Business models for the Internet of Things : can the laws of information build the value proposition

(1)

Business Models for the Internet of

Things: Can the Laws of Information

Build the Value Proposition?

Msc. Business Administration Track Entrepreneurship & Innovation Master Thesis – Final version Dhr. Dr. G.T. Vinig

Marjolein V. van Hage 10171649 18-06-2017 Amsterdam

(2)

Statement of Originality

This document is written by Student Marjolein Valerie van Hage 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.”

M.V. van Hage 10171649

(3)

List of Tables

Table 1. Laws of Information and the Value of Information attribute………..21

Table 2. Survey sample statistics involvement in IoT………...26

Table 3. Survey sample statistics: the use of IoT products………26

Table 4. Survey sample statistics: role in IoT. ………..26

Table 5. Interview participants. ……….28

Table 6. Support for value of information statement……….30

Table 7. Cronbach’s Alpha………30

Table 8. Cronbach’s Alpha. ………..31

Table 9. Mean and Standard deviation of value of information statements………..31

Table 10. Mean and Standard Deviation of economic information attributes..……….32

Table 11. Mean and Standard Deviation for specifics for revenue recognition………33

Table 12. Example quotes related to the value proposition………...35

List of Figures

Figure 1. Business model analysis……….11

Figure 2. Business Model Canvas………..12

Figure 3. Information flows in IoT………15

(5)

Abstract

The Internet of Things (IoT) has enhanced the availability of real-time information as connected sensors provide continuous streams of data. Current trends reveal the possibilities of IoT, as various sets of sensors make it possible to access information at any moment thus improving visibility and expediting decision-making. The number of IoT applications that are entering the market is growing and the interest in the economic value of the information generated by IoT is increasing. The economic approach to the value of information may have greater impact on the success of IoT applications than the technology itself. This study uses an exploratory research approach and, through the triangulation of two methods, it focusses on the understanding of the value proposition of business models for the IoT. It has adapted the laws of information (Moody & Walsh, 2002; Bucherer & Uckelmann, 2011), aligned them with the economic attributes of information (Glazer, 1993; Bisdikian et al., 2013), and extended this to validate the economic value creation through the value proposition of IoT business models. It used this framework to conduct first, a survey among providers of IoT applications and its users and then, eight interviews with founders of IoT companies. The findings show that the laws of information are essential for the value proposition of IoT business models and that the value proposition is to provide users with insights that improve decision-making and offers a solution to a problem. Furthermore, the research reveals that the economic attributes accuracy and relevance, both representing a law of information, are critical for economic value creation. These critical attributes express the quality of the information and determines the probability that the information yields the expected outcome.

Keywords: Internet of Things, Business Models, Value Proposition, Laws of Information, Economic Value of Information.

(6)

1. Introduction

The information theory developed by Claude Shannon, called ‘The Mathematical Theory of Communication’ (1948), is deeply embedded in what we know as the Internet of Things (IoT). The IoT refers to physical objects that are connected via the internet which enables these objects to collect and sent data. The IoT has evolved due to the remarkable development in internet technology and computing. Current trends reveal the possibilities of the IoT, where the use of connected sensors facilitates the collection of enormous volumes of data from various sources. This data can be used to create information for users and the collected information can be sold (Chan, 2015). IoT applications increase the availability of real-time information as connected sensors provide a continuous stream of data and information. Various sets of sensors make it possible to access the information at any moment; this results in increased visibility and expeditious decision-making (Bisdikian, Kaplan & Srivastava, 2013). These technology developments are changing the current economy as organisations increasingly rely on the flow of information rather than on things and money (Azorti et al., 2010).

It is recognized that the long-term success of a company relies on its ability to keep its business model updated, but many companies fail to fully understand business model innovation (Bucherer, Eisert & Gassmann, 2012). One of the reasons that traditional business models are inadequate for IoT applications is because the nature of IoT means that firms in the IoT ecosystem need to collaborate across industries and with competitors (Chan, 2015). This is the opposite of most traditional business models where firms operate with closed ecosystems. As a result, the potential business opportunities for IoT applications have not been leveraged yet (Azorti et al., 2010). Besides, IoT technology is considered a disruptive technology and creating value from disruptive technology is difficult as value remains unidentified until commercialisation is realised through a business model (Chesbrough & Rosenbloom, 2002). Disruptive innovations in technology require innovation in business models and companies need to adjust their products to leverage these innovations. Strategies need to be improved in parallel with the innovative technologies to discover where the opportunities are and realise the economic benefits (Azorti et al., 2010; Sun, Yan, Lu, Bie & Thomas, 2012). There is little assistance for companies from academic literature, as the few studies available on IoT business models focus mainly on the technology platform (Westerlund, Leminen & Rajahonka, 2014). Yet, other studies demonstrate that companies that focus on integrating their information technology successfully with their business model innovations have a competitive advantage over companies that focus on only the technology (Glazer, 1993).

(7)

Innovation in IoT business models is important since it aligns technology developments with the creation of economic value (Bucherer and Uckelmann, 2011). Furthermore, the economic approach to the value of information is a critical issue and may have greater impact on the success of IoT applications than the technology itself (Niyato et al., 2016). In prior literature, information is being progressively recognised as the key economic resource a firm can have (Azorti et al., 2010; Bisdikian et al., 2013). The IoT increases the potential for new value propositions as it is now possible to monitor actual usage of services and all information becomes observable. Besides, IoT applications reveal that the value lies in providing real-time information (Glazer, 1993). Many scholars consider the value of information as a core element of the value proposition and value creation for IoT business models (Moody & Walsh, 2002; Bucherer & Uckelmann, 2011; Niyato et al., 2016).

However, findings of several studies might be aligned about the significance of the value of information for the value proposition of IoT business models, yet, scholars have different opinions about the definition and measurements for the economic value of information. Some scholars argue that the information generated by IoT applications should be treated in the same way as physical resources (Chan, 2015; Niyato et al., 2016). Where other scholars argue that information is not a traditional economic good and thus unqualified to be treated in the same way as physical resources (Glazer, 1993; Bucherer & Uckelmann, 2011). Information has specific economic attributes that are different from those of typical economic goods (Glazer, 1993; Moody & Walsh, 2002; Bucherer & Uckelmann, 2011). It is essential to identify critical elements of the business models for creating value from IoT applications and it is becoming increasingly important to understand how these attributes create value (Ju et al., 2016). Doing so enables companies to extend a greater value proposition to the customers (Ju et al., 2016).

