key: cord-0823439-y301dzi2
authors: Filotto, Umberto; Caratelli, Massimo; Fornezza, Fabrizio
title: Shaping the digital transformation of the retail banking industry. Empirical evidence from Italy
date: 2020-09-06
journal: nan
DOI: 10.1016/j.emj.2020.08.004
sha: 8a55cd9a1219286662f49737160d17d8657f4104
doc_id: 823439
cord_uid: y301dzi2

Although numerous studies have investigated the determinants of users’ acceptance of online channels, little attention has been paid to the continuous usage of a direct bank. To address this research gap, we analysed data collected via semi-structured interviews and a survey questionnaire distributed in 2017 to a sample of Italian bank customers. Responses were evaluated using a different statistical methodology based on the Shapley Value regression analysis. In brief, user-friendliness and economic advantages appear to be more decisive at an early stage of adoption. Moreover, structural assurance mechanisms (e.g. clarifying security policies and guarantees) have a robust impact on keeping customers online and loyal to electronic channels. Surprisingly, accessibility and compatibility do not seem to be determinant factors.

Digital transformation is one of the most significant phenomena in retail distribution (PwC, 2019; OECD, 2014) . The incentive, and thus the drive, is on both sides; companies can use delivery channels that are not only more efficient but that actually allow them to sell products customised to individual needs; consumers, on the other side, can obtain remarkable cost savings and access to services and goods with practically no time or space constraints.

They also confirm the importance of security concerns, and highlight difficulty of access to delivery channels and lack of face-to-face contact as factors that discourage adoption. Liao & Cheung (2002) identify additional factors that affect consumer acceptance of direct banking, i.e. accuracy (operational precision), transaction speed and user involvement (the control that the individual can exercise over transactions). Other factors highlighted include the trustworthiness of the electronic bank and privacy issues (Akinci et al., 2004) . The availability of sufficient support and information from direct channels also features prominently within the research conducted (Lichtenstein & Williamson, 2006) , e.g. Durkin et al. (2008) establish that a customer's propensity to purchase online decreases with the complexity of banking services. In other words, for customers who have simple needs typified by services of low complexity (bank savings accounts, home contents insurance, credit cards), supplementary information about the buying decision is not required and an online environment can be used more confidently and without the need for additional face-to-face support. Conversely, time-consuming information searches contribute negatively to the propensity to use a direct channel as a means to purchase banking services.

In order to ensure the widespread adoption of direct banking, Floh & Treiblmaier (2006) suggest that, rather than simply investigating which is the ideal medium for purchasing banking services, the problem of how to keep customers online and loyal to a specific supplier should be addressed. Thus, they shift the focus from enticing customers to the online environment to retaining them (favouring a lasting virtual relationship). Coherently, their work analyses which factors might induce customers to stay with a particular direct bank (i.e. loyalty) instead of switching suppliers. Their results show that in an online environment, customer loyalty is directly affected by cognitive and emotional (i.e. functional and psychological; Laukkanen et al., 2009) constructs, namely satisfaction -based on the customer's cumulative experience rather than being a transaction-specific phenomenon -and trust in a direct bank. These findings are consistent with the most recent findings collected by Yuan et al. J o u r n a l P r e -p r o o f Yousafzai et al. (2005) clarify that the lack of customer trust -both in the attributes of the bank and in the overall online environment -remains an obstacle to the widespread adoption of direct banking 1 .

Their study highlights that customers' trust in direct banking has some unique dimensions: the distant and impersonal nature of the online environment; the extensive use of technology; and the inherent uncertainty of using an open technological infrastructure for transactions. The authors demonstrate that, given this context, the direct bank's trustworthiness mainly depends on structural assurance, defined as the belief that the success of the banking transaction "is likely because contextual conditions such as promises, contracts, regulations and guarantees are in place". Consistently, they suggest that determining a schema for developing and communicating online trust (e.g. by giving customers clear and obvious access to security and privacy policies) helps to transform a potential customer from a curious observer to one who is willing to perform direct banking transactions.

Although numerous studies have investigated the determinants of users' acceptance of online channels, little attention has been paid to continuous usage of a direct bank, where in fact continuous (i.e. repeated, loyal) behaviour is more important -from the point of view of a banking institutionthan users' initial (i.e. occasional) adoption (Tsai et al., 2014) . To address this research gap, the present work intends to answer the following central research questions: do the levers to encourage the usage of electronic banking change significantly if a customer is at an early stage of adoption rather than in a retention phase? and, moreover, are structural assurance mechanisms (as defined by Yousafzai et al., 2005) essential levers for keeping customers online and loyal to a direct bank?

To tackle these issues, the paper investigates the adoption of direct banking services using data gathered via semi-structured interviews with customers and a unique dataset collected through a survey. The study is carried out based on first a qualitative and then a quantitative investigation conducted in 2017 in collaboration with an international market research institute.

