Adoption of Smart Home Speakers

Explorational Study of User Acceptance Factors

Thèse de Master , 2021 , 104 Pages , Note: 1.00

Gestion d'entreprise - Généralités

Due to technological progress, smart speakers have entered the market in recent years and have already been adopted by certain parts of the population. Although several studies have tried to explain the crucial factors for smart speaker adoption, there is still no common understanding and there are significant gaps in the literature analysing the adoption of smart speakers at their current technical level in specific countries. Therefore, this study aims to identify factors influencing the intention to use smart speakers in Germany.

By reviewing the literature on innovation acceptance with an emphasis upon adoption studies of voice-controlled technologies, the research model of this thesis was developed. The hypothesized effects of the twelve constructs in the model were tested using an online survey with questionnaires as data collection instruments among 233 German non-adopters. The study found that certain constructs specifically enjoyment, privacy concerns, usefulness and social influence were significant predictors of intention to use. System quality and smart speaker complementarity with apps and services additionally emerged as predictors of perceived usefulness. Results suggest that companies should pay particular attention to these factors when marketing and further developing smart speakers. The factors identified might further influence the acceptance of similar IT products.

Table of contents

List of figures

List of tables

List of abbreviations

List of mathematical abbreviations

1 Introduction
1.1 Problem statement
1.1.1 Theoretical perspective
1.1.2 Practical perspective
1.2 Research question and research objectives
1.3 Methodology
1.3.1 Theoretical background
1.3.2 Research method
1.4 Content overview

2 Literature review
2.1 Voice operated technology
2.1.1 Voice assistants
2.1.2 Smart speakers
2.1.3 The German smart speaker market
2.2 Definition of terms
2.2.1 Information technology
2.2.2 Innovation
2.2.3 Adoption
2.2.4 User acceptance factors
2.3 Adoption theories
2.3.1 Diffusion of Innovation Theory
2.3.2 Technology Acceptance Model
2.3.3 Unified Theory of Acceptance and Use of Technology 2
2.4 Identification of acceptance and usage relevant factors
2.4.1 Factors considered in IT innovation adoption studies
2.4.2 Factors considered in VA/smart speaker adoption studies
2.4.3 Findings of previous IT, VA and smart speaker adoption studies
2.5 Research model and hypotheses development
2.5.1 Hypotheses on factors affecting the usage intention
2.5.2 Hypotheses on factors affecting perceived usefulness

3 Empirical study
3.1 Objectives of the study
3.2 Research method and design
3.2.1 Choice of research method
3.2.2 Sampling
3.2.3 Data collection
3.3 Pre-test
3.4 Data analysis
3.4.1 Method of analysis
3.4.2 Assumptions of multiple linear regression
3.4.3 Applied steps within the data analysis

4 Empirical findings
4.1 Presentation of results
4.1.1 Descriptive statistics of the survey results
4.1.2 Multiple linear regression analysis 1 - effects on perceived usefulness
4.1.3 Multiple linear regression analysis 2 - effects on the usage intention
4.2 Hypotheses validation
4.3 Limitations of study

5 Discussion
5.1 Summary of results
5.2 Interpretation of results
5.2.1 Interpretation of the effects on perceived usefulness
5.2.2 Interpretation of the effects on the usage intention
5.2.3 Interpretation of the descriptive results
5.3 Implications of research for theory and practice
5.3.1 Contributions and implications for theory
5.3.2 Implications for practice
5.4 Limitations of research
5.5 Directions for future research

6 Conclusion

References

Appendix

Abstract

Due to technological progress, smart speakers have entered the market in recent years and have already been adopted by certain parts of the population. Although several stud­ies have tried to explain the crucial factors for smart speaker adoption, there is still no common understanding and there are significant gaps in the literature analysing the adoption of smart speakers at their current technical level in specific countries. There­fore, this study aims to identify factors influencing the intention to use smart speakers in Germany. By reviewing the literature on innovation acceptance with an emphasis upon adoption studies of voice-controlled technologies, the research model of this thesis was developed. The hypothesised effects of the twelve constructs in the model were tested using an online survey with questionnaires as data collection instruments among 233 German non-adopters. The study found that certain constructs specifically enjoyment, privacy concerns, usefulness and social influence were significant predictors of intention to use. System quality and smart speaker complementarity with apps and services addi­tionally emerged as predictors of perceived usefulness. Results suggest that companies should pay particular attention to these factors when marketing and further developing smart speakers. The factors identified might further influence the acceptance of similar IT products.

Words: 200

Keywords: smart speaker, voice assistant, adoption, acceptance

List of figures

Figure 1 Smart Speaker Adoption Development in Germany

Figure 2 Perceived Characteristics of Innovating

Figure 3 Research Model for Predicting Smart Speaker Acceptance in Germany

Figure 4 Descriptive Statistics of the Sample’s Age Distribution

Figure 5 Results of MLRA1 and MLRA2 and Revised Research Model

List of tables

Table 1 List of Factors Considered in the Individual Level IT Adoption Literature

Table 2 Clustered VA/Smart Speaker UAFs Considered in Previous Studies

Table 3 Descriptive Statistic of the Sample’s Characteristics

Table 4 Descriptive Statistics of the Sample’s Interest in Smart Speaker Functions...

Table 5 Descriptive Statistics of Sample’s Awareness of Smart Speaker Brands

Table 6 Descriptive Statistics of the Research Model’s Constructs

Table 7 Correlation Matrix of the Research Model’s Constructs

Table 8 Regression Analysis Summary for MLRA1

Table 9 Regression Analysis Summary for MLRA2

Table 10 Hypotheses Validation

Abbildung in dieser Leseprobe nicht enthalten

1 Introduction

Smart speakers are on the rise. Since the release of one of the first voice-empowered home speakers, the Amazon Echo, in 2014 (Keates, 2018, para. 4), sales have in­creased rapidly (Statista, 2021a). At the beginning of 2020, almost one-fourth of the Ger­man population (Beyto Ltd, 2020, p. 16) and even one-third of households in the United States (US) owned at least one smart speaker (Kinsella & Mutchler, 2020, p. 3), making this digital innovation a unique success story in the field of consumer electronics (Subirana et al., 2018, p. 9).

