1 Introduction

Digital innovation has transformed several industries in the past decades. Leveraging digital technology, new entrants are challenging incumbent firms’ existing competencies with great speed [1], building their platform ecosystems [2], and causing radical industry level transformation [3]. Gaming, music, and book industries are proven cases of this digital disruption [4]. Many leading-edge researches indicate that the disruptive forces of digital innovation are also reshaping retailing sector [5, 6] and have enabled platform owners such as Amazon and Alibaba to expand their retail ecosystem on steroids [7]. Although the disruptive architectural features of digital innovation are means for platform entrants to challenge incumbent firms and potentially trigger retail industry transformation [8,9,10,11], innovation is not a unified phenomenon [12]. Some innovations fundamentally destroy and make established competence obsolete; others refine and improve. The foundation of a firm’s position leans upon a set of competencies or competitive ingredients of firms, such as human skills and relationships with partners. Thus, the repercussions of innovation for competition depend on “transilience”, defined as “the capacity to influence the firm’s existing competencies” [13].

The innovation literature traditionally distinguishes incremental innovations from radical innovations by identifying the degree of change associated with it [14, 15]. The former refines or reinforces existing products, processes and practices, and merely signifies marginally departing from existing practice. The latter produces fundamental and profound changes in the activities of the organization, or an industry, and renders established technical and production competencies obsolete. It is manifested by clear and significant departures from existing practices. Moreover, highly radical, competence-destroying innovations disrupt value chains and expose companies to increasing uncertain and competitive environment, resulting in the transformation of industries and revolutionary changes in the market structure [16]. Incremental innovations, on the other hand, alter the market structure marginally and does not transform the nature of competition [17]. Nowadays, the digital innovation (e.g., AI algorithms) adopted by retail platforms such as Alibaba are increasingly intermediating the exchanges between buyers and sellers and orchestrate complementary innovations to make the platform more valuable to customers [18]. This poses threats on the incumbent retailer-supplier relationship, and it is claimed that wholesalers and dealers face the risk of marginalization and disintermediation [19, 20]. Academic researchers and practitioners tend to focus on the transformative potential of digital innovation [9, 11], and a growing leading-edge research is emerging that consider the antecedents and importance of digital innovation [21] and their implications in retail platform ecosystem [7]. However, gaps remain with respect to empirically investigate whether the changes live up to its predicted transformative potential and how digital innovation in retail platform ecosystem transform the incumbent channel relationship. Little empirical research has provided nuanced explanations on the process and degree of changes, ranging from evolutionary to revolutionary, on the established retailer-supplier relationship. This paper undertook an in-depth case study and investigated whether the digital innovation nurtured by retail platform sponsors is transformative or incremental in nature. This issue is important to understand for both digitally native platform entrants and incumbent industry players.

The rest of this paper is organized as follows. We begin by discussing the characteristics of digital innovation and its implications on retail platform ecosystem. We then discuss whether it could destroy the competencies of traditional retail intermediaries. In the subsequent section, we present our research setting, data collection methods and report our preliminary findings. A theoretical discussion on these findings is then followed and based on our initial empirical findings and conceptual rationale, we propose three propositions. The last section is conclusion and direction for future research.

2 Literature review

2.1 The implications of digital innovation in the growth of retail platform ecosystems

Yoo et al. [3] define digital innovation as new orchestrations and ensemble of physical and digital components to produce new products or services [3]. This definition shows that digital innovation transforms the nature of products structure and enables the newly recombined physical components addressable, programmable, sensible, communicable, memorable, traceable, and associable. Fichman et al. [22] further conceptualize digital innovation as “a product, process or business model that is perceived as new, requires some significant changes on the part of adopters, and is embodied in or enabled by IT” [22]. This definition shows that digital innovation is an innovation outcome as well as an innovation process. It is not only about new digital architecture and digital artefacts (technology innovation and product innovation), but also a process innovation. Companies are forced to depart from existing innovation paths, routines and procedures and manage competing concerns created by digital innovation [23]. Given the above conceptualization, Nambisan et al. [21] refine the definition of digital innovation and refer to it as “the creation of (and consequent change in) market offerings, business processes or models that result from the use of digital technology” [21]. This definition is inclusive and captures a broad range of innovation outcomes such as digital platforms and ecosystems [24, 25]. These digital outcomes contain digital components at different levels (e.g. devices, operating systems and applications) and might diffuse, assimilate or adapt to different specific user context, giving rise to wakes of industry digitalization [16, 26]. We draw on the definition of digital innovation developed by Nambisan et al. [21] in this paper and emphasize that processes and outcomes (product/services) are shaping and being shaped by the other [21].

The unique disruptive properties of digital innovation such as generativity and malleability enable new types of innovation outcomes to diffuse rapidly and innovation processes difficult to predict [27]. Generativity is a remarkable disruptive feature of digital innovation, which is characterized by design flexibility allowing for creation and production of new products, structures or behaviour without increasing additional physical inputs by the designers of the digital system [26]. Through loosely coupling at all levels in a layered architecture, innovation can be executed at any level leading to cascading effects at other levels [3]. This leads digital innovation outcomes less bounded and undetermined ex ante with distributed innovation agency. Malleability is another unique feature of digital innovation, defined as the ease with which digital technologies could be reconfigured [28]. These disruptive properties of digital innovation assist firms to expand their product and service portfolios into new markets with great speed [21, 27]. The phenomenal success of iPhone and Kindle demonstrates the power of digital innovation in altering industrial structure and competitive landscape, reshaping industry boundaries and transforming the nature of competition [3].

