Acceptance and Diffusion of Services Based on Secure Elements in Smartphones – Study Design and First Results of the Pretests

Abstract

With the increasing spread of digital services and apps, the risk and damage potential of attacks and identity theft increases. In addition, there are services, such as season tickets on public transport, that must be protected at a high level of security so that there is no significant risk of abuse.

The OPTIMOS 2.0 Project wants to establish an open ecosystem with a Trusted Service Manager as core element serving as a secure storage location for cryptographic keys etc. on the smartphone. The set up of such system depends on suppliers offering their services on that platform as well as customers, who accept, demand and use these services. Both sides have to be acquired to ensure long-term success.

In this paper we would like to present an approach to achieve the user acceptance of different services using an Secure Element to provide hardware secured services. With this we want to support the decision to invest quite a lot money for the initialization of the OPTIMOS 2.0 ecosystem.

1 Motivation

Product innovations developed by researchers and engineers are often faced with the question whether the customer recognizes and appreciates the advantages, benefits and added values of a product. Although the product is superior to competing products, it may be that the innovation does not sufficiently satisfy the users’ needs, thus preventing rapid adoption. The so called technology-push-effect¹ is one of the main problems to be solved.

Fig. 1.

Different roles of the OPTIMOS 2.0 ecosystem and their trade relations. The fields highlighted in blue show the roles represented by partners in the project. (Color figure online)

Furthermore, implementing new digital technologies or new digital products usually require high development and acquisition costs. Interfaces have to be specified, software has to be programmed and servers have to be initialized. The strategic decision whether to invest or not might sometimes made on an inaccurate or uncertain data basis. Decision makers look for reliable information supporting their decision making processes. The knowledge how a product is received by the user and how sales figures will develop in the coming periods can be a good decision-making aid.

The aim of the concept presented in this paper is to support the decision process by providing an acceptance and sales analysis.

In the next section we introduce the project OPTIMOS 2.0. Then we discuss the basics of adoption theory with regard to the different use cases. In the last step, we outline the concept of user acceptance and the establishment of the Bass Model.

2 Project OPTIMOS 2.0

The aim of the project OPTIMOS 2.0² is the development of an open ecosystem for mobile services based on Secure Elements (SEs) integrated in modern smartphones. Being a part of the Near Field Communication (NFC) technology Secure Elements (SEs) can be used as a secure storage of sensitive data and information. This opens up a wide range of services that are not possible today because of data protection criteria or security concerns.

During the project essential system parts will be implemented, demonstrators will be tested and the field introduction will be prepared. Although, the system is initially intended for the German market, all relevant functions and interfaces are developed in an open and transparent process and were introduced into international standardization.

A number of partners from different markets have come together to fulfill these task. In Fig. 1 different roles of the OPTIMOS 2.0 ecosystem are presented.

With respect to the participating project partners we are focusing on three use cases:

• Public transport: In Germany electronic seasonal ticket are only provided by smart cards. Those high priced ticket are not available for smartphones, because of the risk to be fudged. The OPTIMOS 2.0 system promises ticket storage to be forgery-proofen and can even be checked in offline mode. It is also conceivable that a user may purchase a ticket anonymously or pass it on to friends.

• eGovernment: The German identity card can already be used today to identify oneself online and to carry out administrative procedures. Up to now the acceptance to use electronic eGovernment services with the Electronic Identity (eID) card is still low, partly because the service is perceived as not very user-friendly. Storing a digital copy on the smartphone should make it much more comfortable and easier for citizens. By using the SE it seems possible to achieve the substantial level of trust according to the eIDAS³ regulation.

• Car sharing: In car sharing two different use cases are considered. First, during the registration process, the user can transfer his personal data from an eID app (eGoverment use case) or by reading the data directly from the ID card via NFC. In the second case, the user can use the car key stored on the SE of his smartphone to open the rental car. With the NFC technology, the car can also be opened offline (e.g. in underground garages).

A number of other use cases are possible (e.g. online check-in with eID in the hotel sector or access authorizations and room keys on the smartphone), but will not be discussed in this study.