As companies could benefit significantly from the smart speaker market, not only through high profits from the sales of the devices but especially from tying customers more closely to their smart home ecosystems (Su et al., 2018, as cited by Fang & Fu, 2020, p. 554) and gaining valuable customer data (Cheng et al., 2019, pp. 145-146) it is of crucial importance for these enterprises to ensure that smart speaker adoption continues to rise and that the market potential of this innovation is fully exploited. However, despite its significant initial growth, the penetration rate of smart speakers, compared to other household Information Technology (IT) devices like mobile phones, notebooks, and con­nected televisions, is still relatively low, and saturation is far from in sight (Marketing Charts LLC, 2020). In order for smart speakers to penetrate the mainstream market, it is required that the broad majority of the population, whose characteristics differ signifi­cantly from the relatively early adopters, who are in general rather technically affine or find it easy to grasp and appreciate the benefits of new technologies, adopt this innova­tion within the next few years (Moore & McKenna, 1991, pp. 9-36). Since the perception of certain innovation’s characteristics are closely linked to its adoption (Rogers, 2003, pp. 16-17), the question arises: which factors affect the intention to use smart speakers among current non-adopters?

Besides its economic importance this thesis has two additional rationales. The findings may contribute to greater use of smart speakers, which might ultimately lead to more advanced smart speaker systems and consequently to more efficient communication between machines and humans. Furthermore, the study’s findings might enrich the tech­nology acceptance literature in general and provide insights about relevant adoption fac­tors that are not only capable of predicting the adoption of smart speakers among later adopter categories but also similar voice-enabled or smart-home technologies.

1.1 Problem statement

In the following, the relevant research gap within the currently existing studies on smart speaker acceptance and adoption is identified. While the gap within the existing literature is defined in the theoretical perspective, it is argued within the management perspective in more detail why the theme is relevant from a business perspective.

1.1.1 Theoretical perspective

Different factors affect the decision of an individual to adopt IT innovations. Researchers like Rogers (2003, pp. 16-17), Davis (1989, p. 333) and Venkatesh et al. (2012, pp. 170­171) already identified crucial factors that affect the intention towards the use of innova­tions or IT. The results of previous studies were among others that especially perceived product characteristics such as the perceived usefulness and ease of use, characteris­tics of the individual like age, gender and experience and the influence within one’s social system impact the intention to use a certain IT innovation (Venkatesh et al., 2012, pp. 169-171).

Due to their technical and economic potential, factors influencing the acceptance or adoption of the innovation smart speaker and its voice empowered software agent called Voice Assistant (VA) have already been quantitatively and qualitatively investigated by several researchers in different countries (e.g., Arifin, 2020; Burbach et al., 2019; Cha et al., 2021; Chu, 2019; Coskun-Setirek & Mardikyan, 2017; Han & Yang, 2018; Hoffmann & Thuesen, 2018; Kääriä, 2017; Kessler & Martin, 2017; Koon et al., 2020; Kowalczuk, 2018; Lau et al., 2018; Liao et al., 2019; Ling et al., 2021; McLean & Osei-Frimpong, 2019; Moussawi et al., 2020; Nasirian et al., 2017; Pal, Arpnikanondt, Funilkul & Chut- imaskul, 2020; Pal, Arpnikanondt, Funilkul & Razzaque, 2020; Park et al. 2018; Pitardi & Marriott, 2021; Yang & Lee 2019). These findings, among others, were that perceived relative advantage (Ling et al., 2021, p. 5), perceived usefulness (Coskun-Setirek & Mardikyan, 2017, p. 11; Pal, Arpnikanondt, Funilkul & Razzaque, 2020, p. 12), an ap­pealing design (Ling et al., 2021, p. 5; Park et al., 2018, pp. 2126-2127) and platform- related variables such as the perceived network size (Park et al., 2018, pp. 2126-2127) were often found to directly or indirectly positively influence the intention to buy or to use smart speakers, while concerns about privacy, data security and trust negatively influ­enced the adoption intention in some studies (Chu, 2019, pp. 27-28; Han & Yang, 2017, pp. 629-630; Kowalczuk, 2018, p. 426; Nasirian et al., 2017, p. 7).

Although there have been many attempts in the literature to explain smart speaker adop­tion, there is still no common understanding about the factors affecting the intention to use these devices. Previous researchers only considered a small number of potential factors and hardly built on previous studies’ findings in this field. Moreover, there is a lack of studies analysing factors that influence the adoption intention of smart speakers at their current technology level. Thus, since many studies were conducted already several years ago, their findings might no longer be valid as the innovation smart speaker has continued to evolve over recent years. Furthermore, as already stated by Tornatzky and Klein (1982, pp. 28-29) are the results about factors affecting the adoption of a spe­cific innovation not generalisable across different cultures or sites, a limitation that pre­vious researchers in this field have also mentioned (e.g., Han & Yang, 2017, p. 631; Ling et al., 2021, p. 7; Pal, Arpnikanondt, Funilkul & Razzaque, 2020, p. 17).

Only one study has been found that analysed the acceptance of smart speakers in Ger­many. This study of Kowalczuk (2018, p. 423) was based on a survey conducted in June 2017, when smart speakers were available on the German market for barely a year (Hobson, 2016, para. 1) and where the knowledge about these products among the pop­ulation was rather limited. As smart speakers were hardly adopted in 2017 (Strategy Analytics 2019, as cited by Kinsella, 2019a) the study only analysed the factors that affected the acceptance of smart speakers among the category innovators, a rather small and in general technically affine adopter category according to Rogers (2003, pp. 281-282). It did not focus on the factors that affect the usage intention among later adopters something that has already been criticised by Verdegem and De Marez (2011, p. 411) about most adoption studies in general.

Also, the smart speaker segment itself has evolved since 2017. The available voice apps for smart speakers of Amazon, for example, have increased from 10,000 at the beginning of 2017 to over 100,000 at the end of 2019 (Statista, 2019a), many new smart speaker providers like Apple or Sonos have entered the market since 2017 (Statista, 2021b) and the software (Fang & Fu, 2020, pp. 555-556) and hardware of the devices has continu­ously improved. Hence, as already pointed out by Kessler and Martin (2017) there is a need to analyse the factors that affect the user acceptance of smart speakers “once the adoption has advanced to later stages on the adoption curve” (p. 70). Thus, at the current technology stage, there is a significant gap in the existing literature about factors affect­ing the intention to use smart speakers among current non-adopters in Germany — a market that is, due to its high gross domestic product, especially attractive.

1.1.2 Practical perspective

As demand for smart speakers has increased substantially during the last few years (Canalys, 2020, para. 5), competition has intensified. Dominated by Amazon in 2014, especially companies like Alibaba, Apple, Baidu, Google, Sonos, and Xiaomi were able to gain market share by offering their own smart speakers in recent years (Fang & Fu, 2020, p. 555; Statista, 2021b). The reason for the attractiveness of this market lies not only in the considerable revenue potential of the industry, estimated to reach United States Dollar (USD) 35.5 billion in 2025 (Statista, 2019b), but especially in the oppor­tunity to attract customers to companies’ platforms, tying them more closely to their ser­vices, create smart home ecosystems (Su et al., 2018, as cited in Fang & Fu, 2020, p. 554) and use the gathered data in the form of conversation recordings to improve speech recognition algorithms and other offerings (Cheng et al., 2019, p. 146). Gaining shares in the smart speaker market is of such importance that companies like Amazon and Google might even accept neutral or negative margins from the sales of their smart speakers as assessed by estimations from Nellis and Dave (2018, para. 13).