Leveraging on the power of digital innovation (e.g., AI algorithms, digital platform and digital infrastructure), retailers are no longer appropriately described as “merchant intermediates” who purchase from suppliers and sell to consumers. They are now defined as orchestrators or conductors of two-sided or multiple-sided platforms that function as ecosystems in which value is produced, provided to customers, and then seized by the retailer and its business partners [29]. Value creation has shifted during the last decades from individual contributions by single firms to the cocreation of value in an ecosystem [2]. The concept of ecosystems has changed the view from traditional inter-firm competition to an inter-ecosystem competition [30]. In biology, ecosystem is a community of living organism and non-living components of their environment linked together through nutrient cycles and energy flows and interacting as a system [31]. Borrowed from biology, Iansiti and Levien [32] suggest that there is a shared fate of the business community as a whole and each member’s performance depends on the overall performance of the ecosystem [32]. Among ecosystem research, one particular stream is related to platform ecosystem and focusing on the interdependence between platform sponsors and their complementors [2]. Platform ecosystem considers multiple complementors organizing around a platform and form multisided markets [18]. Platform sponsors facilitate the connection between supply and demand and does not necessarily own or control physical assets and infrastructure resources [33]. This could also be viewed as a business model innovation called “platformization” manifested by coordinating the transactions between buyers and sellers on multisided platforms without any own inventories. The roles of platform sponsors include building the regulations that facilitate and govern transactions and network activities [18]. Unlike hierarchical supply chain management, platform ecosystem benefit from the decentralization of infrastructure governance and the unbundling of its use. Ecosystem participation is viewed as a particular type of loose alliance characterized by loosely coupled components [2]. Hein et al. [34] propose a clear definition for digital platform ecosystem. “A digital platform ecosystem comprises a platform owner that implements governance mechanisms to facilitate value- creating mechanisms on a digital platform between the platform owner and an ecosystem of autonomous complementors and consumers” [34]. A defining characteristics of a digital platform ecosystem are the value creation dependencies established among platform sponsors and complementors, a common set of goals shared at the ecosystem level to provide a unique value proposition that differentiate with other ecosystems [34].

In retailing sector, platform sponsors simply act as enablers and facilitators, intermediating transactions between actors without handling the entire supply and logistics chain themselves [35]. Retail platform ecosystems provide value-adding complementary services and emphasize core interactions between ecosystem members, including buyers, sellers, and third-party logistics actors [18]. For example, instead of investing heavily on building inhouse logistics capability, Alibaba Group expand its vast ecosystem by integrating third-party logistics providers for supply chain functions and orchestrate their retail ecosystem [36]. They theoretically create value by aggregating both buyers and sellers, providing long-tail products, reducing intermediaries and information asymmetry, decreasing search costs for both parties, and therefore benefit from positive network externalities (the more buyers they attract, the more suppliers will join, and vice versa) [37, 38].

Multiple ecosystems may compete for survival or dominance [39]. To contend for market share and scale to a critical mass [18], platform sponsors actively attract complementors to join into their ecosystems. Within platform ecosystem, complementors are actors co-creating complementary products or services with platform sponsors to customers and enhancing the attractiveness of the focal product or services [34]. As a result, active complementor participation and alignment that create innovative supplementary items to increase user demand are becoming increasingly crucial to a platform ecosystem’s success. The literature reveals a consensus among academics that complementors are crucial to the survival and development of the overall platform ecosystem. In platform ecosystems, complementors and their various qualities are a crucial core element that directly affects a platform's long-term success [40,41,42]. Through their innovative complementary goods, complementors significantly contribute to creating network externalities by raising the ecosystem's value and meeting the demands of diverse user structures [43]. Apart from contributing to network effects, researchers discover that complementors provide value for customers and platform sponsors by offering and enhancing knowledge of the platform ecosystem [44, 45]. This means that the more complementors join a platform, the more the data quality allows the service to be improved. Examples are platforms such as Uber, where individual drivers are complementors in the ecosystem and provide complementary services to customers, from which Uber collect data and continuously advance its algorithms of route optimization and matchmaking between drivers and customers. With the increasing success of evolving digital innovation such as big data analytics, Uber has a network of 7 million drivers, overshadowing local taxi companies. As of 2019, platform business models are used by seven of the ten most valuable publicly traded companies in terms of market capitalization, including Apple, Amazon, Alphabet, Facebook, Alibaba, Tencent, and Microsoft. These companies were able to create a sustainable platform ecosystem in which innovations are generated by complementors in the platform ecosystem rather than the platform provider itself [46,47,48].

However, the concept of multihoming implies that complementors do not necessarily have brand loyalty to platforms as they could simultaneously participate different platform ecosystem rather than concentrating on one [49]. Multi-homing is also a strategy for complementors to lower the risk related with being too dependent on a single platform ecosystem [50]. The platform sponsor in the market is arguably the one with the best digital innovation orchestrating the best ecosystem of suppliers and consumers [18]. In contrast to non-digital platform, infrastructure lean upon digital innovation to remove its dependence on location for process completion. Consequently, it facilitates the diffuse of expertise across geographical and organizational boundaries [24]. In the retailing sector, platform ecosystems are bounded by and leverage digital innovation to share digital information reciprocally with ecosystem members across a multitude of systems and devices. This is a beneficial and cyclical process of data collection and data sharing as platform sponsors rely on complementors to expand touch points and data sources. Complementors could infuse insights from big data and predictive analytics and have the potential to usher continuous customer engagement, as anticipated by research on omni-channel retailing and customer relationship management [6]. In digital realm, a purchase offers the retailer prodigious amount of multidimensional information, consisting of transactional data (e.g., composition of shopping basket), demographic data (e.g., gender), and environmental data (e.g., geo-spatial location and temperature). Retailers that can draw effective insights from digital innovations could make better predictions about customer expectations, design more appealing products and services, and facilitate buyers to purchase their products [5]. Concomitantly, however, as the examples of Alibaba and Amazon suggest, digital innovation is cultivated in a capital intensive approach and small firms could not establish the same level of computing and networking resources in a short period of time and thus they rely on the effective participation of platform ecosystem to gain access to predictive analytics [26].