Because we are particularly interested in knowing why individuals are slow to adopt direct banking, our purposive sampling places an emphasis on the customers of banks with high branching intensitycommonly called "traditional banks" -including both (occasional) online users and online non-users for banking transactions. The banked population is the market segment thought most likely to adopt online banking (Lichtenstein & Williamson, 2006) . However, traditional banks -that control the largest share of banked customers -are the ones that seem to experience the greatest difficulties in enticing their clientele to the online environment (Dandapani et al., 2018; Corrocher, 2006) . Thus, the rationale for observing direct banking from the perspective of traditional banks is twofold: on the one hand, digitalisation is a powerful solution to the problems of efficiency and cost reduction that traditional banks must necessarily deal with given the low profitability of the retail business (Canato & Corrocher, 2004) ; on the other hand, the general shift to digital services, e-commerce and smart working, is exacerbating the competition from new players (the FinTech companies), which are technology savvy and do not have the burden of legacy infrastructures (Filotto, 2000) . Understanding how customers could move to the digital delivery approach and how fast is thus crucial for restructuring incumbents' supply chain and interface model. Furthermore, our sample group consisted of Italian customers. In fact, the Italian case presents some interesting peculiarities, since the industry has experienced an increase in the number of retail bank branches over time, notwithstanding the potential for network rationalisation due to digital adoption. This growth in physical branch numbers is in marked contrast to the process of network reduction that has occurred in most European countries (Corrocher, 2006) . At the same time, however, the population in Italy appears particularly reluctant to adopt direct banking. According to Eurostat, in 2019, Internet services offered by banks are accessed by only 36% of Italians, with the EU average being at 55%; a figure which places Italy among the countries with the lowest rate in the 27 EU Member States (Dumicic et al., 2015) , despite a significant and growing Internet penetration rate.

Indeed, about 84% of Italian households have Internet connection, against the EU average of 89%.

These features add an interesting dimension to the work and provide a unique insight into the nature and success factors of banking in such an environment. The results of the present study suggest that the development and dissemination of direct banking should be approached by means of different levers if a customer is at an early stage of adoption rather than in a retention phase. Moreover, our findings confirm the contributive effects of structural assurance on continuous online banking usage.

In addition, our results show that current direct banking has catered to the rational hemisphere of our needs: prices, convenience, and transparency. This has worked, but everything that can be labelled "emotional" was not even taken into consideration, limiting the potential for adoption and making the process slower than it could have been. Therefore, traditional banks that want to accelerate the transition should be ready to develop an approach that does not simply emphasise "hard facts" (costs, higher revenues, 24x7, etc.), but also leverage the soft profiles of the customer relationship. What clients are asking for is not simply a better bank but a different kind of bank, able to combine efficiency, relationship and care. Coherently, the study highlights how several service elements, which could increase and complete the current model with regard to relationship contents -such as customised interaction, light branches, mature range of products -are capable of positively driving the propensity to use direct banks.

The paper is organised as follows: we review the academic literature on the adoption of direct banking by customers in this first section and outline the research methodology adopted in the second section. In the third section, we explain our results, which are analysed in relation to the study's central research questions. Finally, in the fourth section, we present conclusions and proposals that assess some of the strategic implications of the work's findings for researchers and practitioners.

From the brief literature review just provided, it is possible to obtain some important indications, which have significantly influenced the structure and conduct of the present work. First of all, there are several different factors affecting consumer adoption of online banking. Therefore, it is necessary to support researchers and practitioners in narrowing the playing field by finding a parsimonious set of success factors able to improve the value proposition on offer through a virtual interface J o u r n a l P r e -p r o o f (Yousafzai et al., 2010) . Second, it is crucial to develop the analysis by including the emotional (rather than cognitive) constructs which are responsible for keeping customers on the web, playing an important role in predicting the usage intention of direct banking (Peral-Peral et al., 2019) . Moreover, existing studies which examine the emotional-based propensity to purchase online are largely grounded on surveys that, alone, are not always able to uncover the deeper issues identified by interpretivist methods (Lichtenstein & Williamson, 2006) . Hence, a two-stage research study is useful (Howcroft, Durkin, 2000) . Stage 1 of the overall work is qualitative and consists of semi-structured interviews with customers. Stage 2 follows sequentially and involves the distribution of a survey questionnaire to a stratified sample of customers. In the second stage, the findings from Stage 1 inform the research propositions.