Compared to other IT innovations, the smart speaker market is far from being saturated (Marketing Charts LLC, 2020). Smart speakers have the potential to become as wide­spread in the average household as televisions and notebooks are today. However, for the potential of this market to be truly realised, the adoption of this device requires to be further promoted. For evidence-based management practices (see Bryman & Bell, 2011, p. 6) information about factors affecting the intention to use these devices among current non-adopters is therefore urgently desired. Insights about smart speaker adoption fac­tors are especially valuable for smart speaker manufacturers and marketers as they al­low them to better design their products according to market needs, to successfully dif­ferentiate themselves from competitors, and to ensure that the devices are actively used by their owners, generating valuable data for the smart speaker service providers. For marketing, this offers the opportunity to target the most relevant adoption factors, be it ease of use, trust within the service provider or an appealing product design, and thus increase not only initial purchases of the devices but also their, in this case more im­portant, actual usage.

1.2 Research question and research objectives

Based on the gap in the existing literature and its practical and academic relevance, this thesis addresses the following research question:

What user acceptance factors affect the intention to use smart speakers in Germany?

The term User Acceptance Factors (UAFs) within the research questions mainly relates to the perception of using smart speakers, but is not exclusively limited to their perceived characteristics. It may also include, for instance, the effect of the social environment or the presence of facilitating conditions on the usage intention. Within the adoption litera­ture, researchers like Hameed et al. (2012. p. 373) summarise under the term UAFs, factors that directly or indirectly affect the usage intention of new technologies or inno­vations in general. Within this thesis, however, the effects of socio-economic conditions such as age, gender, education, or technology experience are excluded from the term UAFs.

By answering the research question, this thesis overall aims to derive a set of factors that affect the intention to use smart speakers among current non-adopters in Germany. To answer the research question, the overall research aim is broken down into specific Research Objectives (ROs) (Biggam, 2015, pp. 67-68). The four ROs are:

i. Descriptive objective

RO1: To elaborate on factors from the literature that potentially affect the intention to use smart speakers.

RO2: To derive hypotheses on how these factors affect the intention to use smart speakers among current non-adopters in Germany.

ii. Analytical-empirical objective

RO3: To empirically test the hypothesised effects between several, through a literature review identified user acceptance factors on the intention to use smart speakers among current non-adopters in Germany.

iii. Prescriptive-normative objective

RO4: To derive a set of factors that affect the intention to use smart speakers among current non-adopters in Germany together with practical recommen­dations to further promote the diffusion of these devices.

1.3 Methodology

In this section the theories and literature on which the thesis is based are introduced, followed by a description of the applied research method.

1.3.1 Theoretical background

To answer the research question, this thesis considers innovation adoption theories and user acceptance models. As already stated by Fichman (2001, p. 3) there is no single theory of innovation adoption. Instead, there are various theories and models describing factors influencing an individual’s or organisation’s decision or intention to adopt a tech­nology or innovation (Hameed et al., 2012, p. 362). Due to the complexness of behav­ioural science, no single theory or model is able to cover all factors that impact the adop­tion decision. Therefore, models and theories differ in the influencing factors considered and have their specific advantages and limitations (Momani et al., 2017, p. 8). Within this thesis, the analysed factors hypothesised to influence the smart speaker usage intention are not taken from a single theoretical model, but are obtained through a literature review of different adoption and acceptance theories. Among the most popular theories for (IT) innovation adoption are the Diffusion of Innovation (DOI) theory (Rogers, 2003), that explains the innovation adoption process among the members of a population or within a specific social context (Mkhomazi & Iyamu, 2013, p. 531), the Technology Acceptance Model (TAM) (Davis, 1986; Davis, 1989), the Theory of Reasoned Action (TRA) (Fishbein and Ajzen, 1975) and the Theory of Planned Behaviour (TPB) (Ajzen, 1991) (Hameed et al., 2012, pp. 362-363). The Unified Theory of Acceptance and Use of Tech­nology (UTAUT) (Venkatesh et al., 2003) and its extension UTAUT2 (Venkatesh et al., 2012) were built on several of the previously named theories (Venkatesh et al., 2003, p. 425) and are additionally considered within this thesis. Different to the DOI theory, TAM and the UTAUT, the UTAUT2 was specifically developed to measure factors that influ­ence an individual’s intention to use technology in a consumer context (Venkatesh et al., 2012, p. 157).

1.3.2 Research method

To identify factors that might have a significant effect on the intention to use smart speak­ers, a literature review of adoption theories and adoption studies is conducted. Since smart speakers are an IT innovation, factors that have been found relevant for the ac­ceptance of IT innovations in general and for the acceptance of VAs or smart speakers in particular in previous studies are considered. Based on this literature review, a set of factors is defined, along with hypotheses about how their perception affects the intention to use smart speakers. Since this thesis adopts a deductive research approach (Bryman & Bell, 2011, pp. 11-13), a quantitative research strategy by means of a survey research with a standardised questionnaire as the data collection tool is considered appropriate (Bryman & Bell, 2011, p. 27) and is thus applied. To gather primary data, the link to the online questionnaire was sent to a sample of the target population, defined as all adults living in Germany who do not currently own or use a smart speaker and are active inter­net users. The survey was conducted between 27th May to 16th June 2021. As sampling method, a mixture of snowball and convenience sampling was applied. The final sample consisted of 233 respondents. The results were analysed via a Multiple Linear Regres­sion Analysis (MLRA) using the software tool SPSS.

1.4 Content overview

This thesis is structured as follows. The second chapter, literature review, is divided into five sub-chapters. In sub-chapter one the subject of this study, smart speakers are intro­duced. In sub-chapter two, relevant terms from the adoption and acceptance literature are defined. In sub-chapter three, commonly applied adoption theories and models are presented. In sub-chapter four, the findings of previous IT adoption studies in general and VA/smart speaker adoption studies in particular are analysed. Based on the findings of the previous sub-chapters, this thesis’s research model and hypotheses about the effect of its factors on the intention to use smart speakers are derived in the fifth sub­chapter.

The third chapter, empirical study, is structured as follows. In the first sub-chapter, the study’s objectives are summarised. In the second sub-chapter the applied research method and research design are presented and discussed. In the third sub-chapter the insights of the conducted pre-test are presented followed by a description of the applied data analysis method in sub-chapter four.