In summary, previous research has tended to focus on disruptive power of digital innovation generated by platform sponsors and why it threats the existing retail business model. Disintermediation is therefore predicted in the retailing sector and is defined as “the replacement of traditional trading mechanisms with alternate trading mechanisms that can be managed either by suppliers or by new, IT-enabled intermediaries” [20]. However, the evidence from studies of the digital innovation and platform literature is inadequate to draw a conclusion that the disintermediation will be pervasive in all markets, especially where social capital is abundant in transaction process. In the next section, we briefly review non-platform practices and assess the transilience of digital innovation on the traditional retail business model, aiming to answer whether digital innovation created by retail platform sponsors could fundamentally destroy the competencies of traditional retail intermediaries.

2.2 Transformational or incremental: analysing the conditions for disintermediation

The concept of disintermediation subjects to two underlying paradigms: transaction cost perspective [51] and social embeddedness perspective [52] in traditional retailing and marketing literature. From the transaction cost theory, retail transactions aim at lowering transaction costs to the minimal. As a transaction proceed, buyers encounter four types of transaction costs, including contact costs (e.g. finding information), contracting costs (e.g. ongoing negotiation), monitoring costs (e.g. quality and deadlines check), and adaptation costs (e.g. modifications of the contract). If transaction costs are high and negotiating and formulating a contract is extremely difficult, buyers may seek assistance from intermediaries who may take advantage of economies of scale and be more efficient at negotiation than the buyer [53]. In the retailing context, by providing specialized and cost-effective distribution services, wholesalers possess inherent advantages of economies of scale and scope and occupy the “middle” space in the value delivery process between retailers and manufacturers [19]. These distribution services include inventory management, warranty support, sorting, repackaging and processing returns and refunds, which could and often be provided by other similar channel players such as dealers and agents of brand owners [19]. The market position of these wholesaling middlemen is based on effective and efficient performance of distribution services needed by retail ecosystem members and it is argued that the services provided by these intermediaries is their only reason for being [19]. From the transaction cost perspectives, if the distribution services could be replaced by other providers with lower prices or better quality, the monetary cost advantage of this displacement may stimulate the retailer to replace original channel members.

However, from the social embeddedness perspective, the existing social capital between the retailer and dealer may act as an impediment when there are new service providers coming to retailers. Social embeddedness perspective focuses on the approach that agents are embedded in social networks and draw on an agent’s social capital for obtaining transactional efficiencies [54]. It is built on the concept of social capital, as derived from Bourdieu [55] and defined by Helliwell and Putnam [56] as “the trust, norms of reciprocity and networks of civic engagement, which increase the efficiency of society by facilitating coordination between individuals”. Nahapiet and Ghoshal [57] further improve this concept from a resource-based view and define it as “the sum of the actual and potential resources embedded within, available through, and derived from the network of relationships possessed by an individual or social unit” [57]. Adler and Kwon [58] develop a framework explaining how social capital operates through providing opportunities, having motivation and possessing ability [58]. Opportunities result from location in a network structure. Networks contain information gap or structural holes, which provide business opportunities for intermediaries [54]. Motivation arises from relational dimension of social capital (shared norms and trusts). Ability comes from resources that could potentially be mobilized through their social relations. In the context of retailing sector, retailers occupy the structural hole between consumer markets and product markets, bridging customer demands with product availability. Manufacturers and distributors are motivated to forge an intimate relationship with retailers to gain network advantages [54]. These relationships develop over time with transactions as the retailers and suppliers develop trust and friendships facilitated by quality products and services. Network members accumulate favours that could be invoked as resources when needed. These intangible and relational assets accumulate and grow with the increase of its usage, which is unlike other kinds of capital that would depreciate with use. Compared with transacting with strangers, trading with acquaintances reduce risks and uncertainty since it makes the commercial exchange subject to sellers’ reputation and their social relations to other network members and therefore manifests a “credible commitment” [59]. Risks and uncertainty lie in product or service quality and sellers’ future performance quality. Sellers have advantages in exchanges and an inclination to exploit these advantages. Pre-existing social relationships with sellers act as a protection mechanism against this opportunistic exploitation. Buyers prefer to use social networks when uncertainty is high since they believe that sellers to whom they are socially tied give them better terms of trade than strangers [59]. In cohesive relations not only parties make economic and informational investments, personal investments are also performed [52] including significant sentimental attachments [60].

In summary, retailer-supplier relationships are epitomized by both impersonal market relationship and personal social relations. Transaction cost economists view the former to be market-like, and efficient since it allows self-interested actors to free from opportunistic behaviours through their flexibility to freely switch to a new transaction party [51]. Economic exchanges that depend upon social capital could lead to inefficiencies since social obligations prevent actors from pursuing trades with higher quality or more cost-effective partners. However, social capital lubricates arm’s length transactions and increase interorganizational commitments. Comparing with arm’s length transaction systems, guanxi-type systems deliver information more effectively, perform activities more flexibly and adapt more quickly to changing circumstances, thus reducing transactions costs [61]. It is unclear which paradigm dominants when digital innovation brought by retail platform sponsors such as Amazon and Alibaba begin to challenge traditional retail landscape. This study aims to investigate the transformative potential of digital innovation in retail platform ecosystem and exam whether it redefines the nature of competition and forces incumbent industry players to confront a dire fate.

3 Methods

An in-depth case study was conducted on Alibaba Lingshoutong (“LST”). We selected this platform company as an extreme case for intensive study, a prototypical or paradigmatic case that is particularly useful for exploring and analysing the phenomenon of interest [62]. Initiated by Alibaba group in 2014, LST is a business-to-business digital sourcing and distribution platform, leveraging its digital infrastructure and intermediating exchanges between brand owners and independent and unfranchised retailers. LST acts as the platform ecosystem sponsor, enabling and facilitating interactions between platform ecosystem complementors including retailers, consumers, producers, and third-party logistics actors. To contend for market share and compete with other rival retail platform ecosystems, LST is now subsidizing millions of mom-and-pop stores owners to install its procurement app and digital devices across townships and villages in China.

