardized. PROCESS model 1 will be used to test the conditional effect of product nature and product involvement on each relationship, while controlling for the other

(33)

33 independent variables that are present. By doing so, repeating multiple models with

different independent variables is a valid method for generating the conditional effect of product nature and product involvement for all the relationships between the various independent and dependent variables (Hayes, 2012, pp. 196 - 197).

4. Results

The results section is structured as follows. First, an overview of the general findings is given, followed by a reliability and factor analysis. Second, the effect of attitude on usage, intention to recommend and intention to keep using is investigated. Third is the investigation of the effect of the three perceptual dimensions of the TAM-model on attitude, usage, intention to recommend and intention to keep using. Fourth, the hypothesized relationship between the five shopping app functionalities and the perceptual dimensions of the TAM-model are examined. Fifth and final is the analysis of the moderating effect of product involvement and product nature. At the end of the result section an overview of which hypotheses are accepted and rejected is given instead of mentioning them through the different sub-sections of the result section for the purpose of readability. The results section is concluded by a visualization of discovered effect by incorporating their Beta’s and significance levels in the conceptual model as shown in the end of the theoretical framework.

4.1 General findings.

In total, the data derived from 163 respondents has been used for the analysis. The average age of the respondents was 27.31 years old (SD = 11.25) and 67% of the respondents was female. Amongst these respondents, the average amount of shopping apps that they had on

(34)

34 their phone or tablet was 2.63 (SD = 1.65). This resulted in a dataset containing 45 different

evaluated shopping apps.

4.2 Reliability analysis.

To test whether the scales used to measure the aforementioned variable can be used for a regression analysis, a reliability analysis is conducted. All the scales were considered to have good Cronbach’s Alpha’s (α > .80). The scale measuring attitude had a Cronbach’s Alpha of 0.848. The PU scale recorded a Cronbach’s Alpha of 0.826 and the scales of PEOU and PE recorded a Cronbach’s Alpha of respectively 0.804 and 0.898. The corrected item-total correlations indicate that all the items have a good correlation with the total score of the scales (all above .30). Also, none of the items would have substantially (Δ > .10) affected the reliability of the scales if they were deleted. The correlation table (table 2) further on in this chapter shows the Cronbach’s Alpha’s of the scales on the diagonal.

4.3 Factor analysis.

To evaluate the goodness of the scales in the measurement model and to verify that the scales do indeed measure different things, a factor analysis is conducted. The Kaiser–Meyer– Olkin measure verified the sampling adequacy for the analysis, KMO = .850. Bartlett’s test of sphericity χ² (66) = 1103.34 p < .001, indicated that correlations between items were sufficiently large for PAF. An initial analysis was run to obtain eigenvalues for each component in the data. Three components had eigenvalues over Kaiser’s criterion of 1 and in combination explained 59.72% of the variance. However, the scree plot revealed a

possible alternative solution, because the logical cut-off point seemed to be at two factors. Since the literature is in line with the results from the eigenvalues, three factors were retained and rotated with an Oblimin with Kaiser normalization rotation. Table one shows

(35)

35 the factor loadings after rotation. The items that clustered on the same factors suggested

that factor one represents PE, factor two represents PEOU and factor three represents PU. The third item of PU had a relatively low loading and also showed cross-loadings on the factor of PEOU. This could be because of the content of the item because it has the word ‘convenience’ in it, which might be linked to ease of use by respondents. Because of the low- and cross-loading of this item, dropping this item was considered. In the reliability analysis, this item was the only item that, if deleted, would increase the reliability of the scale with 0.5 (from .77 to .826). This difference alone was not directly a reason to drop the item, but combined with the results from the factor analysis, this item was dropped for further analysis. The remaining scale of PU still had three items in it and was thus still valid.

4.4 Correlation analysis.

Before testing the hypotheses, a correlation analysis between all the relevant variables in this study was conducted together with an analysis of the descriptive statistics. Table two

Table 1

Factor analysis of scales.