The fourth chapter, empirical findings, consists of the main sub-chapter in which the find­ings of the study are presented, followed by a brief sub-chapter that evaluates whether the hypotheses within this thesis were confirmed through the study or not and a short sub-chapter that highlights specific limitations of the empirical study.

The fifth chapter, discussion, is divided into five sub-chapters. First, the findings of the empirical study are briefly summarised. Second, the findings are interpreted and set within a context of previous adoption studies and theoretical frameworks. Third, theoret­ical and practical implications are derived from the thesis’s findings. Fourth, the study’s limitations are critically reflected. The chapter ends with recommendations for future re­search. The thesis ends with a conclusion about the thesis’s findings in chapter six.

2 Literature review

The following chapter provides a literature review on voice-enabled technologies, rele­vant terminology, innovation adoption theories and the empirical findings of previous IT and smart speaker adoption studies. Based on the findings of the literature review, a research model is derived to analyse factors affecting the intention to use smart speakers in Germany.

2.1 Voice operated technology

To identify relevant factors affecting the intention to use smart speakers, it is critical to first briefly understand this IT innovation. This sub-chapter therefore introduces VAs and smart speakers which are based on this software agent. Since this thesis analyses fac­tors that affect the intention to use smart speakers among current non-adopters in Ger­many, the current adoption state is also briefly presented.

2.1.1 Voice assistants

In the literature on VAs, some related terms referring to this software agent are partly used synonymously. However, as some terms differ in meaning, this sub-chapter first discusses the differences in terminology followed by an explanation of VAs and their functions. Understanding synonyms and differences in terminology is essential for defin­ing key terms for the literature review on previous VA/smart speaker adoption studies in sub-chapter 2.4.2. A basic understanding of smart speakers and VAs is furthermore cru­cial for identifying potential factors affecting their adoption.

An Intelligent Personal Assistant (IPA) also referred to as “digital personal assistant” (Lopatovska, 2019, p. 72) or “smart personal assistant” (Balci, 2019, p. 23), is an appli­cation that evaluates user inputs such as voice, images, text or contextual information and provides an appropriate response, e.g. in the form of an answer or the execution of a task (Canbek & Mutlu, 2016, as cited by Lopatovska, 2019, p. 72; Hauswald et al., 2015, p. 223). While Serenko and Deltor (2004, p. 366) argue that the word intelligent refers to the application’s capability to learn from its experiences, be reactive, work au­tonomously and communicate with people or other devices, Knote et al. (2018, pp. 1086­1089) summarise context-awareness, self-evolution, multimodality, anthropomorphism and platform integration as main functional principles of IPAs. While some researchers argue that all IPAs can interpret spoken language (e.g., Canbek & Mutlu, 2016, p. 595; Han & Yang, 2018, p. 620), others consider this as only a common but not necessary function of an IPA (e.g., Drewer et al. 2017, p. 8; Li et al., 2009, p. 125). Although the terms IPA and VA are often used synonymously (e.g., Drewer et al. 2017, p. 8), in this thesis the term VA based on Hoy (2018, p. 81) refers to an IPA that is capable of inter­preting natural spoken language and possibly also other user input and to respond via artificial voice and other means. Other commonly applied terms that are used almost interchangeably for VAs in the current literature are “voice-activated personal assistant” (Coskun-Setirek & Mardikyan, 2017, p. 1), “voice-activated digital assistants” (Koon et al., 2020, p. 16), “voice-controlled agents”, “conversational agents” (Lopatovska, 2019, p. 72), “virtual voice-assistants” (Burbach et al., 2019, pp. 101-102) or a variation of these and similar terms.

Cloud-based VAs like Siri (Apple) or Google Assistant (Google) are widely spread on mobile devices. Through technological advancement, the capabilities of VAs have in­creased over recent years. In current VAs, various technologies are integrated, e.g. from the fields of Artificial Intelligence (AI), cloud computing and the Internet of Things (IoT) (Balci, 2019, p. 24; Yang & Lee, 2019, p. 67). One of the most crucial technologies is Natural Language Processing (NLP), a branch of AI that enables the software agent to automatically process natural spoken language, encode it into structured information and respond to it (Canbek & Mutlu, 2016, p. 594; ITWissen.info, 2019). The VA runs in the providers’ data centres. An internet connection is needed for a device to access the VA on the cloud server (Hörner, 2019, pp. 10-11).

Although each VA has its specific features, Jiang et al. (2015, p. 507) summarised the main functionalities of VAs like Alexa or Cortana as dialogue, web search and chat func­tions. The dialogue function allows users to interact with the device via voice to carry out specific tasks or access certain features. The web search function enables users to search information online via voice. The chat function enables users to interact with the VA and receive pre-defined responses. Among others, VAs are capable of sending text messages, making phone calls, setting reminders, telling jokes, playing music, or con­trolling other smart home devices like lighting systems or thermostats (Hoy, 2018, p. 83). Some VAs can be individually updated with additional functions, so-called voice apps (Hoy, 2018, p. 83). Voice apps for Amazon’s Alexa are called skills. In 2019 already over 100,000 skills were available for Alexa empowered smart speakers (Kinsella, 2019b, para. 1) which include among others Domino’s a skill that allows to order pizza (Amazon, n.d.a), Spotify a skill that allows to access one’s Spotify account and play music (Ama­zon, n.d.b), or PayPal a skill that allows to make transactions via voice (Amazon, n.d.c).

2.1.2 Smart speakers

As already stated by Koo et al. (2017, p. 3), there is no unique definition for smart speak­ers, which are also often referred to as voice-controlled speakers, artificial intelligent speakers or digital assistant speakers. However, according to Kowalczuk (2018, p. 418) and Hörner (2019, pp. 10-13) a smart speaker can be considered as a stand-alone de­vice with integrated microphones and speakers as main hardware components that can be connected via interfaces to a VA. Devices with an additional touch screen as main components are called smart displays but are not the subject of this thesis.

To use a smart speaker, an internet connection is required to connect with the VA on the server provider’s cloud (Hörner, 2019, p. 10). The smart speaker is in constant listen mode and the VA is most commonly activated via a trigger word like Alexa (Amazon) or ok Google (Google) (Lopatovska, 2019, p. 72). Among the most sold smart speakers globally at the end of 2020 were those from Amazon, Google, Baidu, Alibaba, Apple and Xiaomi (Statista, 2021b). Smart speakers vary, for example, in supported VAs, their size, design, sound quality, price and whether or not they provide visual feedback in the form of lights.