Retailers range in size and ownership from small and independent retailers (e.g., mom-and-pop stores) to department stores and large chains such as Walmart, Tesco, and 7-Eleven, to e-commerce giant Amazon and Alibaba. Mom-and-pop stores are those sole traders, private and independent family-owned located small stores that provide retail services and life essentials to customers living in the neighborhood nearby. In India, they are known as kirana, in Indonesia as warung, in Vietnam as cua hang, in China as xiao maibu, in the Philippines as sari sari, and in Malaysia as kedai runcit. These stores are often run by local residents who have lived in the neighbourhood for a long time, know each of their consumers’ names and favourite beer brands, and are willing to provide door-to-door delivery service with a phone call. According to the research data, 6 million mom-and-pop stores function as the last-mile retailers for customers, selling 40% of the Fast-Moving Consumer Goods in China.

These mom-and-pop stores used to source inventories from multiple levels of wholesaler and dealers. LST enables the mom-and-pop store to become a digitally connected store for brand owners, which might weaken the role of multiple layers of traditional retail intermediaries. Mom-and-pop storekeepers are now scrambled by both digitally enabled retail platform ecosystem and traditional retail supply chain. Multiple interviews on these ecosystem complementors were conducted to qualitatively understand the transilience of digital innovation. All interviews were conducted in China, where social capital is abundant and guanxi-type systems are pervasive and dominate business activities [63].

The advantage of case study is its explanatory power in understanding both process and outcome of a phenomenon through complete observation, analysis and reconstruction of the cases under investigation [64]. Case study acts as a useful tool to investigate the behavioural context and sophistication of real-life business situations through the human agents’ perspective, which might not be captured through experiments and surveys. Rather than reducing the complexities of realties, case study narratives construct realties [65]. In particular, for research on digital innovation, the value of such socio-cognitive sense-making “narratives” increase as Nambisan et al. [21] indicate that innovation agency become diffused and less predefined ex ante and this makes shared cognition become vital component to comprehensively understand digital innovation management [21]. Nambisan et al. [21] further suggest that such narratives facilitate “meaning making”, that is, innovators “plot sets of social and material elements from the past, present, and future into a comprehensible narrative” [21].

3.1 Data collection

Data sources include archives, field observation and semi-structured interviews. Through gaining access to both retrospective and real-time accounts by those agents who have experienced or are experiencing the phenomenon of theoretical interest, this is ‘‘research as engagement’’ [66]. The data collection activities for our case study of LST started in September 2018 and extended into the winter of 2019. Five data sources were incorporated in our case study: (1) semi-structured interviews with 10 mom-and-pop store owners who have more than 5 years of retail work experience and are now actively using digital devices and applications of LST to carry out retail activities in 10 different province in China (2) field observation on 10 mom-and-pop stores in rural and semi-rural area in China (3) interviews with 2 industry experts with 20 years of wholesale and retail work experience (4) informal follow-up dialogues with store owners via short messages (5) archival materials (including Alibaba annual report, LST official websites, press articles, case company press releases, industry conference presentations, analyst reports and publicly available interviews with LST senior executives from 2014 to 2018). We took several steps to ensure the validity of our data. Firstly, interviewing techniques such as non-directive questioning was used, which increased the possibility to yield accurate information.

We recruited 12 interview respondents using purposive, snowball sampling [67]. We started by identifying and subsequently interviewing two mom-and-pop storeowners in Hangzhou, where the headquarter of Alibaba LST is located. They are complementors of LST who have direct user experience with digital devices and applications of LST. They also know traditional retail operation routines and business models very well, as they opened up their own retail stores before the birth of LST for more than ten years. Following the interviews., they provided us with industry participants who were identified as having adequate knowledge of LST and retail operations. As it is the research objective of this paper to identify whether the digital innovation processes and outcomes in retail platform ecosystems are transformational or incremental, it is important that we collect data from complementors who are also traditional retail network participants possessing adequate experience and knowledge that can help us in our exploration.

Non-directive questions are questions in view of an interviewee's reactions, which are examined further and continues like a cordial discussion and will more often than not be more unassuming, open ended, conversational, casual, free streaming and adaptable, which is suitable for researchers to use to answer exploratory question and understand an underreached phenomenon. This interview method is suggested by Unwin and Unwin [68], which allows freedom and flexibility to respond. Respondents are encouraged to feel ‘free’ to relate their experiences and reveal their attitudes and opinions as they see fit, with as little direction as possible from the interviewer. All interviews were conducted at the respondents’ stores, a familiar and relaxing surrounding, with the open but semi-structured interviews, lent the interviewees a reliable atmosphere, with the purpose of encouraging them to respond openly and honestly. Non-directive questions are non-judgmental, as we do not want our respondents to feel pressure and express in accordance with the researcher’s value orientation. This implies interviews that tries not to use leading questions, those that demonstrate a specific response as ‘favored’ by the questioner and utilizing undivided attention to draw out respondents’ mentalities and sentiments. In this sense, the researchers endeavour to permit respondents themselves to set the terms and boundaries of the conversation [69]. During the interview with mom-and-pop storeowners, we never reveal our own judgements about the value provided of LST and traditional middlemen and we always explain that there is no scoring of answers to our respondents. Example questions are “How do you decide what product to buy and which brand to buy?” “From whom you seek for recommendations and advice? Who often helps you make these decisions? Do these recommendation work?”.