Since the start of the project in 2018, our research activities have focused on the following subtasks:

• 1. Analysis of the central Trusted Service Manager (TSM)-platform with regards to the characteristics of digital platforms (e.g. defined by Parker et al. [8])

• 2. Development of a role model including the trade relations of the several partners to give an overview about the complex ecosystem validating by the project partners

• 3. Estimation of the acceptance and diffusion of different use cases. If there are differences in adoption speed, we want to name the causes and derive solution strategies. Once we will have validated the forecasts for the sub-markets, we want to aggregate them, so we can check, whether the TSM and the entire ecosystem can be economically successful in the long term.

Parts 1 and 2 are already finished and some of the results have been published (e.g. [12]). During the business modeling process the project partners asked for a sales forecast and sales figures, because they have to report internally to their strategic deciders, whether the project will or will not be successful. This paper focuses mainly on part 3.

3 Attributes of Innovations

According to Rogers [9, p. 15], successful and fast adopted product innovations promise the user a relative advantage compared to existing products, are highly compatible with previous experience, show a low complexity, are testable without major consequences and are observable (for others).

Kee** this definition in mind, we now want to have a closer look at the innovations of the various use cases.

3.1 Relative Advantage

Each of the innovations considered promises a fast handling process through the use of the smartphone without any media disruption and thus a time saving for the user.

• Public transport: After the implementation of the system it is possible for users to buy seasonal tickets via their smartphone. Compared to the common situation, where tickets are only available paper based or stored on a smart card, the innovation can be relatively advantageous for the user. Also the purchase process could be much faster⁴.

• eGovernment: By storing the identity card on the smartphone, it is no longer necessary to take the identity card out of the wallet at an online identification process. Without a media disruption it seems to be much easier for citizens to use the online identification function.

• Car sharing: As easy onboarding process by automatically transfer of personal data to the registration form can be an advantage for the user compared to the manual input.

  The keyless car function allows the customer to drive off quickly with the vehicle, as there is no need to go to a counter or key cabinet. The perceived barrier before borrowing a vehicle could be minimized.

3.2 Compatibility

The importance of the smartphone for the completion of everyday tasks is constantly increasing with the result, more and more users are becoming better at handling apps and mobile services⁵. They can also draw on this knowledge when using the product innovations considered here.

The rapidly increasing use of mobile payment services trains users in the use of NFC technology and its applications. This will also benefit eGovernment services, as users will have more and more confidence in the NFC technology as they use it. This makes it easier for similar services to establish.

• Public transport: Using the smartphone for seasonal tickets should not represent anything completely new for many users, as they already purchase single tickets via smartphone apps.

• eGovernment: Citizens, who already use the eID function in combination with the identity card will only be able to draw on their experience knowledge if using a similar service, but without the use of the identity card.

• Car sharing: The acceptance of the simplified registration depends on the acceptance of the eID function (eGovernment).

  Using the keyless car function is similar to the on boarding process before a flight. This allows a number of users to draw on their experience.

3.3 Complexity

All product innovations promise a simplified process flow. But installing the service provider app and the associated applet for the SE can make acceptance more difficult.

3.4 Trialability

The possibility of trying out the various functions without making any significant commitment is limited in all use cases.

3.5 Observability

• Public transport: When passengers identify themselves to a ticket inspector with their smartphones, other customers can observe and become interested about those systems and services. Later on, some of them will be become adopters.

• eGovernment and Car sharing (Registration): Registration with an online service is usually private, without other people being able to observe the process (this makes especially sense, because of data protection reasons). Using the eID function with the smartphone will not change this behavior.

• Car sharing (Keyless car): Unlocking a parked car by using a smartphone can be observed by other persons. Some of them can copy and reproduce the procedure.

4 Concept

Whether the user can actually recognize the advantages of the product innovations and how fast the diffusion progresses, we will found out with the now presented acceptance investigation. In Fig. 2 you will find an overview of the entire concept.

Fig. 2.

Schematic overview of the sales forecast concept

4.1 Pretest Public Transport

In January 2019 we started with the first pretest and interviewed over 2000 test persons with an online questionnaire regarding the acceptance of electronic services when purchasing tickets in public transport. Afterwards, we calibrated and validated a Multinomial Logit-Model (MNL) to explore the influencing factors by focusing on the question: Why people prefer tickets on paper, on a chip card or on their smartphones?