2.1.3 The German smart speaker market

Only one study was found that provides relatively current insights into the German smart speaker market. According to Beyto Ltd. (2020, p. 16) 24 percent of the adult German population owned at least one smart speaker in March 2020. Kinsella (2019a) provides further insights into the development of smart speaker household ownership in Germany between 2016 and 2019. Building on these findings, the adopter category of Germans who are currently initially adopting smart speakers based on their innovativeness, which can be understood as the extent to which individuals are “relatively earlier in adopting new ideas than other members of a social system” (Rogers, 2003, p. 280) can be de­fined. The people who are currently adopting smart speakers can be assigned to the adopter category Early Majority on Rogers’ (2003, p. 281) adoption curve. Figure 1 dis­plays the development of smart speaker ownership in Germany among the different adopter categories over time based on Hobson (2016, para. 1), the findings from Beyto Ltd. (2020, p. 16) and Kinsella (2019a). It should be mentioned, however, that there might be a difference between household and individual smart speaker ownership and that the shape of the adoption curve in Figure 1 is only a simplification and might look different in reality. Furthermore, ownership does not necessarily imply adoption.

Figure Smart Speaker Adoption Development in Germany

Abbildung in dieser Leseprobe nicht enthalten

Note. Smart speaker adoption development in Germany and classification of the adopter categories based on their innovativeness. Adapted from Diffusion of Innovations (5th ed., p. 281), by E. M. Rogers, 2003, Free Press. Copyright 2003 by E. M. Rogers.

2.2 Definition of terms

To determine if the theories and models introduced in this paper are appropriate for de­riving relevant factors for the intention to use smart speakers, it is first necessary to clarify what an information technology or innovation is and whether smart speakers and VAs meet the required criteria. To avoid unclarity, it is furthermore important to distinguish between the terms adoption and user acceptance. This sub-chapter therefore aims to define relevant terms applied within this thesis.

2.2.1 Information technology

IT is “the study, design, development, implementation, support or management of com­puter-based information systems, particularly software applications and computer hard­ware” (Information Technology Association of America, 1997, p. 9). IT is concerned with the application of computer hardware and computer software to process, store, secure, transfer and retrieve information (Rouholamini & Venkatesh, 2011, p. 234). VAs are soft­ware agents dedicated to a computer device like smart speakers to process natural lan­guage into data. A VA can therefore be considered as an own or a combination of differ­ent information technologies.

2.2.2 Innovation

The Oslo Manuel (OECD/Eurostat, 2018, p. 20) considers critical aspects of innovations, their novelty, usefulness and implementation, e.g. applied within the firm or put on the market. According to OECD/Eurostat (2018) an innovation is “a new or improved product or process (or combination thereof) that differs significantly from the unit’s previous prod­ucts or processes and that has been made available to potential users (product) or brought into use by the unit (process)” (p. 32). This definition includes the critical aspect of Schumpeter (1939) who argues that an innovation in the field of business includes “any doing things differently in the realm of economic life” (p. 84) and Rogers (2003, p. 12) who considers ideas, practices or objects being perceived as new as innovations. Since VAs and smart speaker fulfil the criteria of OECD/ Eurostat (2018, p. 20), they can be considered as innovations in the field of IT. Thus, the models presented in sub-chap­ter 2.3 are suitable for deriving relevant factors affecting their usage intention.

2.2.3 Adoption

The innovation adoption process can be considered as the steps an individual or com­pany passes from first gaining knowledge about an innovation until fully making use of it (Renaud & van Bijion, 2008, pp. 210-211). The process consists of three stages namely initiation, adoption decision and implementation (Hameed et al., 2012, p. 361). The final adoption can be understood as “a decision to make full use of an innovation as the best course of action available” (Rogers, 2003, p. 177), manifested by the adopter’s actual use of the innovation or technology (Fazel, 2014, p. 82).

2.2.4 User acceptance factors

According to Dethloff (2004, p. 18, as cited by Fazel, 2014, p. 82) the term acceptance can be understood as the positive active acceptance or approval of an idea or product and not just its passive tolerance. User acceptance can therefore be considered as the formation of a positive attitude towards a certain idea or product that affects the usage intention (Fazel, 2014, p. 82). When using the term acceptance in the adoption literature many studies refer to TAM 1-3 (Davis, 1989; Venkatesh & Davis, 2000; Venkatesh & Bala, 2008) or UTAUT 1-2 (Venkatesh et al., 2003; Venkatesh et al., 2012). Since the dependent variable attitude towards use was removed within TAM 2-3 and UTAUT 1-2, the term acceptance is often synonymously applied to not only describing the positive attitude towards a technology but also the positive intention towards using the technol­ogy. To name just a few, Venkatesh & Bala (2008, p. 280) or Kowalczuk (2018, p. 422) refer to acceptance in their study title without including the dependent variable attitude towards use in their research model.

Summing up, different to the concept of adoption, the acceptance of an innovation/tech- nology can be considered as the formation of a positive attitude towards the innovation or the intention to use it. Acceptance can therefore be considered as a crucial phase within the process of innovation adoption, as adoption can only occur if potential users form a positive attitude towards (the usage of) an innovation and accept it (Fazel, 2014, pp. 82-83). Within this thesis, UAFs can therefore be considered as all factors that di­rectly or indirectly affect either the attitude or the usage intention towards an innovation.

2.3 Adoption theories

There is no unified theory of innovation adoption, nor is it likely that one will develop (Fichman, 2001, p. 3). Instead, multiple theories and theoretical frameworks describe different factors that often have an impact on the adoption of innovations or specifically the intention to use IT (Hameed et al., 2012, p. 362). Due to the complexness of behav­ioural science, no single theory or model is able to cover all factors that affect the adop­tion process. Therefore, models and theories differ in the influencing factors considered and have their specific advantages and limitations (Momani et al., 2017, p. 8).

Within this thesis, UAFs affecting the intention to use smart speakers in Germany are not taken from one single theory, but are obtained through a literature review of different innovation adoption and technology acceptance theories and VA/smart speaker adoption studies conducted in different countries. In the following, three commonly ap­plied innovation adoption and technology acceptance theories and theoretical frame­works are presented. These are the DOI theory, the TAM and the UTAUT2 (Hameed et al., 2012, p. 362). The definition of the factors examined in each model are relevant to understand the results of previous VA/smart speaker adoption studies presented in sub­chapter 2.4.

2.3.1 Diffusion of Innovation Theory

The DOI theory developed by E. M. Rogers in 1962 belongs to the oldest theories ex­plaining the diffusion process of innovations in general (Tornatzky & Klein, 1982, as cited by Momani et al., 2017, p. 6). Within this theory, Rogers (2003, pp. 265-266) identified in total five perceived characteristics of an innovation that impact its adoption rate. The factors are presented in this sub-chapter, followed by an assessment of the strength and limitations of the DOI theory.