Secondly, during the interviews, anonymity encouraged respondents to speak with candour [70]. In our exploration, we prepared interview protocols and several predetermined questions before we conduct field visit to each mom-and-pop store. However, after each interview, researchers often use informal follow-up dialogue to dig deeper into the interviewees’ responses, a qualitative research method suggested by Seidman and Creswell and Poth [71, 72]. This is on the grounds that it is generally more suitable to open the interview with a couple “simple” inquiries to make the interviewee agreeable and to acclimate him/her with the subject of the meeting. A couple of model questions are: “If it’s not too much trouble, tell me, how long have you been opening this store here?” What kind of challenges do you confront while running your little store? These questions are easy to answer to our respondents and based on their answers, we improvise the follow-up questions to probe the core competency of LST and traditional middlemen: “How do you overcome those difficulties?” Are you satisfied with the services provided by LST or traditional middlemen?” “How does the LST or traditional middlemen help you overcome those difficulties?” Further into the interview, questions like “In your opinion, what differentiates LST from traditional middlemen?” or How have you experienced the changing workload in your current store after using LST?” are bound to be addressed thoughtfully instead of on the off chance that they were suggested as the main conversation starter of the meeting. An inquiry like “Is there much else you might want to add?” frequently become a reasonable shutting question. In addition, we selectively compared field notes and interview transcripts on mom-and-pop store owners with insights from many media conversations with LST senior managers. Overall, the combination of these different data sources gave us a robust understanding of how LST as a retail platform works, how LST as a digital innovation works and how ecosystem complementors react to LST.

3.2 Data analysis

In qualitative research there is an overlap between data collection and data analysis, which is described as concurrent activities [73]. Data analysis approach suggested by Corbin and Strauss [74] is used during this research [65]. Before interviews, the phenomenon is conceptualized by researchers through reading “non-technical literature and during this process researchers’ sensitivity towards the unique aspects of the phenomenon under study have be enhanced [75]. The initial conceptualization and categorization of secondary data online could help researchers to formulate constructive questions for interviews later on. Then open coding was conducted on first-order data, with the aim of labelling newly discovered phenomenon from the data and facilitate researchers to establish uniformity and varying conditions during the process of identifying incidents and naming them into different nodes. The second step is axial coding, which is establishing the logical relationship between the main categories and their subcategories. Nodes generated by open coding were compared, merged and abstracted into high level. This process enables researchers to identify variations in concepts and facilitates researchers to find pattern. The final step is selective coding, which is to integrate and refine categories generated from axial coding and find the aggregate dimensions, which are mutually exclusive and exhaustive. First and second order data were used to verify the relationship between core dimensions and its subcores until theory saturation has been reached. During this stage, a hybrid approach of inductive and deductive coding process was used, which combined data-driven coding process (codes inductively generated from data) suggested by Boyatzis [76], and theory-driven coding process codes derived from existing literature framework) suggested by Crabtree et al. [77]. Based on literature review on the concept of digital innovation, platform ecosystem and social capital, code manual which contains definition and description of template codes was established. Then the 1st and 2nd order data were compared with template codes in the framework suggested by existing literature [2, 21] with the aim of identifying theory contribution. The whole iterative and reflexive process is used to delineate the “story line” describing the phenomenon sorted out and explaining what is going on [74]. With the focus of nascent concepts that do not have enough explanations in the extant literature, data structure was built.

Two researchers analyzed the transcripts of interviews to identify patterns and themes. Qualitative data from 12 respondents was analyzed using open coding and then grouped into themes based on their similarity. The process of qualitative research was summarized in in Table 1 (Page 17) and Table 2 (Page 20). The data collection continued until a pattern of similarity in the responses was observed, or the 'point of saturation' was found. After conducting 54 interviews, our codes reached the point of saturation where the themes and pattern derived from each response/code were similar. To confirm our findings, six more interviews were carried out. The data analysis and coding method follow the guideline given by [78]. Relevant terms and phrases were found during the initial reading of the transcripts, which were then open coded before the analysis began. We updated the open codes every five responses. We checked the existing literature after each level of coding to make sure our findings were consistent with it. In order to connect open codes to categories and topics, an axial coding technique was developed [68, 79]. The categories were established by arranging the non-hierarchical groupings of the codes used to explain particular words or phrases. A thematic map was created when the categories were further divided into themes based on how similar they were. The overview of the codes, categories, and themes that emerged from the data is shown in Table 2 Coding structure. Next section gives a brief description of our case company and some preliminary observations and findings.

Table 1 Data analysis
Full size table
Table 2 Coding structure
Full size table

3.3 Case study: Alibaba Lingshoutong (“LST”)

Initiated by Alibaba Group in 2014, the LST retail platform is a B2B digital retail sourcing and distribution platform that has been adopted by more than 500,000 neighbourhood mom-and-pop stores across townships and villages in China. The LST enlists mom-and-pop stores to tap into Alibaba vast ecosystem and become a “smart store”. It allows for all the invoicing, procurement, inventory and sales data in the store to be digitally connected and the query could be completed with one click by shopkeepers. They could order and source form a broad selection of brands and products at wholesale price online by using the LST mobile app. A free smart POS system with Alipay e-payment terminal (for data collection) is installed in the store at the checkout counter once the store has ordered a certain amount of goods from LST. With every item or service purchased electronically, retailers could gain a clear depiction and analysis of nearby demographics. Their tech can also be used for merchandising, inventory management and logistics to enhance their business operations. An inventory checking and reporting system is in operation to give real-time advice on what to source, when to source and how to display products assortments in their stores based on big data analytics of LST. After online ordering, goods are delivered to stores by third-party logistics providers within certain days. In addition, these stores could have a deeper level of cooperation with LST from sourcing to tapping into other complementary services of Alibaba retail platform ecosystem. For example, these stores could become searchable online and become an order-and-delivery stations for Tmall e-commerce (a business-to-consumer online retailer operated by Alibaba Group). In summary, LST provides mom-and-pop store owners with digital devices and applications and change their daily procurement and inventory management routines, which is both a product and process innovation for storeowners. Figures 1, 2 shows the main differences of business logic operated by LST compared with traditional retail model. The next section presents our preliminary findings from interviews and field observations.

Fig. 1
figure 1

Source: authors’ research

Traditional Retail Models Versus LST Business Logic.