Moody and Walsh (2002) define seven laws of information, derived from the information theory, which are used to determine quantitative measurements for the economic value of information as an asset. These laws are the foundation for understanding the value proposition of business models for the IoT industry (Bucherer & Uckelmann, 2011). This study builds on prior research and uses a developed theory to understand the essential elements for the creating economic value of information. The following question is proposed for this research:

“How to create economic value for IoT applications through business models by using the seven laws of information as design for the value proposition?”

(8)

The laws of information are used as approach to explore business models for IoT. The use of the laws of information for the economic value of information is novel and related literature is scarce. This study aims to gain insights for creating economic value from information gathered by IoT applications. More specific, it aims to use a framework based on the laws of information to provide novel insights to address the value proposition in the business model for IoT applications. It is aimed to achieve results that companies can use to understand appropriate business models for operating in the emerging industry of the IoT.

The following chapter describes and explores existing literature on the IoT, business models, the value of information and the economic attributes. The chapter thereafter illustrates the conceptual model and the next chapter describes the research methods used in this study and the relevant data collected. The next chapter discusses the results of the research. The final chapter contains the conclusions and limitations of this research.

2. Literature Review

The IoT area is one in which technology is continuously improving and developing. The Economist Intelligence Unit (EIU) found in their study that businesses worldwide are using IoT and many of them are transitioning from research and development to market deployment (2013). As there is no single, superior business model in this area, existing literature addresses many relevant economic models to see how value is created by IoT products (Westerlund et al., 2014; Chan, 2015; Dijkman et al, 2015; Ju et al, 2016). The business model innovation for IoT faces many challenges and limitations. The technology is improving rapidly and many technology companies their business models differ substantially from each other (Westerlund et al., 2014). This chapter explains the important concepts and elements and discusses the theory that has been used. It also introduces relevant business models for IoT available in existing literature. The first paragraph explains IoT and its architecture. The second paragraph focusses on the relevance of business models for disruptive technology innovations. It also analyses current business models that existing literature uses to address IoT industry business models. The next paragraph explains information theory and the laws of information. The final paragraph examines the established conceptual framework.

2.1 Internet of Things and Architecture

The IoT is a very broad concept and there are slight differences in how it is defined in existing literature. This study uses the following basic definition: IoT is a concept that describes technology applications that extend the Internet into objects; these objects are all connected

(9)

with the Internet and each other. They can communicate with one another and they generate data (Mattern & Floerkemeier, 2010). The IoT study by Goldman Sachs (2014) states that IoT:

“Connects devices such as everyday consumer objects and industrial equipment onto the network, enabling information gathering and management of these devices via software to increase efficiency, enable new services or achieve other health, safety or environmental benefits” (p. 2).

IoT moves away from the traditional desktop and into the objects we use in our everyday life. The IoT is an emerging field, driven by technology developments, where most new products produce an extraordinary amount of data (Gubbi, Buyya, Marusic & Palaniswami, 2013). Gubbi et al. (2013) explain that “novel fusion algorithms” have been created that are able to understand the data and learn more from it as the number of connected items increase. Furthermore, Ju, Kim & Ahn (2016) state that the data created by IoT can support consumer needs and excellent data analysis will help to improve existing products or create innovative new ones. This data analysis will change the nature of existing products or services as it create new information and lead to additional value. The opportunities for companies are enormous; they can use data analysis to deliver customised and personalised products and services to their customers. Moreover, decision-making becomes faster when companies have a continuous stream of relevant information (Ju et al., 2016).

These opportunities are possible because of the innovative architecture of IoT technology which consist of three layers (Sun et al., 2012). The first layer of IoT is the sensors, which is the hardware component. These sensors are connected to the internet and continuously create enormous amounts of data. Each sensor has an Internet Protocol (IP) address which it uses to communicate with other network nodes and objects connected to the internet (Mattern & Floerkemeier, 2010). Prior to the IoT, it was not possible to connect enormous quantities of different IP addresses to the internet (EIU, 2013). However, The Internet Protocol Version 6 (IPv6), introduced in 2012, enhanced the availability of unique Internet addresses. This enhanced the use of IoT as unlimited number of sensors can be connected to the Internet (EIU, 2013). The market of sensors is growing exponentially as the costs of sensors decreases (EIU, 2013). The second layer of IoT is the network layer where all the data from sensors is received. The sensors share data with other sensor nodes and send data to this central database, which is often a cloud based solution (Sun et al., 2012). After the data is send to the cloud, it needs to be processed. This happens in the application layer, the third, and last, layer of IoT. The application layer is referred to as the intelligence centre of IoT (Sun et al., 2012). It is a centralised system that enables data storage and computing tools for analysis of data. This is necessary because sensors collect data more autonomously than in the past (Ju et al., 2016) and

(10)

this does not provide a competitive advantage to companies unless it is correctly analysed (EIU, 2013; Ju et al., 2016). Companies face challenges in this last layer because they often lack excellent analytical capabilities. Analysing the data and providing information is done by highly skilled data scientists but they are scarce (Ju et al., 2016).

The Economist Intelligence Unit (EIU, 2013) expects an explosion of data generated by IoT because of IoT sensors. Sensors have “transformed daily life in human society, as they generate, process and store the amount of data increasing at exponential rate all over the world” (Hai Jiang et al, 2014, p. 133). Atzori et al. (2010) state that it is unquestionable that the strength of IoT will impact the behaviour and everyday life of potential users. Therefore, economic value creation for users gains increasing interest because that is key for revenue streams (Gubbi et al., 2013). Identifying the value for the end user remains challenging as they are not interested in IoT itself (Wu, Li, Cheng & Lin, 2016). The end users want to know what value the IoT has to them and what benefits they can obtain from it (EIU, 2013). Wu et al. (2016) state that generating the right value is important to yield the satisfaction of the user and improve the use of the application to increase the contribution of additional benefits. The Goldman Sachs study (2014) explains that “the IoT building blocks will come from those that can web-enable devices, provide common platforms on which they can communicate, and develop new applications to capture new users” (p. 1).

This research makes a clear distinction between the concepts of data and information. A sensor transmits data, but the data itself has no meaning for the user. The data is processed into information and this information enhances the knowledge of the user. Information is obtained from sensor networks where the data is properly organised (Bisdikian et al., 2013).