Perceived characteristics of an innovation. Within this theory, Rogers (2003, p. 170) developed a model consisting of five mental stages, the so-called Innovation-Decision Process, that a decision making unit like an individual sequentially passes through before adopting an innovation. At first, the individual gains knowledge about the innovation, upon which an attitude towards the innovation is formed. Based on this attitude, the in­dividual decides to reject or adopt the innovation. In case of a decision to adopt the innovation, this stage is followed by the implementation of the innovation and finally, the decision to continue its usage or not (Rogers, 2003, p. 170). In the persuasion stage, the individual forms either a positive or negative attitude towards the innovation based on the perception of five innovation characteristics relative advantage, compatibility, com­plexity, trialability and observability (Rogers, 2003, pp. 170-177). Relative advantage is “the degree to which an innovation is perceived as being better than the idea it super­sedes” (Rogers, 2003, p. 15). Compatibility is defined as “the degree to which an inno­vation is perceived as being consistent with the existing values, past experiences, and needs of potential adopters” (Rogers, 2003, pp. 15-16). Complexity refers to “the degree to which an innovation is perceived as difficult to understand and use” (Rogers, 2003, p. 16). The construct trialability is defined as “the degree to which an innovation may be experienced with on a limited basis” (Rogers, 2003, p. 16). Observability, the last con­struct, refers to “the degree to which the results of an innovation are visible to other” (Rogers, 2003, p. 16). The higher the characteristics relative advantage, observability, compatibility and trialability, and the lower the complexity of a particular innovation are perceived, the more likely it is that this innovation will be adopted (Rogers, 2003, pp. 15­16). The characteristics were identified through analysing thousands of innovation studies (Rogers, 1983, as cited by Moore & Benbasat, 1991, p. 193). It should be noted, however, that according to the DOI theory other factors besides the perceived charac­teristics affect an innovation’s adoption rate such as communication channels or the structure of a social system (Rogers, 2003, p. 222), which will however, due to the limited scope of this thesis, not be further discussed.

Critical review of the DOI theory. The strength of the DOI theory lies, among others, in its ability to analyse the adoption of any innovation type (Momani et al., 2017, p. 9) among individuals or other decision-making units like companies (Rogers, 2003, p. 170). However, the former point is also criticised as the theory does not consider the specific perceived characteristics that more precisely explain the adoption of specific types of innovations, e.g. IT innovations (Momani et al., 2017, pp. 9-10). Tornatzky and Klein (1982, p. 34) furthermore criticised that the perceived characteristic relative advantage is defined too broadly and is therefore of little use for understanding its effect on the adoption decision. They have also identified several innovation characteristics, not ex­plicitly covered by the DOI theory but considered by other researchers to affect the inno­vation adoption. The additional characteristics include e.g. communicability, social ap­proval and costs (Tornatzky & Klein, 1982, pp. 33-38). Moore and Benbasat (1991, pp. 192-194) furthermore refined and supplemented Rogers’s original characteristics to more precisely measure the factors affecting the intention of individuals in organisations to use IT innovations. Based on Rogers (2003) findings, they extended the original five perceived characteristics with two additional constructs (voluntariness and image) (Moore and Benbasat, 1991, p. 195), and subdivided the observability characteristics into the constructs result demonstrability and visibility (Moore and Benbasat, 1991, pp. 203-204). They further developed a set of items to measure the constructs (Moore and Benbasat, 1991, pp. 215-216). Moore and Benbasat (1991, p. 196) also slightly rede­fined the meaning of perceived characteristics and labelled them Perceived Character­istics of Innovating (PCI). In their definition, the perceived characteristics do not influence the innovation perception itself, as in Roger’s (2003, p. 174) model, but actually the per­ception of using the innovation. A simplified visualisation of Moore’s and Benbasat’s (1991) PCI is displayed in Figure 2.

Figure Perceived Characteristics of Innovating

Abbildung in dieser Leseprobe nicht enthalten

- Relative advantage: “The degree to which using the innovation is perceived as being better than using its precursor'’(Moore & Benbasat, 1991, p. 196)
- Compatibility: “The degree to which an innovation is perceived as being consistent with the existing values and past experiences of potential adopters” (Moore & Benbasat, 1991, p. 199)
- Ease of use (Complexity): “The degree to which an innovation is perceived as being difficult to use" (Moore & Benbasat, 1991, p. 215; San Martin & Herrero, 2012, p. 343)
- Visibility: “The degree to which one can see others using the system in the organization” (San Martin & Herrero, 2012, p. 343)
- Result demonstrability: “The tangibility of the results of using the innovation, including their observability and communicability" (Moore & Benbasat, 1991, p. 203)
- Image: “The degree to which the use of an innovation is perceived to enhance one’s image or status in one’s social system" (Moore & Benbasat, 2001, p. 195)
- Voluntariness of use: “The degree to which use of the innovation is perceived as being voluntary or free of will” (Moore & Benbasat, 2001, p. 195)
- Trialability: “The degree to which an innovation may be experimented with before adoption” (Moore & Benbasat, 1991, p. 195)

Note. Extended and refined perceived characteristics of (using) IT innovations that more precisely explain the intention to use IT innovations in organisations. Own illustration.

2.3.2 Technology Acceptance Model

Models of technology acceptance in general aim to explain human acceptance factors that affect the intention towards using a certain technology. In the literature, there are three main models that describe factors influencing the acceptance and subsequently usage of technologies or the execution of a certain behaviour (Fazel, 2014, p. 103). These are the TRA (Ajzen & Fishbein, 1980; Fishbein & Ajzen, 1975), the TPB (Ajzen, 1991), and the TAM (Hameed et al., 2012, p. 362), whereby the TPB and the TAM are both built on the TRA (Fazel, 2014, p. 103). While the TAM specifically aims to explain factors affecting the acceptance of IT, the TRA and TPB can rather be considered as frameworks explaining factors influencing human behaviour in general. In the following, the TAM, which was originally developed to predict the acceptance and subsequently usage of IT by individuals at their workplace, is described in more detail (Davis et al., 1989, p. 985; Venkatesh et al., 2003, p. 428).

User acceptance factors considered in the TAM. The model introduced by Davis (1986) states that two constructs (beliefs), Perceived Ease of Use (PEOU) and Per­ceived Usefulness (PU), have a primary effect on users’ attitude towards IT and their consequently usage (Davis, 1989, p. 333; Davis et al., 1989, p. 985). The TAM of Davis, (1989, pp. 333-334) furthermore states that PEOU and PU are affected by external var­iables and that PU is also positively affected by PEOU as an IT system is considered to be more useful if it is easy to use (Fazel, 2014, pp. 111-112). According to the TAM, PU also directly influences the Behavioural Intention (BI) (Davis, 1989, pp. 333-334). Davis et al. (1989, pp. 989-990) empirically tested the original TAM model in a non-organisa- tional setting and found that PEOU and PU both directly had a positive effect on BI. They therefore adapted the model slightly and removed the dependent variable attitude to­wards use (Davis et al., 1989, pp. 996-997; Venkatesh et al., 2003, p. 428).