The first qualitative step of our investigation was conducted in March 2017 and entailed the following activities: a focus group lasting 2.5 hours with eight customers of traditional banks who use branches but also the Internet for their banking transactions (i.e. Internet banking); a focus group lasting 2.5 hours with eight multi-bank customers (account holders at both traditional and direct banks); an insight group lasting 3 hours with four customers of traditional banks who do not use Internet banking; eight individual interviews lasting one hour and an online forum with customers of direct banks; and a T-group lasting 3 hours with six customers of both traditional and direct banks, half and half.

The participants in the qualitative step are current account holders, all ATM users (reflecting an interest in technology), who have carried out at least four different operations with the bank in the last 3 months. They are between 35 and 65 years old, mainly belonging to the upper mass segment rather than to the mass market, with Internet literacy. Men make up 54% of the respondents and women make up 46%. The decision to interview predominantly middle-aged individuals with higher incomes and males is consistent with previous studies reporting that these demographic characteristics are associated with the adoption of different banking channels, especially Internet banking (Akinci et al., 2004; Mattila et al., 2003; Wang et al., 2003) .

In the qualitative step of the study, a key focus in the interview guide is the nature of the bankcustomer relationship and the reasons behind the passage from "active" behaviour on the Internet (such as e-commerce) to "active" behaviour in the usage of online banking.

The perceived innovation attributes (functionality, accessibility, user-friendliness and cost-savings) are found to be important determinants of consumers' adoption decisions. The availability of sufficient information from electronic channels can also significantly affect the acceptance of direct banking. However, according to the semi-structured interviews with customers, direct banks appear to be an "evolving" model with strong growth potential. Indeed, the "emotional" component of the relationship with customers (e.g. the feeling of being protected and cared for in managing their financial resources) is still evolving.

Issues arising from qualitative Stage 1 -combined with the literature findings -are used as a basis for designing the survey questions of quantitative Stage 2.

The results from the second stage -which represent the focus of this paper -are derived from a survey questionnaire distributed in May 2017 to a stratified sample of Italian bank customers extracted from a panel of 4,000 households (around 10,000 individuals), perfectly representative of the Italian population, who are periodically interviewed about their purchases of banking services by the international market research institute involved in the study. Panellists were thus classified according to the bank (traditional or direct 2 ) where their main current account is held and -if they are account holders at traditional banks -based on their usage (or not) of Internet banking. This classification method -founded on information available a priori with respect to individuals' engagement in the interviews -potentially reduces bias due to customer self-selection (Xue et al., 2011) .

The total sample is rendered representative of the Italian banked population 3 , through a weighting process. In fact, since we are particularly interested in the usage of electronic channels, our sampling placed an emphasis -compared to the universe of reference -on customers of direct banks (i.e. account holders at direct banks) and customers of traditional banks who use branches but also the Internet for their banking transactions. This categorisation is clearly functional to our analysis, which overcomes the classical concept of adoption as a binary variable, i.e. consumers either have not adopted or have adopted the innovation under study (Yiu et al., 2007) . In contrast, we test our research hypotheses by breaking down the customers into three target groups: those who have not adopted online channels (here called the "non-adopters" or "TB customers"); those who have adopted them but only occasionally (the "users" or IB customers); and those who only operate through direct banking (the "devotees" or DB customers).

Within the weighted sample, DB customers represent about 30% of our group of respondents, while the remaining 70% includes IB and TB customers, half and half.

The target of "non-adopters" is more mature in terms of age and has a lower percentage of graduates compared to the other groups considered. These characteristics fit the profile of the non-users of electronic banking as described in previous studies (Karjaluoto et al., 2002) . Moreover, respondents are predominantly male -as well as for the other survey groups -given that gender difference in banking activities is a strong characteristic of Italian households (CONSOB, 2017) . With regard to employment, more self-employed workers (with higher earnings) and a lower share of employees were identified among the segment under consideration. This finding seems to contradict the empirical findings collected so far, according to which customers in high-level employment are more likely to use online banking (Jayawardhena & Foley, 2000) . It is also noteworthy that, for the target of "non-adopters", the frequency of use of the Internet (not for banking activities) is equal to that of the other groups interviewed (IB and TB customers). Furthermore, the "users" and the "devotees" are similar in terms of socio-demographic profile: greatest concentration between 18 and 44 years, and higher level of education; they are also long-term users of online services (more than 35% of them have been using web banking services for more than five years). Both targets use online banking for routine operations (acquisition of information and payment orders), while advanced usage (orders on loans and investments) connotes the "devotees" much more than the "users" (31% versus 10% of the respective group). These features are consistent with Lassar et al. (2005) , who show that it is more likely for individuals with web experience to adopt Internet banking. Table 1 summarises the principal characteristics of the sample investigated. In detail, we collected 688 interviews: 203 involve the "devotees", 243 concern the "users", and to conclude 242 involve the "non-adopters". Panellists are provided with a touch screen tablet, and this device enabled the real-time submission of a questionnaire containing 26 questions (19 of them closed-ended with a single choice) and respondents answered by simply touching the screen. Filling out the questionnaire took approximately 10 minutes.