2.2 Business Models

Even though most businesses are aware of the greater potential of IoT, existing literature on technological management shows that firms find it difficult to manage innovations in which they have no experience (Chesbrough, 2010; Chan, 2015). The definition of a business model is “the method of doing business by which a company can sustain itself – that is generate revenue” (Chesbrough, 2010, p. 533). Firms need to find the right architecture of revenue if they want capture value from new technology innovations. They might pursue similar purposes whilst implementing diverse revenue models. As the technology for IoT is developing radically, firms need to adjust their business models and their economic value creation at a rapid pace as well. It is demonstrated that old business models do not apply for IoT because in

(11)

designing these traditional business models the firm is viewed as the central, where for firms in the IoT industry, the nature of the technology means that collaboration is necessary player (Atzori et al., 2014; Chan, 2015).

Nonetheless, it is essential to identify new, appropriate business models because they are the mediator between technology development and economic value creation (Chesbrough & Rosenbloom, 2002). The success of a business model depends on the degree of understanding the dependent factors for the technological potential that achieve economic value for the user, which is illustrated by figure 1 (Bucherer & Uckelmann, 2011). Successful business models for innovations should use the core of innovative technology to bring value to the user (Chesbrough & Rosenbloom, 2002). The end user ascribes the value of the technology to the extent in which the technology offering is a solution for a certain problem (Chesbrough & Rosenbloom, 2002). Furthermore, technology offerings have an appealing value proposition to its user when it directly reduces cost and improves efficiency. Glazer (1993) believes that information technology creates economic value when it improves decision making. The value of technology is not inherent for every user, it has different value attributes for various users (Chesbrough & Rosenbloom, 2002). For IoT, the actual carrier of value is the information (Glazer, 1993) that offers a solution to a certain problem.

Figure 1. Business model analysis (source: Bucherer & Ucklemann, 2011).

Several researchers (Chesbrough, 2010; Bucherer & Ucklemann, 2011; Dijkman et al, 2015) have used the “business model canvas”, introduced by Osterwalder and Pigneur in 2010, to address the IoT industry. This framework identifies four main elements: value proposition, infrastructure, customer, and financials. The business model canvas creates and analyses the business activities of a company and helps to understand its business model (Ju et al., 2016).

(12)

Figure 2. Business Model Canvas (source: Osterwalder & Pigneur, 2010)

The business model canvas is a time-consuming analysis tool to develop a business model that has multiple levels. Using it to design the business model is complex (Sun et al., 2012). Furthermore, it lacks a clear relationship between cause and outcome and does not clearly specify what should be done to achieve success. Also, the business model canvas is rather descriptive. Service-dominant logic is another well-developed framework that has been investigated for its applicability in IoT (Chan, 2015). In this framework, business models have four essential elements: the target customer (who), the value proposition (what), the value chain that delivers the value proposition to the targeted customer (how), and the underlying economic model that captures value (why) (Chan, 2015). Information about customer behaviour is collected in a connected world, and the service-dominant logic is used to construct a business model for IoT (Chan, 2015). In this framework, the value of information is central.

Of the building blocks that these existing models demonstrate, many scholars acknowledge that the value proposition is the most critical element for IoT business models (Bucherer & Ucklemann, 2011; Chan, 2015; Dijkman et al., 2015). This is supported by the 72 IoT professionals involved in the study by Dijkman et al. (2015). The study demonstrates that the value proposition is believed to be most important, which is discovered at a 95% confidence interval with a significance level smaller than 1%. The purpose of any business model is the revenue generated, which results from the value proposition successfully offered to customers (Osterwalder & Pigneur, 2010). Ju et al., (2016) conclude that it is essential to identify critical elements of the value of information for business models that create economic value from IoT applications. Doing so enables companies to extend a greater value proposition to the customers. As mentioned earlier, several scholars believe that information itself is the main component that delivers value for IoT applications and thus the value of infromation (Glazer, 1993; Dijkman et al., 2015).

(13)

2.3 Information Theory

As stated in the introduction, it is widely acknowledged that the information theory developed by Claude Shannon in 1948 has led to the development of the internet. The theory quantifies the limits for compressing, storing, and transmitting data and originates the concepts on which the Internet is based, that is, high-speed communications, data transfer and file compression. Shannon developed mathematics that made it possible to quantify information which enabled for “setting up probability distributions on the basis of partial knowledge, and leads to a type of statistical inference which is called the maximum-entropy estimate” (Jaynes, 1957, p. 620). The focus was on solving a communication problem where a message, that is compressed at one point, can be transmitted and exactly reproduced at another point (Shannon, 1948). In other words, Shannon’s information theory made it possible to transmit, process, and extract messages. The theory is focused on the utilisation of the information sent. Shannon acknowledged that, in communication, messages often have meaning: “that is they refer to or are correlated according to some system with certain physical or conceptual entities” (1948, p. 1). However, he considers these semantic characteristics to be irrelevant when solving the engineering problem of information in communication. Shannon’s theory ignores that a message has meaning and is neutral about the value of the information transmitted. Furthermore, the theory expresses information as means of reducing uncertainty (Lawrence, 1979). Several researchers note that the mathematical definition of information ignores its consideration of content (Glazer, 1993; Lawrence, 1979; Moody & Walsh, 2002). A message can be information but when it holds no value, it has no meaning. The unique structure of information, especially the mathematical structure, causes complications in the attempts that use traditional approaches to value information (Glazer, 1993).

Even though the information theory ignores the meaning of information, it is highly valued in economics. Information is deemed valuable for decision-making because it reduces uncertainty. Therefore, it is critical to understand how economic benefits are extracted from information. In economics, the problem of decision-making under uncertainty has been widely researched. The consequences of such decisions are the result of incomplete knowledge or imperfect information (Lawrence, 1979). In contemporary society, many decisions, both business and personal, are made daily and these decisions are based on the information available to the decision-maker. Managers make decisions based on management information, doctors make decisions based on patient information obtained from tests, and consumers make decisions based on comparing products. Information is widely available to everyone as anyone can easily search the internet for the necessary information (Atzori et al., 2010). As IoT

(14)

provides a continuous stream of real-time information, it is more important than ever to analyse the usability of the transmitted information and identify its value. Lawrence (1979) posits that the value of information depends on the decisions made based on the information.