Critical review of the TAM. The strength of the TAM lies in its simplicity, which makes it well suited for practical use such, as assessing the acceptance of an IT system (Sha- chak et al., 2015, p. 507). Overall, the TAM has proven its reliability and validity as a model (King & He, 2006, p. 751). Among different settings, PEOU (Keil et al., 1995, as cited by Moon & Kim, 2001, p. 217; Malhorta & Galletta, as cited by Moon & Kim, 2001, p. 217) and especially PU (Keil et al., 1995, as cited by Moon & Kim, 2001, p. 217; King & He, 2006, p. 751; Malhorta & Galletta, as cited by Moon & Kim, 2001, p. 217) have been found to be capable of predicting (IT) acceptance.

Besides its strengths, the model has also been criticised. Some researchers argued that BI did not affect the actual behaviour in several studies. These researchers found low or no significantly positive effect of BI on the actual IT usage (Fazel, 2014, p. 115; Szajna, 1996, pp. 89-90). However, in a meta-analysis of 60 studies that applied the TAM, Turner et al. (2010, pp. 468-469) identified that the construct BI overall, can be considered a valid indicator of actual system usage. Over-simplicity is another criticism of the TAM as the model considers only a small number of variables (Shachak et al., 2019, p. 1). Due to its simplicity, some researchers argued that the TAM might not be precise enough to explain the acceptance of certain technologies or the acceptance within certain contexts (Moon & Kim, 2001, p. 217). Furthermore, researchers found that the TAM hardly ex­plains above 40 percent of variance in people’s self-reported system use (Legris et al., 2003, p. 202).

Therefore, Davis et al.’s (1989) model was extended to include additional external vari­ables that affect the BI to use IT in different settings, e.g. TAM2 (Venkatesh & Davis, 2000) and TAM3 (Venkatesh & Bala, 2008), that additionally consider external variables affecting PEOU, PU and BI, or UTAUT (Venkatesh et al., 2003) and UTAUT2 (Venkatesh et al., 2012), that additionally consider external variables directly affecting BI to use a system. Within all the models the dependent variable attitude towards use of the original TAM model (Davis et al., 1989, p. 985) was excluded. While TAM 1-3 and UTAUT were developed to determine factors that affect the acceptance of IT by individuals within an organisation, the UTAUT2 was mainly developed to study the acceptance of IT in a con­sumer context (Venkatesh et al., 2012, p. 157). As this thesis aims to identify factors that affect the intention to use smart speakers, which are consumer products, the UTAUT2 model is briefly presented in the following sub-chapter.

2.3.3 Unified Theory of Acceptance and Use of Technology 2

The UTAUT model, which has been proven capable of explaining about 56 percent of variance in BI and 40 percent of variance in the actual use of IT (Venkatesh et al., 2012, p. 157), was the basis for developing the UTAUT2 model, which aims to explain factors for the acceptance and usage of IT among end consumers (Venkatesh et al., 2012, p. 158). Based on the original UTAUT (Venkatesh et al., 2012, p. 158), UTAUT2 has inte­grated the factors of eight former technology/innovation acceptance theory models, in­cluding the DOI-theory, TAM, TPB and TRA (Venkatesh et al., 2003, p. 425).

User acceptance factors considered in the UTAUT2. Based on the findings of prior studies that applied the original UTAUT model, Venkatesh et al. (2012, pp. 158-166) formulated five hypotheses about the effects of certain variables on consumer’s BI to­wards using IT and empirically tested them among users of mobile internet technology. Besides the constructs PEOU and PU of the TAM model, reformulated to performance expectancy and effort expectancy, the UTAUT2 includes five additional factors (social influence, facilitating conditions, hedonic motivation, price value and habit) that affect the continuous usage intention, which again directly affects the actual usage behaviour of IT by individuals. Furthermore, different to the TAM, three moderating variables age, gen­der and experience, have been added that partly have a moderating impact on certain effects within the model (Venkatesh et al., 2012, pp. 169-170).

Critical Review of the UTAUT2. The strength of the UTAUT2 lies in its capability to explain about 70 percent of the variance in BI and 50 percent of the variance in actual technology usage (Venkatesh et al., 2012, p. 157). A further advantage is that it is spe­cifically designed to analyse the acceptance and usage of IT within a consumer context. Another strength is the comprehensive factors considered in the model (Tamilmani et al., 2017, p. 46). However, many studies applying the UTAUT2 included additional vari­ables from other theories in their research model (Tamilmani et al., 2020, p. 4). The UTAUT2 may therefore not be applicable as a stand-alone model without additional context-specific factors. Some studies also excluded the moderating variables in their analysis due to increasing complexity (Tamilmani et al., 2017, p. 46).

2.4 Identification of acceptance and usage relevant factors

“Building your research on and relating it to existing knowledge is the building block of all academic research” (Snyder, 2019, p. 333). Instead of conducting a separate netnography or applying other qualitative research methods like semi-structured or focus group interviews, potential user acceptance factors for smart speakers included in this thesis’s research model were identified through a literature review. Therefore, the factors hypothesised to affect smart speaker acceptance in Germany are based on the consid­eration of general adoption and user acceptance factors defined in the previous chapter and on a literature review of factors considered in previous user acceptance or adoption studies. Within the next two sub-chapters, factors that have been considered within dif­ferent IT innovation acceptance studies in general are reviewed, followed by a review of factors considered in previous VA/smart speaker acceptance studies in particular.

2.4.1 Factors considered in IT innovation adoption studies

As already stated in sub-chapter 2.2, smart speakers can be considered as an innovation in the field of IT. Thus, the factors that were relevant for IT innovation acceptance in previous studies were analysed. Instead of reviewing individual studies, studies that have already conducted a meta-analysis on relevant IT innovation acceptance factors were sought. Although a considerable literature review was conducted, only one meta­analysis study that analysed relevant factors for the acceptance of IT innovations by individuals across multiple studies was identified. This study of Hameed et al. (2012) analysed the effect of different factors, gained through a literature review of 151 adoption studies, on IT innovation adoption by organisations and their usage through individuals within an organisational setting (p. 362). Most of the studies analysed by Hameed et al. (2012, pp. 362-367) were based on theoretical frameworks, commonly applied in the innovation adoption or technology acceptance literature, including the theoretical frame­works from the DOI theory, TAM and TRA already presented in sub-chapter 2.3, and some additional frameworks also capable of explaining relevant adoption factors.