The main results⁶ are,

• Test persons, who like a fast purchase procedure, tend to prefer smartphone tickets

• Smartphone tickets are seen as more environmentally friendly

• Test persons, who wish to remain anonymous or who would like to be advised at a counter, usually prefer traditional tickets

• Younger people and heavy smartphone users tend to use smartphone tickets more often.

The results will be used to limit the number of relevant alternatives, attributes and attribute levels of the Choice Based Conjoint-Analysis (CBCA), so we can focus on aspects, the people really know or like to have.

4.2 Pretest Carsharing

We conducted a second survey in early 2020 focusing on car sharing users and using the Kano model to analyze customer needs. A total number of 205 respondents took part in this online survey. The main results are:

• To register, users want to enter their data manually into an input mask as they have long been accustomed to. Alternatives, such as using the eID or platform IDs, are not preferred.

• The obligatory driving license check by a copy stored on the smartphone or by using electronic services (e. g. Video-Ident) was well accepted by many test persons.

4.3 Choice Based Conjoint-Analysis

The Conjoint-Analysis (CA) is one of the most used method to determine the preferences and perceived benefits of customers. It is assigned to the decomposition methods. We are observing selection decisions and draw conclusions about the use of the several product components. Because we would like to forecast the market development, we looked for a CA, where the test persons have to choose there favorite product from a set of products (Choice-set), so we can simulate a (real) purchase decision. We also want to grant a non-selection option. By using a Choice Based Conjoint-Analysis (CBCA), all of these requirements are met.

Stimuli Refinement and Experimental Design. In Table 2 the first steps of the CBCA are presented⁷. After describing the problem to be investigated, we have determined the population from which we want to draw a sample. In the next step, we shape the stimuli by looking for relevant alternatives and attributes of the products using the findings of the previous pretests.

In public transport we identified 4 relevant attributes with 3 to 4 attribute levels each (see Table 1). The total number of Stimuli S is

$$\begin{aligned} S = \prod _{j=1}^{J} M_j \end{aligned}$$

(1)

So there is a total of 144 stimuli in public transport, 192 stimuli in car sharing and 256 in eGovernment.

Table 1. Estimated attributes and their different levels in public transport use case

Table 2. Experimental design of the CBCA

To avoid overburdening the test persons we decided to present only three stimuli plus a non-selection option ($K=3+1$) per decision. The number of possible choice-sets C is

$$\begin{aligned} C = \left( {\begin{array}{c}S\\ K\end{array}}\right) = \frac{S!}{K!\left( S-K!\right) } \end{aligned}$$

(2)

Because the number of stimuli is very high, the total number of choice-sets in public transport sums up to 17 178 876. However, for a test user it is impossible to evaluate all these variants.

To deal with this problem, we have to select a number of choice sets R randomly. Following the recommendations of Backhaus et al. [1, p. 185] we present a total number von $R=12$ choice-sets to every test user.

Utility Model. A number of models are available for modelling the utility $u_s$ of a stimulus s, including the vector model, the ideal point model and the part worth model. The latter we regard as the most suitable for our use cases.

$$\begin{aligned} u_s = \sum _{j=1}^{J} \sum _{m=1}^{M} b_{jm} \cdot x_{jm} \end{aligned}$$

(3)

where

$$\begin{aligned} \begin{array}{lcl} u_s &{}\quad =&{} \quad \text {utility of stimulus }s \\ b_{jm} &{}\quad =&{}\quad \text {Partial utility of attribute }j\text { and attribute level }m \\ x_{jms} &{}\quad =&{}\quad 1,\text { if stimulus }s\text { has the attribute }j\text { and attribute level }m,\text { else }0. \end{array} \end{aligned}$$

With the so called compensatory utility models, a lower utility of one attribute level is outweighed by the higher benefit of another.