Respondents' answers were evaluated using a different statistical methodology based on the Shapley Value (SV) regression analysis, as described in the following subparagraph.

The questionnaire consists of four sections aimed at investigating respectively: awareness and usage of direct banking; image of and propensity to use direct banks; main resistance to adoption and J o u r n a l P r e -p r o o f possible levers of development; relationship with the traditional bank and multi-channel distribution;

and a final section containing information on the socio-demographic profile of the interviewee.

In the second section of the questionnaire, participants were asked to respond to a series of statements regarding their perceptions about direct banking. These statements were created using the measures found in previous research (Juwaheer et al., 2013; Akinci et al., 2004; Howcroft et al., 2002) and in response to critical aspects that emerged from the qualitative stage of the study, such as the degree of simplicity, comfort, safety, assistance and customisation of the relationship, transparency and economy. Moreover, in this section, aspects and elements for a new direct bank model -arising from the qualitative stage -were analysed. In particular, the interest accorded to each of them was tested.

Then, their consequences in terms of image and propensity were measured.

As in previous studies, the statements were adapted to fit with the specific technology being examined (in this case, online banking; see, e.g. Chau & Hu, 2001) . Overall, five statements (with a total of 31 items inspected) were developed, specifically relating to direct banking. Responses are on a 6-point Likert scale (Preston & Colman, 2000) with anchors from strongly disagree (3) to strongly agree (8) 4 .

At the beginning of the quantitative step, we focussed on how the "current" model of direct bank (explicitly defined in the questionnaire as a bank that delivers its services primarily online, through the Internet) is perceived by the three target groups involved in the survey.

The aim, at this stage, was to provide answers to the following questions:

Can, and to what extent, different perceptions explain the choice of using the direct bank model as compared to the other models (traditional bank with or without Internet banking)?

What is the relative importance of each aspect of this choice?

In order to answer these questions, we carried out two regression analyses using binary models comparing customers with respect to the segment to which they belong (specifically the "devotees"

versus the "non-adopters", and then versus the "users").

The first regression involves a sub-sample of 445 observations belonging to the following two target groups: the 203 DB customers; and the 242 customers of traditional banks who do not use Internet banking (TB customers).

For the predictor variables, we used the responses (on a six-point scale) collected from participants with about 11 attributes designed to describe the image they have of direct banks (see Table 2 ). These items can be grouped into three macro-areas/dimensions, which emerged from the previous research and the qualitative stage of the study: cognitive constructs/utility (user-friendly; accessible; economically advantageous); psychological constructs/emotionality (secure; cold/detached; with a high level of support and assistance; with customised interaction, even remotely); and compatibility (it is suited to more complex needs and services; it is only suitable for certain types of consumers). A few attributes (e.g. transparent) are multidimensional concepts strongly affected by both cognitive and emotional components (Yousafzai et al., 2005) .

The second regression analysis, on the other hand, is based on the sub-sample of 446 observations belonging to the following two target groups: the 203 DB customers; and the 243 customers of traditional banks who also use Internet banking (IB customers). Moreover, we still consider the 11 attributes already mentioned as predictor variables.

In both cases, we employ the Shapley Value (SV) regression analysis as developed by Lipovetsky & Conklin (1998) .

Many different techniques are used in applied regression analysis to evaluate the relative importance of predictors. One particularly useful technique is a decomposition of the coefficient of multiple determination into direct, indirect and net effects associated with each variable identified (Ferber, 1964 Various techniques were elaborated for choosing and averaging among the regression models to find the best subset of variables (Lipovetsky, 2012) . However, given the power reduction caused by multicollinearity, it is difficult to be sure that the best models found by such search are really superior to the many other models that are rejected (Israeli, 2007) . Thus, we needed a decision tool that could produce clear results for the estimation of regressors, even if they are collinear and when there are many possible models by various subsets of the predictors. We found in the field of co-operative games a useful analysis and decision tool. This tool is the Shapley Value imputation.

The Shapley Value imputation satisfies the Nash equilibrium, and as such it is a method for choosing an optimal strategy under uncertainty. The Shapley Value creates a score for each player in a game that represents that player's contribution in the total value of the game. This applies to cooperative games.

In the regression, we can think of attributes as players and the total value of the game as the quality of the regression model or the R 2 value. The Shapley Value of a single attribute can be defined as: predictor variable in the model, in fact, is equivalent to the R 2 value. This means that it can be thought of as a breakdown of the total R 2 value into components associated with each predictor j.

The Shapley Values (SV j ) are nothing else but estimations via averaging by all possible linear models with different subsets of the regressors.