Full size image
Fig. 2
figure 2

Source: authors’ research

Multiple actors within LST platform ecosystem.

Full size image

4 Findings and discussion

Across the interviews with 12 respondents who are the complementors of the LST platform ecosystem, they mentioned that LST retail platform created values for complementors in terms of inventory management, intelligent product recommendation on sourcing, provided shared services across different platforms, and enhancing breadth of assortments, but not to the same extent as from traditional middleman (wholesalers and dealers) in terms of credit purchase, after-sale service flexibility and relationship maintenance. Illustrative quotes are provided below, illuminating the impacts of digital innovation brought by the retail platform LST. This outcome is more indicative of a future for digital innovation in the retail platform ecosystem reflecting evolutionary changes rather than revolutionary changes, as one of our interviewed mom-and-pop storeowners elaborated:

There is no substitute between LST and traditional wholesalers, and one channel or platform can't dominate. At present, they are complementary to each other. Tradition has unique edges, LST has its own advantages, and traditional dealers will also undergo corresponding evolution. Therefore, if you are talking about one day, LST completely replaces traditional dealers and wholesale departments, which I think is unrealistic. This market should be in a state of blooming flowers.

Our research found that social capital possessed by traditional middlemen and retailers tends to act as a protection mechanism for traditional intermediaries to defend their market position in the retailing industry. Traditional intermediaries possess unique advantages in dealing with flexible and non-routinized transactions, such as after-sales refunding and exchanging. While LST helps standardize, automate, and customize routinized B2B transactions such as sourcing, inventory management, marketing, and product recommendation, leveraging its cross-platform data analytics advantage. The sections below discuss our findings about the unique edges of the digital innovation of LST and the superiority of traditional intermediaries.

4.1 Digital innovation of LST

4.1.1 Automation

Automation is defined as any activity or process that runs automatically without active human input or control. In our case, the B2B digital platform LST create value for complementors through automation in terms of the inventory management, brand marketing and procurement process. Storeowners who adopt LST’s Intelligent POS Terminal in the store and LST in their mobile phones could easily know when to source, what category to source and at which level of price, without driving to wholesale markets and collecting market data in person. They received automated reminders, in-stock alerts, chatbots, providing them with real-time brand information and responses.

4.1.2 Individualization

Individualization is the modification and personalization of marketing efforts based on a person's current requirements and preferences. Through collecting ecosystem-wide data and leveraging on our big data analytics, LST could continuously deliver personalized buying recommendations to each storeowner, differentiating themselves in terms of the product assortments and satisfying the customers’ future demand. as two of our interviewed mom-and-pop storeowners elaborated:

LST predicted that red masks would sell well as the Chinese Lunar New Year was coming, so they informed us to stock red masks in advance. Happily, the sales results were as expected; they sold very well and made me earn extra 5000 yuan during that month.

LST can provide assortment guidance based on Alibaba's big data capabilities. For example, LST once suggested a store to sell dog food because it found that the neighborhood owns significantly more dogs than average based on consumers’ online purchase behavior on Taobao and Tmall.

4.1.3 Cross-fertilization of data

LST Retail platform leverages its digital innovation to diversify horizontally and provides a comprehensive ecosystem of value-adding complementary services (e.g., Gawer, 2009). The ecosystem of Alibaba platforms involves core commerce, cloud computing, finance and payment, travel businesses, and on-demand video and music streaming services.

  1. (1)

    Shared services across platforms Overall, by tapping into LST retail platform ecosystem, complementors could have the potential to enhance their business scopes. One of the mom-and-pop storeowners elaborated:

    Every time when I click the LST application to source food and beverage, I feel like I am part of Alibaba cohorts and the future of my store has more possibilities. The homepage is updated basically every time, popping up many service interfaces. Drawing on the functions provided by this application, my store might be a travel agency, a delivery post, a high-tech store in the future, thus have more possibilities to increase passenger flow and business types. Now, an obvious improvement is that my store is searchable on Ele.me (a Chinese takeaway app acquired by Alibaba) and customers can access my store through the digital channel.”

  2. (2)

    Breadth of assortments Digital innovation also contributes to a possible extension of offerings. [80] attribute value creation of online businesses to complementarities, which include possible bundling by expanding the selection of products and services provided and abundant opportunities to combine and recombine these “online” and “offline”. Leveraging this, the retail platform aggregates both buyers and sellers, provides long-tail products and decreases search costs for both parties. One of storeowner explained:

    Some food categories could only be sourced from LST such as imported goods and internet-famous snack food. These categories are rarely procured by offline dealers in my district. But these new-fangled products are typically characterized by high gross profit. I usually source them from LST as an auxiliary channel to refresh the assortments within my store.

4.2 The superiority of traditional middlemen

However, traditional intermediaries enjoy an inherent relational advantage over third-party logistics providers affiliated with retail platform ecosystem for performing distribution services in the neighbourhood communities. Moreover, as frontline workers providing both brand promotion and distribution services through which they obtain commissions, these traditional middlemen have the potential to focus intensely on distribution problems and opportunities that help them to be more specialised, imaginative and innovative in solving emergent problems for business customers than other ecosystem members. They accumulate hard-to-break social capital with retailers through daily relationship maintenance activities, lubricating arm’s length transactions, increasing the trust and interorganizational commitments, preventing them from being replaced by digital innovation of LST.

4.2.1 After-sales service flexibility

Two of the mom-and-pop storeowners elaborated:

The wholesaler outlet is located in my district, very close to me. It has been open for many years. After delivering the goods, sometimes they help me sort out the shelves and check whether the goods are out of date. When the goods expire and are not sold, they help me take the goods off the shelf in time. We have a good relationship, like friends.