Item PE PU PEOU PU1 0.04 0.78 -0.03 PU2 0.03 0.90 -0.02 PU3 0.01 0.20 0.26 PU4 0.01 0.57 0.20 PEOU1 0.11 -0.06 0.69 PEOU2 0.34 0.07 0.45 PEOU3 0.04 0.08 0.61 PEOU4 -0.06 0.03 0.83 PE1 0.95 -0.06 -0.05 PE2 0.75 -0.01 0.15 PE3 0.76 -0.02 0.21 PE4 0.72 0.22 -0.12

Note. Factor loadings above .40 appear in bold

(36)

36 provides an overview of the mean value and standard deviation of each variable. All the

correlations between each variable are shown and if applicable, the Cronbach’s Alpha’s are mentioned on the diagonal.

As expected, it turns out that attitude towards the app was positively correlated with usage r(200) = .24, p < .01 (weak), intention to recommend r(200) = .54, p < .01 (strong) and intention to keep using r(200) = .66, p < .01 (strong). Also according to predictions, perceived usefulness r(200) = .53, p < .01 (strong), perceived ease of use r(200) =.53, p < .01 (strong) and perceived enjoyment r(200) = .66, p < .01 (strong), were positively correlated with attitude towards the app. Furthermore, it turns out that good informational functionalities were positively correlated with perceived usefulness r(200) =.30, p < .01 (moderate), perceived ease of use r(200) = .38, p < .01 (moderate) and perceived enjoyment r(200) = .40, p < .01 (moderate). Good location-based functionalities were positively correlated with perceived usefulness r(200) = .17, p < .05 (weak), but not with perceived ease of use r(200) = .13, ns and perceived enjoyment r(200) = .05, ns. Good experiential functionalities were positively correlated to perceived ease of use r(200) = .18, p < .05 (weak) and perceived enjoyment r(200) = .43, p < 0.01 (moderate), but not with perceived usefulness r(200) = .12, ns. Good transactional functionalities were positively correlated with perceived usefulness r(200) = .37, p < .01 (moderate), perceived ease of use r(200) = .26, p < .01 (weak) and perceived enjoyment r(200) = .26, p < .01 (weak). Finally, good reward-based functionalities were positively correlated with perceived usefulness r(200) = .19, p < .01 (weak), but not with perceived ease of use r(200) = .09, ns and perceived enjoyment r(200) = .14, ns.

(37)

. Table 2

Descriptives and correlations between variables (Cronbach's Alpha's on the diagonal).

Variables M SD 1 2 3 4 5 6 7 8 9 10 11 1. Usage 3,81 4,21 2. Recommendation 6.77 2,06 .22** 3. Repeat behaviour 7.50 1,97 .24** .69** 4. Attitude 5.43 1,01 .24** .54** .66** (.848) 5. PU 4,8 1,23 .24** .40** .41** .53** (.826) 6. PEOU 5,75 0,86 .08 .35** .43** .53** .53** (.804) 7. PE 4,89 1,11 .5 .47** .52** .66** .50** .61** (.898) 8. Informational 5.43 1,26 .18** .23** .31** .42** .30** .38** .40** 9. Location-based 4.73 1,57 .07 .13 .16 .07 .17* .13 .05 .30** 10. Experiential 3.64 1,544 .03 .20** .15** .23** .12 .18* .43** .29** .15* 11. Transactional 5.22 1,52 .22** .28** .36** .41** .37** .26** .26** .39** .09 .20** 12. Loyalty-based 5.23 1,75 .04 .04 .10 .08 .19** .09 .14 .15 .09 .21** .10

(38)

4.5 The effect of attitude on usage, intention to recommend and intention to keep using. To start the analysis, the effect of attitude on usage, intention to recommend and intention to keep using was tested by using a multivariate regression analysis. The results of the regression analysis supported the predicted effect of attitude on usage (B = 1.033, SE = .315, p < .01), intention to recommend (B = 1.182, SE = .134, p < .01) and intention to keep using (B = 1.289, SE = .115, p < .01). Thus, the more positive consumers’ attitude towards the shopping app, the more likely it is that they 1) use the app more often, 2) recommend the app to friends and family and 3) will keep using the app in the future. An overview of these results can be found in table three.