Instead of calculating the net effects from the ß coefficients and the correlation matrix, we can use the Shapley Value estimates as the net effects (NEi) and solve for the βs. This is a non-linear system of equations but it can be solved with any non-linear solver routine.

The result of this procedure is a new set of regression coefficients that have net effects with good properties. Indeed, they are not so volatile as regular net effects and are not prone to multicollinearity distortion (Lipovetsky & Conklin, 2001) . In brief, the Shapley Value is an alternative way of measuring the relative importance of variables in a regression model. It is similar in concept to the net effects and can be viewed as a more robust estimate of those net effects.

In the second part of the quantitative investigation, we analysed the reactions of the "prospect" group composed of the 243 IB customers (i.e. customers who, while using the Internet for their banking activities, still do so only occasionally) to the proposal of a new concept of direct bank -the "evolved" model -to which we added seven new characteristics/services -one for each field/area of potential development which emerged from the qualitative stage of the study (see Table 4 ) -which improve the "current" model along several unusual lines compared to the current offering by direct banks. Besides emphasising the advantages in terms of costs/prices, there are also improvements regarding interactions, training, consulting, protection, involvement in governance, and even the proposal to provide a limited, albeit significant, number of "light" front offices.

The aim of the analysis is to understand:

if, and to what extent, the concept of evolved direct banks can increase the propensity of the "users" towards hi-tech banking solutions;

which (service and/or image-related) elements contribute to this increased propensity, and to what extent.

In order to respond to the first question, we compared the values assigned (on a six-point scale of interest) by the IB customers (the "users") regarding their propensity to become a DB customer by opening an online account in a new evolved model rather than in a current direct bank.

To respond to the second question, we once again used the Shapley Value regression approach, but comparing customers with respect to the group of interest to which they belong (specifically the "reactive" group versus the "non-reactive" group). We define as the "reactive" group the IB customers that attributed a propensity value towards the evolved model higher than the one assigned to the current model. The remaining IB customers were allocated to the so-called "non-reactive" group.

We consider the seven new characteristics/services and the 11 traditional attributes used in previous regressions as predictor variables.

J o u r n a l P r e -p r o o f 16 The first regression -as already mentioned -was performed to investigate the decision to adopt the direct banking model by traditional bank customers who do not use the Internet for their banking transactions (the "devotees" versus the "non-adopters"). The following are the results of this first analysis: The particularly high R 2 value (0.818), besides being an indicator of the effectiveness of the model, also confirms -in combination with the limited number of attributes that account for a sizable percentage of the total variance observed -the existence of well-established, robust determinants of users' acceptance of online channels. This robustness appears to be consistent with the predominantly cautious transition towards dematerialised ways to interact with a bank experienced in many countries (Durkin et al., 2008) .

In particular, the analysis highlights that the key factor underlying the decision to switch over to direct banks by "pure" customers of traditional banks (i.e. those who access financial services without the support of any form of Internet banking) is the assessment of their user-friendliness. This provides a significant marginal contribution equal to 29.53% of the model's explained variance. This finding is not surprising, given that ease of use is generally regarded as an important quality attribute in computer-technological services and in the perceived usefulness of Internet-based retail banking (Liao security, with a smaller variance contribution of 18.22%. As clarified by Yousafzai et al. (2005) , security concerns arise from the use of an open network, where customers are afraid that their personal and financial information will become available to others via the Internet and may be used for fraudulent purposes. The need for security has been widely recognised as one of the main obstacles to the adoption of electronic banking (Aladwani, 2001) . Therefore, we did not expect it to be placed only second in our ranking, some distance behind the first consideration. Economic advantage ranks third, with a contribution of 12.46%. This finding is also in line with previous research (Ahmad & Buttle, 2002) .

It might be useful to highlight that the first two aspects alone (user-friendliness and security) account for 47.75% of the model's total explained variance. These are so-called "prejudicial" aspects, i.e. they are due more to the a priori judgements of potential customers -who have never used Internet banking -than to personal experience. This perception is obviously affected by emotional uncertainty deriving from lack of experience (Patsiotis et al., 2012) . The findings described are consistent with the thesis according to which difficulty in using devices combined with lack of experience is a major,

if not the most critical, impediment to Internet banking (Cheung & Lee, 2001) . The confirmation of this thesis would support the rationale for an educational model targeting the potential customers of direct banks, based on the core criteria of primary socialisation: relational involvement of the bank's front-end network, e.g. for face-to-face training at the branch, or at the customer's home, and a gradual approach, with the development of basic direct bank models for the first stages of the training process (e.g. simple screens -as in the case of ATMs -with limited functions and access through familiar passwords, subsequently upgrading to more complex versions once the basic skills have been mastered).