Thus, it is concluded that the purely mathematical form of the communication theoretic terms is coherent in its measurement but its use is limited because it does not attempt to measure the meaning of the information. The meaning of information is undoubtedly correlated to the measure of the value of the information (Glazer, 1993). The primary concern of economic users of information is the value of the information and not the abstract definition of the theory (Glazer, 1993). Therefore, the value of information is the focus of this research to increase economic benefits.

2.4 Value of Information

Prior literature on business models for the IoT, and technology in general, states that when economic value is created for the user of information, this leads to success (Chesbrough & Rosenbloom, 2002; Dijkman et al., 2012; Chan, 2015; Wu et al., 2016). Niyato et al. (2016) argue that the economic approach to the value of information is a critical issue and may have greater impact on the success of IoT applications than the technology itself. Bucherer and Uckelmann (2011) explain that through IoT, information will become a key determinant of value and thus the value proposition. However, there is no complete satisfactory understanding of what constitutes information and how economic benefits are extracted from it (Glazer, 1993). Some scholars argue that the information generated by IoT should be treated in the same way as physical resources; it should be optimised and its utilisation should be maximised (Chan, 2015; Niyato et al., 2016). In contrast, other scholars argue that information is not a traditional economic good and thus unqualified to be treated in the same way as physical resources (Glazer, 1993; Bucherer & Uckelmann, 2011). In IoT, the physical product is used to provide a continuous stream of information and the willingness to pay for digital services is increasing (Bucherer & Uckelmann, 2011). This is in contrast with the traditional economics, where a price for information did not exist as the cost of information was part of a physical products’ price. The figure below shows the information providers in IoT and the flows of information between them (Bucherer & Uckelmann, 2011). This flow is called “the triangle of information exchange”.

(15)

Figure 3. Information flows in IoT (source: Bucherer & Uckelmann, 2011).

As the figure illustrates, information flows bilaterally between the things, consumers, and businesses. Glazer posits that “Any discussion of an information or knowledge theory of value must begin with an analysis of the peculiar attributes of information as a commodity” (1993, p. 101). He states that information differs from typical economic goods because “it is not easily divisible or appropriable”, “it is not inherently scarce (although often perishable)”, and, as the value often increases with use, “it may not exhibit decreasing returns to use”. Moreover, in contrast to other commodities that are non-renewable, information has a self-regenerative aspect and is non-depletable (Glazer, 1993, p. 101). This means that “the identification of a new piece of knowledge immediately creates both the demand and conditions for production of subsequent pieces” (Glazer, 1993, p. 102). Glazer (1993) also mentions that the unique features of information as a commodity are problematic when attempts are made to assign value to information using traditional approaches. The reason for this is that there is an extreme divergence between the value when information is used and its exchange value, as it is not scarce in the traditional way.

These unique economic elements of information, described by Glazer (1993), are used by Moody and Walsh (2002) to analyse the value of information from a value asset perspective. Information is recognised as a company asset because it has great value when it is treated and analysed properly (Moody & Walsh, 2002). However, because it is difficult to measure its value quantitatively, it is not recognised on the balance sheet of a company. The study by Moody and Walsh (2002) develops seven laws of information for valuing information as a company asset. These laws build on the prior research of Glazer (1993) and are derived using the specifics of the information theory. The study by Moody and Walsh (2002) attempts to build a conceptual model for the valuation of information as they believe this is the main value proposition for business models in the future. Bucherer and Uckelmann (2011) are the first to introduce these laws for business models for IoT and use these to analyse the opportunities in the IoT industry. Their study identifies attributes for the value creation of IoT (Bucherer &

(16)

Uckelmann, 2011). The IoT increases the potential for new value propositions as it is now possible to monitor actual usage of services and all information becomes observable.

The seven laws of information that govern information behaviour as an economic good (Moody & Walsh, 2002; Bucherer & Uckelmann, 2011) are as follows:

Law of information one: Information is (infinitely) shareable and can be shared with others without loss of value.

This may be the most exclusive characteristic of information as an economic good. Information can be shared by an unlimited number of users without losing value. The value of information can increase when shared, as a company can benefit from more users. According to Perera et al. (2014), information that is shared in the open market potentially increases the value of IoT applications. One potential opportunity, for example, is when one institution deploys sensors for a specific purpose and a second institution is interested in purchasing the data for a different information purpose. The more people that can access the data, the more economic benefits can be extracted from it (Bucherer & Uckelmann, 2011).

Law of information two: The value of information increases with use and it does not provide any value, if it is not used at all.

In contrast to most resources, whose value decreases with use, the value of information increases when it is used. In order to use it, potential users should be aware that the information exists. Bucherer and Uckelmann (2011) discuss how important it is that the decision maker is able to use the information in a valuable way.

Law of information three: Information is perishable and it depreciates over time.

On an operational level, the useful lifetime of information is relatively short compared to, for example, historical data about customers. Once the moment passes, in which information could have been used to make a decision, the information decreases in value (Moody & Walsch, 2002). IoT offers high-value information as the information is available continuously in real-time.

Law of information four: The value of information increases with accuracy.

Moody and Walsch (2002) state that the more accurate the information, the higher its value. Accuracy describes how precise the information is and therefore the probability that the information is correct. In some use cases, 100% accurate information is necessary whereas in other use cases, such as gambling, any percentage above 50 could be very beneficial. “If decision-makers know how accurate (or inaccurate) the

(17)

information they are working with is, they can incorporate a margin of error into their decisions” (Moody & Walsch, 2002, p. 7).

Law of information five: The value of information increases when combined with other information.

Information can be combined, compared, or integrated with other information and thereby become more valuable because doing this creates more information. With more information, comparisons can be made and information, from different sources, that relates to one object can be linked (Moody & Walsch, 2002; Bucherer & Uckelmann, 2011).

Law of information six: More information is not necessarily better.

For most economic resources, the more you have, the better it is. However, with information, decision-makers often try to find more information than it is humanly possible to process efficiently which may lead to inefficient decision-making (Moody and Walsch, 2002). This is referred to as “information overload” where the quantity of the information exceeds the boundaries of efficient decision-making (Bucherer & Uckelmann, 2011).

Law of information seven: Information is not depletable.

Glazer (1993) observes that contrary to other economic goods, information is self-generating. Standard economic goods are depletable, that is, as they are used their amount decreases. In contrast, using information by combining, integrating, or processing it, increases the amount of information that the user has. While standard economic goods can be scarce resources, information is not. The potential value of information increases when the remaining data is used to generate new information (Moody & Walsch, 2002).