One assumption of the aggregated model is, that the benefit expectations of the consumers are homogeneous [1, p. 215]. But this is not the case in reality. This problem can be solved by breaking the group down into subgroups with similar behavior by using a A Priori Segmentation, the Latent Class Approach or the Hierarchical Bayes Approach [1, pp. 217] estimating a separate choice model for each subgroup.

Choice Model. Because we want to use the Bass model afterwards, we were looking for a model, which designs the individual choice of every test user and calculates the probability how likely it is that a user will choose this alternative. The Multinomial Logit Model was developed for such tasks, so we chose it.

$$\begin{aligned} P_i(k) = \frac{\mathrm {e}^{\beta _i u_k}}{\sum _{k=1}^{K} \mathrm {e}^{\beta _i u_k}} \end{aligned}$$

(4)

$P_i(k)$ is defined as the probability a test user i chooses alternative k. $\beta _i$ can be interpreted as the measure of rational behavior of the test user. If $\beta _i=0$, the user selects random alternatives. If $\beta _i \rightarrow \inf$, the Logit Model behaves like the Max Utility Model. This indicates that the respondent makes extremely rational decisions.

Fig. 3.

Relationship between S(T) and Y(T) on the basis of an exemplary product

4.4 Bass Model

The model of Bass [2] was originally published in 1969. It allows to model product growth rates and adoption processes. Bass was inspired by the first approach made by Rogers [9] in the early 1960s, because he tried to model the assumptions of the theoretical research of the time [2, p. 216].

The probability P(T) a new customer will buy a product at time T is calculated as follows

$$\begin{aligned} P(T) = p + \frac{q}{m} Y(T) \end{aligned}$$

(5)

where

$$\begin{aligned} \begin{array}{lcl} p &{}\quad =&{}\quad \text {Innovation coefficient} \\ q &{}\quad =&{}\quad \text {Imitation coefficient} \\ m &{}\quad =&{}\quad \text {Potential initial sales in the observe period}\\ Y(T) &{}\quad =&{}\quad \text {Number of customers at time }T\\ \end{array} \end{aligned}$$

The variable p could be interpreted as the growing number of innovators who come up with the idea to consume the product totally independently. In contrast, the variable q indicates how many new customers choose the product by imitating customers already adopted the technology, service or product.

The number of sales S(T) at time T is

$$\begin{aligned} S(T) = P(T)\left[ m-Y\left( T\right) \right] = pm + \left( q-p\right) Y(T)-\frac{q}{m}Y^2(T) \end{aligned}$$

(6)

In Fig. 3 you can see the characteristic curves of the model using a exemplary product.

Estimating m. Usually it is difficult to determine m, because you can look at a market historically, but you cannot estimate, how a new product would be adopted by the market in the upcoming periods. By using the CBCA we are identifying the probability users would choose the new product in relation to the probability that they use a product already established on the market. With this probabilities, we can split the market forecast into the several products and can focus in particular on the diffusion of the product in whose development we are most interested (the usage of services based on OPTIMOS ecosystem).

The potential initial sales m(k) of product k can be estimated by

$$\begin{aligned} m(k) = m \cdot \sum _{i=1}^{I}{\frac{P_i(k)}{I}} \end{aligned}$$

(7)

Estimating q and p. The innovation coefficient p and the imitation coefficient q are normally estimated on the basis of historical market data. For product innovations there is usually no data available, because they are not sold yet. To solve this problem Schühle [10, p. 152] published his approach, in which he identified analogue products⁸. Schühle uses the customers’ choice of products to draw conclusions about goods of which he has the historical adoption curve. Subsequently, he estimated their innovation and imitation coefficient and used this for the innovative product he is focusing.

When looking for analogue goods, it may help to look for products that are similar to the attributes identified in Sect. 3.

5 Outlook

The presented concept is to be implemented in the spring. The first results are expected in June 2020.

Acronyms

Notation	Description	Page List
CA	Conjoint-Analysis	7
CBCA	Choice Based Conjoint-Analysis	6–8, 10
eID	Electronic Identity	3, 5, 7, 8
MNL	Multinomial Logit-Model	6
NFC	Near Field Communication	2, 3, 5
SE	Secure Element	1–3, 5, 8
TSM	Trusted Service Manager	1, 3