The aim of the second regression is to investigate the decision to adopt the direct banking model by the customers of traditional banks who use branches but also the Internet for their banking transactions (the "devotees" versus the "users"). This second analysis produces the following results: In this case too, the significantly high R 2 value (0.809) confirms the strong relationship between perceptions about online banking and the decision to switch over to the direct model. In this second regression, however, the marginal contributions of the different image items are more balanced and the priority aspects explaining the adoption of the direct banking model (as compared to Internet banking) are more varied. Obviously, convincing a segment (the "users") that is already familiar with electronic banking, but uses it only occasionally, requires more structured and better defined arguments.

If we classify the contributions in order of marginal value, the contribution of the capacity to offer clear and transparent conditions to explaining the propensity (or willingness) to adopt this model is 18.92%. A high degree of support and assistance with regard to products and services ranks second, with a contribution of 18.57%. This second finding is not surprising, given that the online environment does not allow customers the natural benefits of face-to-face communications to directly observe the service provider's behaviour or look into the service provider's eyes, assurance mechanisms on which humans have depended for untold ages (Citera et al., 2005) . In third place we find security, which adds a further contribution of 17.21%, for a total -with regard to these three aspects -of 54.70% of the model's total variance. Yousafzai et al. (2005) , examining the effectiveness of potential trust-building strategies for ebanking, demonstrate that structural assurance mechanisms (e.g. clarifying security policies and guarantees) had an impact on customers' trustworthiness perceptions, suggesting that banks need to use a portfolio of strategies to build the customer's trust. According to Furnell & Karweni (1999) , customers with a greater awareness of security will be more likely to use Internet-based services. The awareness can be built by giving customers clear obvious access to security policies on the websites.

Nevertheless, as reported by Koller (1988) , customers are mostly concerned with the monetary aspect of e-banking transactions, as opposed to the means through which they happen. Thus, Yousafzai et al. (2005) propose that if liability for an unauthorised transaction is removed, customers will develop a high degree of trust in the online service provider. Our findings, which reveal a distinct contribution of security and awareness of economic conditions (and also expectations, if customers are no longer liable for the risks of disputed transactions), robustly support the vision of the cited authors, as also discussed below.

If we examine the results of the two analyses together, we can find information of use for better understanding the levers for enhancing the desirability of hi-tech banking solutions.

First of all, we can highlight that -in a perspective of further development and dissemination of the direct banking model -the clientele cannot be approached in a uniform manner. The levers that could foster the initial acceptance of the online model by "non-adopters" (e.g. user-friendliness and security) are, in fact, profoundly different from those that emerged in connection with the switch-over from an early stage of adoption (typical of "users") to a retention phase (typical of "devotees"). In brief, user-friendliness and economic advantages appear to be markedly more decisive at an early stage of adoption. Surprisingly, accessibility does not appear to be a determinant factor. Security is stably relevant, although slightly more so at an early stage of adoption. Awareness of economic conditions seems to be an essential lever for keeping customers loyal to a direct bank, although it is also relevant at an early stage of adoption. Extensively, structural assurance mechanisms (e.g.

clarifying security policies and guarantees) have a robust impact on keeping customers online and loyal to electronic channels. Moreover, several previous studies have shown that compatibility is significant and positive for e-banking services (Kolodinsky et al., 2004) . However, our results alter this prediction, as compatibility is never relevant. Furthermore, the factors influencing the adoption of Internet banking appear to be consistent across different interviewee socio-demographic profiles. This J o u r n a l P r e -p r o o f statement is consistent with the significantly high values of R 2 , suggesting that the attributes considered explain much of the total variance observed. This result has precedents in the literature (Floh & Treiblmaier, 2006) . Therefore, we can see the emergence of an (online) customer lifecycle, which requires the development and dissemination of direct banking to be approached through a gradual process, featuring a number of steps, and different activities and support during each step. From this point of view, levers on two different levels can be activated. Tier one levers (or actions) respond to basic needs, such as the above-mentioned user-friendliness and security, which are more closely related to primary ontological security, consisting of the acquisition of a sort of preliminary competence for tackling the virtualisation experience of direct banks with a sense of order and continuity comparable to the current physical experience, while tier two actions are typical of customers at a more advanced stage with regard to online banking skills. These levers seem to respond to more complex needs arising in connection with the new type of banking and are primarily practical in nature, focussing first and foremost on transparent conditions. As already mentioned, the expression "transparent conditions" may be interpreted in two different ways, i.e. as a practical request (direct banks have clearer conditions than the online banking services offered by traditional banks), or as an implicit request related to a different type of bank altogether, based on a "contract" between bank and customers, in which the two parties are on more equal terms. In the latter case, the bank would be characterised by a greater degree of horizontality and the sharing of business fundamentals as well,

and not just more open external communication. This hypothesis does not yet seem to be confirmed by the research carried out and, therefore, needs to be further investigated. However, certain findings presented in the following pages would seem to strengthen the hypothesis that expectations for a different type of banking exist; the issue of "governance", according to the evolutionary concept presented and discussed in the next paragraph, refers precisely to this.