2.5 Economic attributes for the value of information

The laws of information, and especially the influence they have on the value proposition of business models for IoT, are conceptually difficult to measure. Therefore, prior literature is analysed to identify elements that are appropriate to test the value of information provided by technology and whether they align with the laws of information. Specific economic attributes for value of information are identified from two main streams of research and their leading articles. The first is the technology acceptance theory which is studied by Wang and Strong (1996), Wixom and Todd, (2005) and Frokjaer et al. (2010) and more. The second is the

(18)

research into the quality of information for sensor networks by Bisdikian et al. (2009) and Bisdikian et al. (2013). These scholars discuss the processes and factors that mediate the realisation of economic value of information technology applications. Their applicability for the value of information for IoT business models is explained hereafter.

For the economic value of information, research often states that created value is a result of the utilisation of the information and the user satisfaction (Wang & Strong, 1996; Wixom & Todd, 2005; Frokjaer et al, 2010). Wixom and Todd (2005) explain that the value of information to the user is highly influenced by the user satisfaction with the information. However, Wixom and Todd (2005) assign a high weightage to the perceived usefulness of information, thus meaning that the usefulness of the information depends on the perceives of the user. Furthermore, Bisdikian et al. state that the value of information is “an assessment of the utility of an information product when used in specific usage context” (2013, p. 48). Thus, the value of information is subject to the use of the information and the descriptor context is related to the attributes of information. This means that the value of information depends on (1) the degree in which the user perceives usefulness and (2) the context in which the information is used. Though, the two combined should focus on the value of gaining information to reduce uncertainty (Coyle & Oakley, 2008).

The same is applicable to the quality of the information provided, which is strongly correlated with the value of information (Bisdikian et al., 2013). Quality of information cannot be defined. It is the users believe what superior quality of information is. The quality of information depends on its fit-to-use and thus, the context in which the information is used as described above (Bisdikian et al., 2013). According to Wei et al., for sensor networks, the quality of information:

“represents the goodness with which the information available describes the state of the world of interest, and is expressed using quality attributes such as accuracy (e.g., probability of detection of an event, accuracy in tracking an object, probability of misclassification), latency, and provenance (or lineage, describing all entities that have produced or touched this information, and hence may have altered its qualities). The quality of information directly impacts the effectiveness of the action plans that it recommends to decision makers”(2010, p. 93).

Information will be utilized by the decision maker when the quality and the value of the information reduces uncertainty, where the latter depends on the prior for generating economic value (Glazer, 1993; Bisdikian et al., 2013). Quality of information and value of information are both expressed by using specific attributes. The integration of user satisfaction and technology acceptance is characterised by attributes such as information relevance,

(19)

completeness, accuracy, timeliness, presentation, and so forth (Wixom &Todd, 2005; Coyle & Oakley, 2008; Bisdikian et al., 2013). Wixom and Todd (2005) argue that prior literature on user satisfaction explicitly considers information-specific attributes such as system reliability and information accuracy. Existing literature uses these attributes to investigate whether the design of IT system information, whose output is measurable and thus tested, is relevant to the user. These attributes can be linked to the concepts of the seven laws of information and potentially help to understand the realisation of economic value for the users of IoT applications. The value proposition of the business model should successfully offer the created value to customers (Osterwalder & Pigneur, 2010) and the value of information attributes delivered to the user creates economic value when the user perceives usefulness and satisfaction (Wixom & Todd, 2005; Bisdikian et al., 2013).

Furthermore, analysing the literature on the value of information, there are two leading attributes for economic value of information (Wixom & Todd, 2005; Frokjaer et al., 2010). The first one is accuracy and the second is relevance. Accuracy is an attribute that can be quantified in a percentage and is a metric that presents the uncertainty of the information (Bisdikian et al., 2013). It is critical to achieve a prominent level of accuracy to ensure that the economic benefits extracted from this information reduce uncertainty in the decision-making (Frokjaer et al., 2010). Jaynes states that “The great advantage provided by information theory lies in the discovery that there is a unique, unambiguous criterion for the "amount of uncertainty" represented by a discrete probability distribution, which agrees with our intuitive notions that a broad distribution represents more uncertainty than does a sharply peaked one, and satisfies all other conditions which make it reasonable” (1957, p. 622). This relates to the study of Lawrence (1979), who describes that the maximum expected utility for de decision maker under uncertainty crucially depends on the probability that the information is correct or yield the expected value. The economic concept that measures this is the expected value of perfect information (Coyle & Oakley, 2008). The expected value of perfect information is weighted against the probability of having perfect information (Coyle & Oakley, 2008). The probability that the provided information will happen or is true, is measured for IoT applications by accuracy, which is the likelihood that a predicted event will occur. Accuracy measures to what percentage the information is correct and results in the predicted event (Coyle & Oakley, 2008). The accuracy of information shows the probability or the precision of the information provided by IoT application. Lawrence’s formula (1979) for calculating the expected utility indicates that knowing the accuracy of the information is as important as the accuracy itself, because it allows the decision maker to calculate the expected utility. In addition, decisions increase in

(20)

accuracy when more information is provided to the decision maker (Lawrence, 1979; Frokjaer et al., 2010). However, more information does not automatically mean that the accuracy increases (Bisdikian et al., 2009), it should also be relevant. Relevance relates to the correctness of the information by how closely or complete the provided information is to the information that is requested (Bisdikian et al., 2009).

2.5 Conceptual Model

This study examines the cohesion of the seven laws of information and the value proposition to understand their applicability to the economic value creation through IoT business models. These laws are used to construct approaches to help understand the value proposition and characterise the economic value of information. Prior literature indicates that information has economic attributes that generates the economic value of information for technology applications. This study uses these economic attributes for the value of information to test the relevance of the laws of information for the value proposition and the economic value creation of IoT business models. It aims to find evidence concerning IoT business models in relation to the seven laws of information for the value proposition as the value proposition is considered the core element of IoT business models (Dijkman et al., 2015). The conceptual model is constructed to provide an answer to the following research question: “How to create economic value for IoT applications through business models by using the seven laws of information as design for the value proposition?"