The author is aware that factors affecting the usage intention of IT innovations in com­panies considered by Hameed et al. (2012) such as enterprise instant messaging or electronic supply chain management systems (p. 382) may differ from the factors affect­ing the usage intention of IT innovations within a consumer context. However, the find­ings were considered as a good reference to identify potential factors that may also be relevant for the acceptance of smart speakers. Table 1 presents factors analysed in previous studies about IT innovation acceptance by individuals and the relative fre­quency the analysed factors were considered as significant for IT acceptance (Hameed et al., 2012, pp. 373-374).

Table List of Factors Considered in the Individual Level IT Adoption Literature

Abbildung in dieser Leseprobe nicht enthalten

Note. User acceptance factors considered in previous IT adoption studies and the rela- tion how often they have been considered as significant drivers for acceptance and ac­tual usage of IT innovations by employees. Adapted from “A conceptual model for the process of IT innovation adoption in organizations”, by M. A. Hameed, S. Counsell & S. Swift, 2012, Journal of Engineering and Technology Management, 29 (3), p. 373. https://doi.org/10.1016Zj.jengtecman.2012.03.007. Copyright 2012 by M. A. Hameed, S. Counsell & S. Swift.

As shown in Table 1, the most often analysed UAFs in the individual adoption literature were PU and PEOU, which also had a significant effect on the acceptance of IT in most studies. Perceived behaviour control, which is in its meaning quite similar to the PEOU concept (Ajzen, 1991, pp. 183-184), self-efficacy, which refers to one’s perceived capa­bilities of performing the behaviour in question (Bandura, 1982, p. 122), user experience, facilitating conditions, compatibility and perceived playfulness were also identified as sig­nificant acceptance factors in several studies. To sum up, among different IT acceptance studies within an organisational setting, PU, PEOU, compatibility, perceived playfulness, user experience, self-efficacy, perceived behavioural control and facilitating conditions were frequently identified to be significant UAFs.

2.4.2 Factors considered in VA/smart speaker adoption studies

To identify factors likely to influence the intention to use smart speakers in Germany, an additional literature review on previous VA/smart speaker adoption studies conducted in different countries was performed.

This approach was chosen because it allows the consideration of the results of several previous VA/smart speaker adoption studies, which have themselves largely derived their research factors from various other theories, e.g. the parasocial relationship theory (Han & Yang, 2017, pp. 620-621) or the theory of network externalities (Park et al., 2018, p. 2120), literature reviews of other technology acceptance models like the value-based adoption model (Pal, Arpnikanondt, Funilkul, & Chutimaskul, 2020, pp. 10853-10855), or qualitative research methods such as netnography (Kowalczuk, 2018, p. 420), focus group interviews (Kessler & Martin, 2017, pp. 26-32) or semi-structured interviews (Koon et al., 2020, p. 17). Therefore, the findings on potential adoption factors from several methodologically diverse studies could be considered for the creation of a research model analysing the intention to use smart speakers among current non-adopters in Ger­many.

Identification of relevant VA/smart speaker adoption studies. As a first step, criteria for the literature review were set. Selection criteria for studies were: (a) the subject of the study analysed factors influencing the adoption, acceptance or (continuous) usage of VAs/smart speakers, (b) among the studies analysing VAs, only studies were considered that analysed VAs accessible via smart speakers like Siri or Alexa and, (c) the studies analysed the acceptance or use of smart speakers by individuals instead of organisa­tions.

Peer-reviewed articles were preferred. However, as the literature review only concerns identifying user acceptance factors that have been considered in previous studies, sev­eral conference papers and master thesis results were also reviewed. Still, the assess­ment of the findings within peer-reviewed articles were considered most reliable.

Relevant studies were identified via the internet. In particular, relevant studies were searched for on Google Scholar, ScienceDirect, Springer Link, Emerald Insight, and in­dividual potentially relevant journals, e.g. in the field of IT or innovation, which received a relatively high ranking according to the Association of Business Schools (2015). Key­words like smart speaker or different terms for VA, defined in Chapter 2, e.g. IPA, voice- activated personal assistant, voice-controlled agent in combination with keywords relat­ing to adoption, acceptance or actual usage, e.g. adoption, acceptance factors, usage, etc. were applied to identify potential articles. The research took place between February and March 2021. The studies of Pitardi and Marriott (2021) and Cha et al. (2021) were identified after the literature review and additionally added. It should be noted, however, that there may be further studies about VA/smart speaker adoption, which have not been identified through this literature review, which does not claim to be exhaustive but rather aims to provide an overview of factors that influenced the intention to use smart speakers in previous studies.

In total 22 studies have been identified, fourteen journal publications (Cha et al., 2021; Coskun-Setirek & Mardikyan, 2017; Han & Yang, 2018; Koon et al., 2020; Kowalczuk, 2018; Lau et al., 2018; Ling et al., 2021; McLean & Osei-Frimpong, 2019; Moussawi et al., 2020; Pal, Arpnikanondt, Funilkul & Chutimaskul, 2020; Pal, Arpnikanondt, Funilkul & Razzaque, 2020; Park et al., 2018; Pitardi & Marriott, 2021; Yang & Lee, 2019), three conference papers (Burbach et al., 2019; Liao et al., 2019; Nasirian et al., 2017) and five master theses (Arifin, 2020; Chu, 2019; Hoffmann & Thuesen, 2018; Kääriä, 2017; Kess­ler & Martin, 2018).

Identification of previous considered VA/smart speaker adoption factors. The quantitative and qualitative studies analysed different UAFs for VAs/smart speakers and were only partly based on common technology acceptance models like those described in sub-chapter 2.3. The factors considered in previous adoption studies, irrespective of whether a direct or indirect effect on the intention to use VAs/smart speakers was actu­ally found or only hypothesised, were clustered into categories of fairly similar factors. Table 2 presents the result of the literature review. It aimed to include as many individual factors as possible in the table, but since some factors could not be assigned to a suitable cluster and seemed irrelevant for this thesis, they were excluded from the table.

[...]

• Université: Management Center Innsbruck (Management Center Innsbruck)
• Note: 1.00
• Auteur: Jakob Summer (Auteur)
• Année de publication: 2021
• Pages: 104
• N° de catalogue: V1339687
• ISBN (PDF): 9783346851413
• ISBN (Livre): 9783346851420
• Langue: anglais
• mots-clé: Smart Speaker, Voice Assistent, Alexa, Siri, User Acceptance, Adoption, Innovation