4.6 The effect of the TAM-model on attitude, usage, intention to recommend intention to keep using.

Next, the relationships between the perceptual dimensions of the TAM-model and attitude

are tested. Since there is one DV and three IV’s, a multiple linear regression was used to test the relationships between those variables. The model was statistically significant, recording F (3, 159) = 53.91 p < .01, and explained 50% of the variance in attitude. The result showed the predicted effect of PU and PE. PU thus had a significant on attitude (B = .212, SE = .058, p < .05) and the same could be said for the effect of PE on attitude (B = .435, SE = .067, p < .05). Unexpectedly, PEOU did not significantly affect attitude (B = .435, SE = .067, ns). In other words, if consumers’ attitude towards the shopping app increases with one, PU and PE Table 3

Regression results of the effect of attitude on usage, recommendation and repeat behaviour.

Coefficient SE Beta Coefficient SE Beta Coefficient SE Beta

Attitude 1.033** .315 .250 1.182** .134 .570 1.289** .115 .663

R2 .063 .325 .439

Note. *p<.05. ** p<.01.

(39)

39 increase with respectively 0.212 and 0.435, but PEOU remains the same. However, PEOU is

still an important facet for consumer attitude towards shopping apps because even though it did not have a direct significant effect on attitude, it did significantly affect PU (B = .731, SE = .091, p < .05). An overview of the results can be found in this section can be found in table four.

Additionally, it was also checked whether there was a direct relationship between the perceptual dimensions of the TAM-model and usage, intention to recommend and intention to keep using. The results of a multiple multivariate regression analysis suggested that there was a small positive effect of PU (B = .045, SE = .020, p < .05) and a small negative effect of PE (B = -.037, SE = .018, p < .05) on usage. Regarding intention to recommend, none of the perceptual dimensions of the TAM-model were found to have a significant effect. Regarding intention to keep using however, all three of them did, with PE (B = .264, SE = .053, p < .01) recording a higher B value than PEOU (B = .171, SE = .046, p < .01) and PU (B = .151, SE = .046, p < .01). In other words, if PU increases with one, usage and intention to keep using the app increase with respectively 0.045 and 0.151. Also, if PEOU increases with one, intention to keep using the app increases with 0.171. Finally, if PE increases with one intention to keep

Table 4

Regression results of the effect of the TAM-model constructs on attitude.

Coefficient SE Beta Coefficient SE Beta

PU .212** .058 .249 - - -PEOU .112 .088 .096 .731** .091 .534 PE .435** .067 .479 - - -R2 .504 .285 Note. *p<.05. ** p<.01. Attitude PU

(40)

40 using the app increases with 0.264. PE was thus the strongest driver of attitude. An overview

of these results can be found in table five.

4.7 The effect of the shopping app functionalities on the TAM-constructs

Multivariate multiple regression analysis was performed to investigate the effects of the shopping app functionalities on PU, PEOU and PE. The individual models were all statistically significant with PU recording F (5, 157) = 8.79 p < .01, PEOU recording F (5, 157) = 6.47 p < .01 and PE recording F (5, 157) = 13.38 p <0.01. The models explained 21.9% of the variance

in PU, 17.1% of the variance in PEOU and 29.9% of the variance in PE.

Regarding PU, only the effects of informational and transactional functionalities were statistically significant, with transactional functionalities (B = .248, SE = .058, p < .01) recording a higher B value than informational functionalities (B = .172, SE = .075, p < .05). Non-significant effects were found for location based functionalities (B = .070, SE = .059, ns) experiential functionalities (B = -.054, SE = .059, ns), and loyalty based functionalities (B = .077, SE = .050, ns).

Concerning PEOU, only one type of functionalities was found to be statistically significant, namely informational functionalities (B = .218, SE = .057, p < .01). Non-significant effects were found for location based functionalities (B = .032, SE = .045, ns), experiential

Table 5

Regression results of the effect of the TAM-model constructs on attitude.

Coefficient SE Coefficient SE Coefficient SE

PU .045* .020 .111 .057 .151* .061

PEOU -.013 .015 .037 .042 .171** .046

PE -.037* .018 .091 .049 .264** .053

R2 .226 .199 .341

Note. *p<.05. ** p<.01.

Retail applications : what does it take for a shopping app to be successful?