The conceptual model (Figure 4.) describes the relation of the laws of information for the value proposition of IoT business models and creation of economic value. The seven laws of information characterize the elements that information holds for the value proposition and are expressed through the unique economic attributes for the value of information. These attributes are important for the economic value of information (Glazer, 1993). The quality of the information that brings value to IoT is described in the laws of information and expressed by the expected value of perfect information, which is presented in the conceptual framework (Moody & Walsh, 2002; Bucherer & Uckelmann, 2011; Bisdikian et al., 2013).

(21)

Figure 4. Conceptual Model

To summarize the context of value of information for IoT applications:

1). Information itself is the main holder of value for IoT applications (Glazer, 1993).

2). The value proposition is the essential part of IoT business models to deliver value as it entails the context for creating economic value for users (Dijkman et al., 2015).

3). The seven laws of information enable to outline essential elements for the value proposition to create economic value of information (Moody & Walsh, 2002; Bucherer & Uckelmann, 2011). These laws are expressed by specific economic attributes of information (Wixom & Todd, 2005; Bisdikian et al., 2013).

4). Economic value of information is weighted against the expected value of information delivered through the value proposition (Coyle & Oakley, 2008). The expected value of information depends on the attributes accuracy and relevance.

5). An IoT application delivered through a value proposition above reduces uncertainty and should offer a solution for a problem that the users face (Chesbrough & Rosenbloom, 2002).

Table 1 illustrates the essential elements of the value proposition that have been identified from literature. It depicts the Value of Information (VoI) attribute corresponding to each law of information.

Table 1. Laws of Information and the Value of Information attribute.

Law of Information

VoI Attribute

1 Information is shareable and can be shared with others without loss of value. Accessibility

2

The value of information increases with use and it does not provide any value, if it is

not use at all. Usability

3 Information is perishable and it depreciates over time. Timeliness 4 The value of information increases with accuracy. Accuracy 5 The value of information increases when combined with other information. Completeness

6 More information is not necessarily better. Relevance

(22)

3. Research Method

This study uses a combination of qualitative and quantitative research methods to explore the essential elements for economic value creation and test the relevance of the laws of information to the value proposition of IoT business models. The first section of this chapter provides an in-depth explanation of the research approach and the research design. The second section explains the process for sample selection and data analysis in both the research methods used.

3.1 Methodology

As there is limited existing research on the relevance of the laws of information to the value proposition of IoT business models, this study uses an explorative research approach (Saunders, Lewis & Thornhill, 2009). Exploratory research focusses on the causal relationships between concepts or variables and explanation building is used to test a theoretical proposition (Yin, 2009; Saunders et al., 2009). This study uses two methods for its exploratory research design: it begins with the collection of quantitative data and subsequently it collects qualitative data. The results from each of the research methods are analysed separately and then integrated in one discussion. In quantitative research, only numerical data is collected and analysed, whereas in qualitative research, non-numerical data is collected and analysed (Saunders et al., 2009). The construct validity of this research is increased by combining the two different research methods (Green, Caracelli & Graham, 1989), and this leads to a stronger conclusion (Teddlie & Tashakkori, 2003). Multiple methods for collecting data results in the triangulation of evidence and the combination of qualitative and quantitative data engenders a synergistic assessment of evidence (Eisenhardt, 1989).

Clarity is obtained by contrasting the results of the qualitative research with the surveyed quantitative research, which “emphasises measurement and analysis of casual relations among variables” (Denzin & Lincoln, 2000:8). Unlike qualitative results, quantitative methods more often generate results that can be generalised. However, quantitative methods are considered “thin” whereas qualitative methods are categorised as holistic (McClintock et al., 1979; Jick, 1979). This study triangulates results to ensure construct validity and rigour in research. Research triangulation means that different data is collected from various sources resulting in different interpretations of the researched phenomenon (Denzin & Lincoln, 2000; Yin, 1994). This is implemented by a mixed method and results in the contextual portrayal of the studied items (Jick, 1979). Using more than one method for the validation process decreases the probability that the error of the variance reflects the method rather than the trait (Jick, 1979).

(23)

3.1.1 Quantitative Method

The first research method is quantitative and collects data through a survey of IoT providers and its users. The aim is to test whether these IoT specialists believe that information value statements are important for economic value creation. The structure and questions in the survey questionnaire are designed using the literature on the laws of information and the economic specific attributes of information (Saunders et al., 2009). The laws of information are mainly descriptive and lack the elements to clearly quantify the data. However, the laws do not quantify data itself, but highlight ways in which the value of information is increased. Absolute value gains are hard to assess from the survey, but IoT specialists are questioned whether they exploit the laws of information to gain value for their customers and users of IoT products are asked whether these laws gain value for them. The specific attributes for the value of information, as proposed in the conceptual framework, are used to build the questionnaire. Each attribute was included in the questionnaire using three or more value of information statements, partially based on the value of information statements used by Bisdikian et al. (2013). The questions should be reliable, which means they should consistently produce the same equal answers. For this reason, each attribute is analysed using several statements, some of which are contradictory, in order to realise construct validity (Saunders et al., 2009). The questions were designed to be valid, that is they measure what the study aims to measure (Saunders et al., 2009). The survey questionnaire can be found in appendix 1.

The survey is conducted with the Likert scale as the measurement method. This is the most popular method used in contemporary research surveys (Park & Jung, 2009). On this scale, developed by Likert (1932), the typical test items are statements. In the Likert scale test, the participant comments on statements by indicating the degree of agreement with each statement. Dillman (2000) suggests that the Likert scale is best used with a 7-point scale, namely, strongly agree, agree, somewhat agree, neither agree nor disagree, somewhat disagree, disagree or strongly disagree. In this study, this means that respondents indicate their degree of agreement with statements regarding the value of information for the IoT product and the contribution of information attributes to its value.

3.1.2 Qualitative Method

As discussed in the conceptual model, existing literature identifies seven laws of information and attributes of the economic value of information. These are included in the conceptual model to validate its relevance to the value proposition in the IoT business models. This is mainly

(24)

theoretical as there is little or no empirical evidence for the relevance and validity of the developed theory in real-world cases. Qualitative research is the most appropriate design (Eisenhardt, 1989) for a better understanding of the relationship between the dynamics of the laws of information and the value proposition in IoT business models. As this study aims to answer the “how” and “why” questions, an exploratory approach is suitable (Yin, 2009). In the quantitative method, data is collected to validate the proposed model for value of information, but the understanding and context behind the answers is missing. To understand the real-life context, the unit of analysis is case studies and the data collection method is interviews. Case studies are deemed appropriate when there are no obvious boundaries between the events and the context (Yin, 2009). Yin (2009) states that case studies have a distinct advantage in situations where the investigator has limited or no control over the current events that are relevant for the research question. Interviewing professionals, as for example founders of IoT startups, in the field of the research is the most appropriate approach for explorative research (Saunders et al., 2009).