LST focuses on attracting more complementors to join into its platform ecosystem and amassing a large share of users to sharpen its data analytics capability and digital marketing capacity, while leaving third-party logistics providers to handle the majority of actual transaction of goods with consumers on the platform. Although platform-mediated business models are less capital intensive compared with managing inhouse supply chains, loose alliance and governance within the retail platform ecosystem could not guarantee that services provided by third-party logistics are sensitive and personal. Two of the mom-and-pop storeowners explained:

Deliverymen are hired by third-party logistics providers. They are not hired by Alibaba LST. Every time I see different guys coming to our stores. Some of them are not familiar with our layout and we do not have renqing and ganqing with them. Very numb and like moving machines, putting away the goods and leaving, they are not warm to us.

When they (traditional dealers) come to my store, they always help me relocate boxes of very heavy drinks neatly in the store. They could always find a suitable place to lay them out. They are very familiar with the layout in my store.

4.2.2 Relationship maintenance

In the literature, guanxi is construed as a kind of social capital that created by interconnected social network and maintained by reciprocal gift and favour exchanges between people [63]. Renqing, xinren, and ganqing are three dimensions of the construct of Guanxi in empirical studies on guanxi [81]. Qi [82] suggests that renqing represents a human imperative to give sympathy and timely assistance to friends and acquaintances when they are in trouble [82]. A person who is versed in renqing is well equipped with empathy. In particular, if an individual shows abundant empathy and are ready to help those that are in need, this person is viewed to be morally right. Renqing can be evoked or requested in the future due to intimacy. While renqing does not request for instant returns, renqing obligations are not free lunch without price. They must be repaid somehow and somewhere in due course, otherwise the person once granting renqing will terminate the relationship and the person once accepting renqing might be morally censured [37]. Guanxi gradually develops when renqing accumulates. The second dimension xinren is a Chinese word, representing trust. It is gradually developed through repeated favour-exchange business activities. The higher level of trust between transaction parties, the better the guanxi are [81]. The third dimension ganqing represents affection, emotion and sentiment, which is used to measure the quality of a relationship between individuals. Undertaking social interactions is proved to be an approach to increase ganqing, for example having dinner together, participating in one’s family and business events, assuming that both sides have enjoyed the interaction. The better the ganqing, the closer the guanxi [81].

In our case, by exchanging reciprocal gift and favour, renqing and ganqing is enhanced between retailers and traditional middlemen, which increase the cohesiveness and endurance of their business relationship and therefore increase retailers’ loyalty to source from them.

I think the dealer treat me as a loyal customer as well as an old friend. He always gives me some gifts during festivals. Last new year I received two bottles of aged wine and I felt quite delighted. This gave me mianzi (a Chinese word representing fame) in this neighbourhood. This renqing is very different with LST, who treat us just like their normal commercial customers.

In the absence of human capital, the significance social capital is highlighted, which could compensate and act as a substitute for other resources if “connections” are superior. It could also complement other types of capital by reducing transaction costs and improve the efficiency of economic capital [58]. In particular, as renqing could be evoked or requested when acquaintances confront difficulties, this increases intimacy and sentimental attachments [60] between buyers and sellers. Two of the mom-and-pop storeowners explained:

Small stores like ours reply on the consumption of acquaintances living in the community nearly to maintain business. If these acquaintances urgently need to buy something, we generally do not charge them for delivery. They do not order our assortments via Eleme application (a takeaway software affiliated with Alibaba Group) since the deliveryman charge them some fees to pick up the order. Therefore, either my wife or I am usually busy with delivering the goods to the customers in the community in person. When we are busy, we will ask the nearby dealers to help take care of our store.

In the absence of financial capital, social capital also plays an important role in improving transaction efficiencies and flexibilities. Here, it is manifested by extended account payable due to xinren, one dimension of guanxi, lubricating the transactions between buyers and sellers.

4.2.3 Account payable

The accumulated social capital from the past transaction records enables retailers to overcome temporary financial constraints and enjoy the service of zero- interest credit purchase provided by their acquainted wholesalers. One of the mom-and-pop storeowners explained:

We developed a high degree of trust and a considerable personal rapport. They give us procurement-on-credit service. Payment days could be monthly, half-monthly, and 10-day. It is a huge advantage for us compared with arm’s length transactions with big retail platforms.

To sum up, interviews on complementors confirm that digital innovation in retail platform ecosystem provides a technical solution to replace traditional intermediaries in terms of warehousing, order entry, order processing and order fulfilment to customers. Digital innovation free retailers from the constraints of geography and enable them to access to a broad selection of brand owners. This reduces search costs (a type of transaction costs). However, it does not fundamentally destroy and make obsolete established core competence of traditional retail middlemen (e.g., wholesalers and dealers). The traditional paradigm that has justified the intermediaries’ role in the distribution channel is still very much operative nowadays albeit in digital realm. Retailer-dealer relationship involves unspecified personal obligations, intrinsic rewards, and is the middle ground between pure calculation and pure love. Traditional middlemen are versed in hands-on small-scale operation and daily guanxi maintenance with retailers. This become their distinctive edges when competing with retail platform sponsors. Distribution services with significant human touch are value-adding services for retailers, especially those who lack financial capital to hire additional staff. This acts as a protection mechanism against disintermediation threat posed by platformization. The findings of this case study indicate that digital innovation in retail platform ecosystem has not been transformative, as many early scholars envisioned. The core competency has not been fundamentally disrupted by digital innovation in retail platform ecosystem and the social-economic nature of ecosystem member relationships are the impetus for technological disintermediation. Platform-mediated market leveraging digital innovation tend not to be a displacement of people-mediated market in in the business-to-business (B2B) market in the retailing sector.

Social embeddedness researchers imply that there are basically two types of relationships through which economic transactions proceed at the extremes: (1) arms-length ties manifested by short-term orientation and frequently changing ties among loose connections of companies or actors, and (2) embedded ties manifested by stable and long-term orientation [83]. Guanxi between retailers and traditional retail middlemen lie in the middle, characterized by interrelated economic and renqing-centered relationships [81]. Guanxi-type systems depend on renqing, xinren and ganqing between individuals [84]. They lubricate transactions because these transactions occur within a flexible, but relatively permanent, network. Thus, there are some preliminary empirical and conceptual rationale to propose:

Proposition 1

Adaptation and flexibility of guanxi-type system is superior to that of the retail platform coordination lean upon digital innovation.