Moreover, regarding the 243 interviewees belonging to the "prospect" group (the "users"), 54% (131 individuals) expressed a higher propensity towards the evolved direct banking model compared to the current model and 46% (112 individuals) revealed a predisposition towards the new direct bank model J o u r n a l P r e -p r o o f equal to or lower than the propensity towards the current model. Therefore, the new evolved model seems able to increase the propensity of "users" towards direct banking. Accordingly, the third regression is performed to investigate which elements contribute to this increased propensity and to what extent, supporting researchers and practitioners in narrowing the playing field by finding a parsimonious set of success factors able to improve the value proposition on offer through a virtual interface. The following are the results of the analysis: (Chau & Lai, 2003) , is significant for Internet banking adoption, even if in an evolved model.

Overall, the three above-listed elements contribute no less than 76.04% of the total explained variance of the model. Therefore, the analysis highlights that several service elements, which could increase and complete the current model with regard to relationship contents (customised interaction, light branches, mature range of products), are capable of positively driving propensity towards direct banks.

The conditions which, therefore, can further drive propensity towards direct banks are sufficiently clear and -once again -can be interpreted at two different levels. One is a "practical" level of demand, featuring technical and functional solutions suited to its implementation, e.g. customised remote relations may consist, for instance, of tailor-made areas of the web portal, or even teams of staff serving individuals or small groups of customers, as is the case today in certain financial and insurance business models already operating. The other is a higher, "pact-based" level, to which the direct bank may be raised to become an institution characterised by a greater degree of horizontality, and governed with a philosophy (values, practices, emotional and symbolic content) that differs from the models hitherto established.

In any case, all these findings remind us that (customer) relationship management remains a key factor, and perhaps not just at an operational and material level; in fact, elements relating to the symbolic and value plane attributed to the bank come into play. This too is an assumption that needs to be further investigated. Consistent with this suggestion, however, "governance" (the proposal of a direct bank directly owned by its customers, according to cooperative systems and people's bank models) ranks fourth among the aspects that contribute to fuelling customer propensity (even though its contribution is much lower compared to the first three factors).

Our results suggest that migration from traditional to direct banking does not happen by itself, simply because it is logical. Considering that the upsides of direct banking are impressive, the choice could not be more obvious for banking institutions: to emphasise accessibility (in time and space), underline transparency and usability, but, above all, be ready to share cost advantages with customers by slashing the prices of services and introducing better rates for deposits. It worked, but with price sensitive, rational, financially savvy, and digitally educated individuals. They are indeed a part of the market, and, often, they belong to the more affluent and wealthy segments. Still, they are a minority, while the majority of the business is made up of individuals who have a mixed relationship with their financial affairs; in fact, the findings tells us that people who keep completely away from direct banking feel incompetent and unable to master the technology. What could drive them to shift to direct banking is user-friendliness and security; factors that, indeed, reveal that what they feel is a lack of competence, better, of basic competence necessary to enter the digital banking world. But once the great leap forward in competence is made, what really motivates people is not a faster, more efficient, direct bank (of course these factors still play a positive role); what people want is a different bank, where technology is the enabling factor that makes transparency real, delivers high-quality customer care, and guarantees security and easy personal relationships. This means that acquiring the necessary digital skills is the pre-requisite; but once people have climbed the steep cliff of competence, they move on to demand services, personal touch, care, i.e. factors that it would be difficult to deem purely logical, coldly accountable.

Therefore, direct banking has catered to the rational hemisphere of our needs: prices, convenience, transparency; it worked, but everything that can be labelled "emotional" was not even taken into consideration, limiting the potential for adoption and making the process slower than it could have been.

The suggestion we have is twofold: on one side, retargeting the communication, and on the other, reallocating some of the advantages of the direct model to a different beneficiary.

Communication policies -we have repeated several times -have always sung the rational tune: price, efficiency, accessibility. But the web does not only enable the delivery of services round the clock at J o u r n a l P r e -p r o o f lower costs; it also gives transparency and makes people able to communicate, interact and establish intimate and micro relationships. Moving on to the economics, while it is true that some direct banks are still far from reaching breakeven point (but, at present, this is an endemic problem of all retail banking), it is clear that the model is inherently more efficient than the traditional one. The decision was to share the cost benefits with customers, to make the proposition more appealing to them; but if the benefit of an increase of 25 base points in the interest rate of an average deposit is marginal, while there is a huge unanswered demand for protection, wouldn't it be more rewarding to use cost savings to build up some sort of insurance that protects clients from the possible dangers of financial markets?