This qualitative method has a holistic unit of analysis as multiple cases are analysed against the seven laws of information. The aim is to find patterns in the value propositions of the cases and their natural processes. The survey results were used to provide direction for the interviews. This was done with the aim of gaining an understanding of the rationale behind the survey results and to capture a more holistic, complete, and contextual interpretation of the unit of analysis. Furthermore, the interviews aimed to explore the opinions of providers of IoT applications on the relation of the laws of information to the value proposition for IoT business models. The interviewees were asked about their IoT business model and the value that their IoT generated information delivers to their customers. The interviews were semi-structured and contained open-ended questions. Survey results were shared with the interviewees and they were asked about the logic and understanding behind their beliefs and understanding. The data was collected from multiple different companies to gain consistency in the results and the same interview guide was used for all interviewees. Founders of startups were asked more in-depth questions about how they create economic value of information through their IoT applications. A sample question is: “How does your company creates economic value from the information you provide?”. Moreover, the design of the questions aimed to check the significance of the key elements of the value of information in IoT businesses and validate whether these are complete for the value proposition. Questions like this were asked in the interviews: “The survey participants all agreed that increase in accuracy increases the value of information. Do you agree and why?”. Interview participants were thereby able to express their opinion first

(25)

and then go deeper into the reasons behind their beliefs. The question also gave them the opportunity to disagree with a statement and explain why. As IoT is a new technology, participants were often asked to provide examples for their reasons so that the researcher could gain a better understanding of their beliefs. This was useful as participants would describe why certain concepts are different for some use cases.

The results from the survey showed that the collected data was not sufficient for analysing the preference of the participants for certain value of information attributes, thus the interviews are used to ask participants to rank their preference. Furthermore, they had to explain which attributes were most important and why they believed this. The interview guide is included in the appendices (appendix 2).

3.2 Sample Selection & Data Analysis

This section explains the sampling and the data analysis for both the quantitative and qualitative methods. The first paragraph explains the sampling for the IoT specialists and the data analysis of the collected survey data. Subsequently, there is an explanation of the sampling of relevant cases for the interviews followed by the analysis of the data to determine the relevance of the laws of information to the value proposition for IoT business models.

3.2.1 Quantitative approach: Survey 3.2.1.2 Sample Selection

Quantitative data was obtained from a survey among technology specialists. A population in research defines a cluster from which a sample is drawn (Eisenhardt, 1989). The population for this study comprises the 500 technology specialists working in “Tech Quarters (TQ)” Amsterdam, founded by The Next Web. Through snowball sampling, this study aims to reach more technology specialists who are willing to fill out the survey (Atkinson & Flint, 2001). This was deemed necessary after the first survey attempt had a low response rate. Therefore, technology specialists were approached individually and asked to approach colleagues in their field. A sample size of 63 was involved in the study. Out of this sample size, only 52 (82.6%) participated actively in the filling of the questionnaire. Out of these 52, there were 12 participants that did not fill out the questionnaire completely. These partial responses are filtered out in the results below. Only complete results have been used for analysis which resulted in 40 responses to analyse. The sample statistics are shown in Table 2, Table 3 and Table 4 below.

(26)

Value Proposition for IoT Business Models, M.V. van Hage

Table 2. Involvement in IoT

Degree of involvement % Count

No Involvement currently 22.50% 9 0-1 year 32.50% 13 1-3 years 35.00% 14 More than 3 years 10.00% 4 Total 100% 40

Table 3. The use of IoT products

% Count

Yes 82.25% 33 No 17.50% 7 Total 100% 40

Table 4. role in IoT

Role in IoT % Count

User: I am a Customer 60.0% 24 Provider: I developed/co-developed a IoT application 37.50% 15 Provider: I obtained an IoT application and provide it to customers 2.50% 1

Total 100% 40

The data collection above shows that participants had to fill out whether they were involved in IoT and to what degree. They also had to answer whether they use IoT products right now. For a good analysis of the results, the participants that filled out that they are not involved in IoT and do not use any products have been extracted. The nine respondents answering, “no involvement in IoT at the moment” included two respondents that are not involved but are using an IoT product. This resulted in only 33 relevant participants for the data analysis.

3.2.1.2 Data Analysis

The data collected from the survey was imported into IBM SPSS 21.0. SPSS is a predictive analytics software that is appropriate for the analysis of the survey data. The survey data analysis aimed to test whether the variances between the IoT providers ant its users in the sample are equal. This means it analysed whether the beliefs of the providers of IoT applications are aligned with the users’ perception of the value of the information provided by the IoT applications. Levene’s “robust” test for the equality of variances was used to analyse this. Herbert, Hayen, Macaskill and Walter (2011) show that with this method, the generated confidence intervals are near to nominal coverage, even for small sample sizes or unequal ones.

(27)

The test shows the differences in the mean(μ) outcomes of two or more groups in a sample (Gastwirth, Gel & Miao, 2009). Gastwirth et al. (2009) explain that there are an extensive number of statistical tests for this, but most are not robust as they assume normality. The Levene test is an alternative that applies the F-test for the absolute deviations of the comments from the groups mean. In contrast to other tests, the Levene test is robust to non-normality (Gastwirth et al., 2009). Therefore, the Levene test is preferred for the examination of homogeneity between variances. For analysis of the variances, the Levene test assumes equal variances. The variances are unequal when the outcomes of the t-test are significant. For the analysis of the data from this study, a two-tailed significance level of p <0.05 was adopted.

In this study, the groups tested were group one: User-Customer (0) and group two: Provider (1). The following parameters were set: Strongly agree = 1, Agree = 2, Somewhat Agree = 3, Neutral = 4, Somewhat Disagree= 5, Disagree = 6 and Strongly Disagree = 7. This results in outcomes of the mean being between 0 and 4 when the respondents agree with the statement and between 4 and 7 when the respondents disagree with the statement. When the mean is close to 4, most respondents were neutral with the statement.