Cluley and Brown (2015) argue that digital technologies do not lead to the individualization of customers but instead to a trend of “dividualisation” [85]. People are regarded “not as individuals with emotions, attitudes, behaviours, but as data that can be divided and reassembled through analysis”. This view ignores that the social and cultural context that enable smooth transactions, yet in struggling for increasing transaction efficiencies, most efforts in creating electronic network resources neglect and even bypass this context. Retail transactions are seen as a value flow from producers—through intermediaries—to end-users. Although intelligent machine agents are powerful in data mining and automate repetitive tasks, distribution services still need human agents to execute in order to fulfil customers’ affective needs and expectations. The transactions between retail ecosystem members are not only commercial but also social, including exchange of favours, face (mianzi) and social status. Merely relying on self-organizing and intelligent machine agents to execute business process is not adequate to satisfy the concerns of ecosystem members. Through an intense focus on providing superior sensitive and personal distribution service, wholesalers as human agents enjoy an inherent advantage in terms of intimacy and sentimental attachments over other channel participants. Consequently, there are some preliminary empirical and conceptual rationale to propose:

Proposition 2

The development of retail platform ecosystem is not a generative process, but requires careful attention by human agents to sustain over time.

Social embeddedness researchers have found disparate advantages of relying on social capital during critical economic exchanges. These include greater emphasis on joint problem solving and an increased possibility that new exchanges will sustain within an existing relationship rather than be directed towards strangers [52, 86]. Experimental results also show that when sentiments are positive (e.g. consisting of trust and commitment), current suppliers are unlikely to be displaced even when alternative new partners enter the market with better deals [87]. Thus, economic and relational advantages of social capital create disincentives for retailers to replace distributors who provide them with personal and day-to-day services in need of a human touch. All the discussions aforementioned provide a theoretical and conceptual rationale to propose:

Proposition 3

Digital innovation in retail platform ecosystem develop incrementally where social capital is abundant.

Although the threat of digital disintermediation has been claimed for many years [88, 89] due to the information-related advantages for manufacturers resulting from digital networks and the potential margin increase resulting from replacing trade intermediaries [90], our research proves that social capital is a protection mechanism that prevents trade intermediaries from being obsolete. Digital innovation will not be able to serve end customers as effectively as small retailers since the latter have a considerable advantage in terms of human relations empathy and distribution flexibility. But digital innovation could empower the manufacturers to have comprehensive data on the purchasing and consumption habits of their end consumers, which normally could only be collected and processed by traditional wholesalers and then passed on to upstream manufacturers with a time lag, and there was an information asymmetry between the manufacturer and the consumer. The manufacturers can use these end customer data to tap new market potential through targeted design and improvements of products, and one-to-one marketing, which is consistent with the extant research on technology-enabled omnichannel retailing [91].

Our research concludes that the development of digital innovation in retail platform ecosystems is a gradual and incremental process rather than an abrupt and comprehensive change since the digital innovation-mediated platform ecosystem could not fully replace the traditional social capital-mediated traditional retail networks. In another word, digital innovation seems could only complement, rather than replace the function of social capital accumulated in the established human distribution network so far. By utilizing the advantages of social capital in handling the flexibility and complexity of distribution processes, traditional wholesalers can try to avert disintermediation. Our research findings respond to the call for searching coping strategies to navigate the chaos of digital disruption [92].

5 Conclusion and future research

Academic researchers and practitioners tend to focus on the exploratory radical nature of digital innovation, leaving exploitative incremental nature of innovation in the backstage. Little empirical evidence has captured either the magnitude of digital disruption or how incumbents are reacting on a broad scale. This case study shows that the advent of digital innovation in retail platform ecosystem does not fundamentally change the position of traditional intermediaries and disrupt their core competencies. This study has illustrated the oppositional nature of digital innovation built from a transaction cost rationality and existing practices based upon social capital. The evidence from studies of the digital innovation and platform literature is inadequate to draw a conclusion that the disintermediation will be pervasive in all markets, especially where social capital is abundant in the transaction process. In the future, research could approach social capital and digital innovation as a duality [93] and see if there are any opportunities for synergy between these two. What types of synergies exist and how might they be enhanced through ecosystem design? New sociotechnical theories are needed to theorize how digital innovation and social capital collectively integrate supply chain processes and simultaneously pursue both standardized and humanized as well as offering stable and flexible service delivery process to retail ecosystem members.

The limitation to our research is related to the specificities of the case itself; the focus on Chinese respondents does not consider cultural aspects that could have an impact on their responses. We call for future research to replicate this study in other countries and regions for more robust theorizing, taking into account the different preferences for social capital from consumers from different age groups, cultural backgrounds, and countries. A broader range of geographic scope would enhance the robustness of the retail platform ecosystem research and potentially reveal important relationships, or other region-specific and culture-specific dynamics that have thus far been overlooked. We also call for future research to conduct comparative study between pre-epidemic and post-epidemic retail platform ecosystems and check the robustness of our findings about the established retailer-supplier relationship. Although the COVID-19 epidemic accelerated the digital transformation of the retailing industry and forced many customers to order groceries online [92, 94], the pandemic also triggered the increase of the social capital as a public good [95], especially the neighbourhood social capital derived from the social commerce and offline exchange of groceries between people living in the same compound [96, 97]. Therefore, we posit that our findings on social capital advantages in dealing with distribution flexibility and human empathy might be still very relevant in the post-epidemic era. We call for future research to investigate how digital innovation in retail platform ecosystems replace, complement or reinforce the established and the emerging social capital embedded in retailer-supplier relationship in the post-epidemic era.