And also, wouldn't using the proceeds of efficiency to rebuild trust and confidence between the bank and the client (i.e. advisory services, customer relations, tolerance vs. bureaucratic formalism) be appreciated much more than a few extra euros in your deposit account? Suggestions, hypotheses and proposals. Nothing definite, certainly, but a possible alternative.

Lastly, our survey was conducted in 2017, thus in a pre-COVID 19 era. In the spring of 2020, most countries went through long periods of lockdown, which forced individuals to make intensive use of e-commerce, social media, coworking tools, etc. Some, indeed, say that the real leadership in digital transformation of business and society has come not from CEOs, CTOs, or digital gurus but from COVID 19 itself. Banking was profoundly affected by this phenomenon, as customers were strongly driven to the use of all remote banking options; an accelerated, strong learning process was activated and a new familiarity with digital access to financial services was the result.

It is still too early to assess the long-term effect of what has happened on digital transformation, but while the role of rational and emotional drivers remains crucial, future research should investigate the impact of this exceptional forced change of customer habits on the shift from traditional to high-tech banking.

Retaining telephone banking customers at Frontier Bank

Adoption of Internet banking among sophisticated consumer segments in an advanced developing country

Online banking: a field study of drivers, development challenges, and expectations

Bricks versus Clicks: The Changing Nature of Banking in the 21st Century

Retail-Banking Distribution 2025. Up Close and Personal

Retail bank services strategy: a model of traditional, electronic, and mixed distribution choices

Information and communication technology: organisational challenges for Italian banks

Information technology acceptance by individual professionals: a model comparison approach

An Empirical Investigation of the Determinants of User Acceptance of Internet Banking

Trust in internet shopping: instrument development and validation through classical and modern approaches

An experimental study of credibility in enegotiations

Report on financial investments of Italian households. Behavioural attitudes and approaches

Internet adoption in Italian banks: An empirical investigation

Perceived usefulness, perceived ease of use, and user acceptance of information technology

Multivariate analysis of determinants of Internet banking use in European Union countries

Key influencers and inhibitors on adoption of the Internet for banking

Behavioural Study on the Digitalisation of the Marketing and Distance Selling of Retail Financial Services

Marketing Research

E-finance e e-commerce: Banche e nuovi competitors

What keeps the e-banking customer loyal? A multigroup analysis of the moderating role of consumer characteristics on e-loyalty in the financial service industry

Security implications of electronic commerce: a survey of consumers and business

The Mid-Point on a Rating Scale: Is it Desirable

Measuring extreme response style

A systematic review of Internet banking adoption

Reflection on bank-customer interactions in the new millennium

Consumer attitude and the usage and adoption of home-based banking in the United Kingdom

A Shapley-based decomposition of the R-Square of a linear regression

Changes in the banking sector -the case of Internet banking in the UK

Factors influencing the adoption of Internet banking: a case study of commercial banks in Mauritius

Factors underlying attitude formation towards online banking in Finland

Risk as a determinant of trust

The adoption of electronic banking technologies by US consumers

The Relationship between Consumer Innovativeness, Personal Characteristics, and Online Banking Adoption

Communication strategies to overcome functional and psychological resistance to Internet banking

Internet-based e-banking and consumer attitudes: an empirical study

Understanding consumers adoption of internet banking: an interpretative study in the Australian banking context

Interpretation of Shapley Value regression coefficients as approximation for coefficients derived by elasticity criterion

Analysis of regression in game theory approach

Choosing Product Line Variants: A Game Theory Approach

Factors affecting the adoption of mobile banking services

An Integrative Model of Organizational Trust

Explaining the application of IVR: lessons from retail banking

Measuring the Digital Economy: A New Perspective

Adopters and Non-Adopters of Internet Banking: a Segmentation Study

Self-efficacy and anxiety as determinants of older adults' use of Internet Banking Services. Universal Access in the Information Society

Optimal number of response categories in rating scales: reliability, validity, discriminating power, and respondent preferences

Global Consumer Insights Survey

Not So Different After All: A Cross-Discipline View of Trust

Adoption of Internet banking by Australian consumers: an empirical investigation

The influences of system usability and user satisfaction on continued Internet banking services usage intention: empirical evidence from Taiwan

Determinants of user acceptance of Internet banking: an empirical study

Balances and accounts of online banking users: a study of two US financial institutions

The Determinants and Outcomes of Internet Banking Adoption

Factors affecting the adoption of Internet Banking in Hong Kong. Implications for the banking sector

Explaining Internet Banking Behavior: Theory of Reasoned Action, Theory of Planned Behavior, or Technology Acceptance Model?

Strategies for Building and Communicating Trust in Electronic Banking: A Field Experiment

Understanding customer-specific factors underpinning Internet banking adoption