The reliability of the survey has been tested with the Cronbach’s Alpha, which measures internal consistency of a scale or test (Saunders et al., 2009; Tavakol & Dennick, 2011). Internal consistency means the extent to which objects in a test measure the same construct. In this study, internal consistency means the extent to which the individual laws of information all measure the economic value of information for IoT applications. The appropriate threshold for the Cronbach Alpha is 0.7 of higher to be accepted as reliable coefficient (Tavakol & Dennick, 2011).

3.2.2 Qualitative approach: Interviews 3.2.2.1 Sample Selection

The interview participants were chosen based on theoretical sampling, which means by analytical generalisation of cases and whether they met the following criteria: IoT specialist, IoT startup or established business, and availability at a convenient location. Interview participants were carefully chosen using detailed criteria, which required that cases must contain a value proposition based on IoT applications or information services. The participants in the interviews are outlined in table 5 and their involvement with IoT is described. Within the network of the researcher there were four cases willing to participate in this research. Three cases were found using web search and LinkedIn. Several interesting IoT companies were

(28)

approached, of which three agreed to participate. The final participant was found using snowball sampling, in which the researcher’s office building partners were approached. One partner, who provides the office building with sensors, was willing to participate. The utility of the information collected by the cases was maximised by selecting cases based on the information evidence they were expected to provide (Flyvbjerg, 2006). All interviews were conducted in English, except for one, where the participant preferred Dutch. In one participant company, both founders were willing to participate in the interview.

Table 5. Interview participants

CASE COMPANY INTERVIEWEE INVOLVEMENT WITH IOT

1 A IoT startup Co-Founder Developer of an IoT product for agriculture.

2 A IoT startup Co-Founder Developer of an IoT product for agriculture.

3 B IoT startup Co-Founder Connects smart and IoT products in one mobile application.

4 C Internet Company

Founder Collect data over the internet and provides information to its customers.

5 D IoT startup Founder Offers wearable and IoT solutions for people who need protection and security

6 E Advisory Company

Partner Advisory for IoT startups & founder of a tech startup himself.

7 F IoT startup Founder Sensor manufacturer

8 G IoT startup Co-Founder Provides smart building technology by making data available from sensors

3.2.2.2 Data Analysis

As explanation building is testing a theoretical proposition, this study uses a pattern matching procedure to test whether the data collected supports the theoretical proposition (Yin, 2009; Saunders et al., 2009). The participants’ responses to the interview questions helped to identify the economic value creation of IoT and to understand its value proposition. The participants were asked about the economic value their IoT product creates and to explain their business models. The interviews were all recorded with the permission of the participants. The researcher transcribed all recorded data and systematically coded and analysed the transcriptions to identify common themes and patterns in the interview responses about the value of information created by IoT applications. A Computer Assisted Qualitative Data Analysis Software (CAQDAS) program (Atlas.it) was used to organise the analysis of the collected data. Using a software has the advantage that it can impose continuity and, if systematically used, enhances the “methodological rigor” and transparency (Saunders et al., 2009). Saunders et al. (2009) explain that in explanation building, the expectation is that the interview question design will result in the emergence of groups of codes during analysis.

(29)

Consequently, the data collection and analysis commenced with an initial group of concepts that was derived from the conceptual framework and from the theoretical propositions allied to the research question (Saunders et al., 2009). Therefore, the data was coded based on the seven laws of information and their related economic specific attributes. Subsequently, as interviewees were asked about the value proposition and the value their products deliver to their customers, these creating economic value and the value proposition were also used as categories of code. Categories can change as their appropriateness depends on the answers the participants provide (Saunders et al., 2009).

There are three ways of coding and the first is axial coding. This means that theoretical terms are used to derive names for code (Strauss & Corbin, 2008). The two other ways are open coding and in vivo codes. Open coding means that names for codes emerge from the collected data; in vivo coding uses code names that are actual terms used by the interview participants (Strauss & Corbin, 2008). Code categories need to be meaningful in two ways: they should be linked to each other and they should be representative of the data. A well-structured framework improves the data analysis (Saunders et al., 2009). Although axial coding was used for the first line of code categories, actual terms used by the participants made it clear that there were additional concepts that relate to the value of information and the laws of information. Therefore, in vivo coding was used to categorise these concepts.

The first line of code categories starts with the value creation – value proposition for every interviewees IoT product. The codes thereafter are derived from the following laws: Law 1 – Sharing information (Accessibility), Law 2 – Usability, Law 3 – Perishable and timeliness, Law 4 – Accuracy, Law 5 – Combining information (completeness), Law 6 – Quantity of information (Relevance) and Law 7 – Self-generating (Reliability).

(30)

4. Results

4.1 Survey results

The data collected through the survey is used to analyse whether the participants support the elements, that are identified in literature, for value creation for IoT applications. Table 6 illustrated the degree of support for the main value of information statements.

Table 6. Support for value of information statement.

Value of Information Statement Support

The VoI does not reduce when shared. 82% The VoI increases when more people utilize the information. 97% The VoI depends on the quality of the information. 97% The VoI depreciates over time. 47% The VoI is higher with a higher accuracy. 100% Knowing the accuracy is equally important as accurate information. 76% The VoI increases when information is combined. 92% The VoI increases with more complete information. 95% The relevance of the information is more important than the quantity. 84% Too much information clouds the decision making. 81%

After the analyzing the support for the theoretical framework, the data has been used to evaluate whether the believes of providers of IoT applications about the value that IoT creates are aligned with the believes of its users for the essential elements for value creation by IoT. Before findings can be drawn from the collected data, the reliability of the survey has been tested by the Cronbach’s Alpha, which is shown in table 7.

Table 7. Cronbach’s Alpha

Reliability Statistics Cronbach's Alpha N of Items

Complete test ,853 43 ,00 User ,752 43 1,00 Provider ,901 43

The SPSS output of the single statements influences on the Cronbach’s Alpha analysis showed that statement 4.4 “Information holds value when it has a 50% margin of error” has been poorly understood by the respondents. This might be a result of many respondents that remained neutral to this statement. They might not have understood that a 50% margin of error is equivalent to the chance of randomness. Tavakol and Dennick (2011) explain that excluding an item needs to be done cautiously and should prevent a decrease of the validity. As accuracy has five value of information statements, thus four remaining, the statement 4.4 has been taken

Business models for the Internet of Things : can the laws of information build the value proposition