Investigating the on-demand service characteristics: an empirical study

2.1 Approach literature review

The focused literature review is based on the methodology of Tranfield et al. (2003). In total, 11 databases were selected to cover most scientific publications in the field of business and related areas as economics, IT and engineering: Thompson Reuters Web of Science, Academy of Management, Elsevier Science Direct, Taylor and Francis online, EBSCOhost Academic Search Premier, EBSCOhost Business Source Premier, Wiley online Library, InderScience Publishers, Informs, Sage, and Emerald Insight. The following search string was used to search the title, abstract, and key words to find articles on on-demand services: “on-demand service” OR “on-demand service” OR “service on-demand” OR “service on-demand.” No limitations were set on publication year or literature type (e.g. papers, book chapters, conference proceedings, and working papers). This resulted in 441 unique papers.

All papers were individually assessed by reading the abstract and performing a quick scan of the search terms throughout the paper to check whether business or management aspects of on-demand services where discussed. In case of doubts, papers were included. In total, 47 papers were selected. The reduction was mainly attributable to many software engineering papers on video on-demand services focusing purely on IT issues without providing insights into the on-demand concept. Then, a full paper review was performed to assess whether the papers provided insights into on-demand service characteristics. Ultimately, ten papers were selected (Table I).

Lastly, ten remaining papers were analyzed following an inductive approach (Neuendorf, 2002; Tähtinen and Havila, 2018). Because no single paper specifically studied on-demand services’ characteristics, this open coding approach was used to derive characteristics of the on-demand services concept without pre-existing themes. All papers were independently coded by two researchers (Tranfield et al., 2003). Subsequently, both assessors’ analyses were compared and discussed by all four researchers to agree on the characteristics derived.

2.2 The on-demand service literature

The literature mentioning the term “on-demand services” has been increasing. Specific forms of on-demand services, such as video on-demand (Kalvenes and Keon, 2008), cloud computing and SaaS applications (Hou et al., 2018; Ma and Seidmann, 2015) and, more recently, ride-hailing services (Alemi et al., 2018), have received specific attention in the literature. Further, on-demand services are often linked to access-based services (Lawson et al., 2016; Schaefers et al., 2016) and the sharing economy (Fehrer et al., 2018; Kumar et al., 2018). In some cases, there is indeed a clear link between these concepts, but also an important difference: sharing resources and offering access to goods does not automatically mean these are supplied instantaneously on demand. On-demand provisioning is also often related to multi-sided platforms (Andreassen et al., 2018; Hagiu and Wright, 2015; Parker et al., 2016). Such platforms are frequently used by service providers to collect available resources for on-demand supply. At present, these platforms are frequently studied, focusing on topics such as pricing and capacity management (Bai et al., 2018; Taylor, 2018). However, again, it is important to note that multi-sided platforms are not necessarily on-demand.

Studies on the characteristics of on-demand services are lacking. Most papers mentioning on-demand services do not focus on conceptualizing and defining such services. Some, however, describe exemplary on-demand services to clarify the service type studied (e.g. Alemi et al., 2018; Bai et al., 2018; Ng et al., 1999; Taylor, 2018). Others provide some insights into the on-demand service concept. These papers are discussed below and an overview of the literature review’s findings is presented in Table I, which shows that present literature provides no consensus on the attributes of on-demand services.

2.3 Characteristics of on-demand service offers

Weinman (2012) discusses the economic value of on-demand services for the customer and argues that “on-demand” implies that the customer can be allocated the “right quantity of resources at the right time for the right amount of time at any given time” and where only the actual usage of resources is priced. Chen and Wu (2013) also discuss “on-demand” from the adopting firm’s perspective to describe the procurement of (mostly IT) externally owned resources with usage-based pricing. They argue that on-demand services offer customers direct and unlimited access to resources as they would have through owning them, but with different cost structures, which change from mostly fixed costs to variable costs. Similarly, Künsemöller and Karl (2014) argue that on-demand computing provides capacity for processing and storage similar to a physical server owned by the customer, but that fees only apply when the capacity is used, which corresponds with the usage-based pricing argued by Chen and Wu (2013) and Weinman (2012). Although these three papers address customer implications, they hardly address what constitutes an on-demand service.

Taylor’s (2018) modeling study on on-demand platform pricing argues that on-demand services supply immediately when customers experience a need. Ng et al. (1999) similarly state that on-demand services such as tow truck, lift maintenance, and emergency services differentiate themselves through short waiting times. Slightly different, Bai et al. (2018) state that on-demand service platforms offer time-sensitive services anywhere and anytime. Weinman (2012) further argues that on-demand services should react responsively, but that acceptable response times vary by service. To illustrate, 24 h is acceptable for physical book delivery, while 24 s is for eBook delivery. Lastly, Weinman (2012), argues that if cloud services can react instantaneously to changing demands and with the right amount of resources requested, perfect capacity can be offered to customers. Responsiveness or immediate supply is not mentioned by Chen and Wu (2013), Künsemöller and Karl (2014), Ma and Seidmann (2015), Weinman (2011), or Yao et al. (2005), the other papers covered in this literature review.

This responsiveness or immediate supply requires on-demand services to be continuously available. High availability for on-demand services or availability on-demand anywhere and anytime is mentioned by Bai et al. (2018), Bratianu (2018), Ng et al. (1999), Weinman (2011, 2012) and Yao et al. (2005), making this a key aspect of on-demand services. Bratianu (2018) argues in the case of access-based services in the sharing economy that a shift in the dominant business model is imminent: all consumer goods will be available as a service and all consumer services will be available on-demand anywhere and anytime. Ng et al. (1999) further argue that the availability of services on-demand is necessary to establish and maintain service quality and firms’ differentiation efforts, and that unavailability or delay is poor service. Yao et al. (2005), in their modeling study on aircraft routing, state that in the specific case of on-demand aviation, availability of the service is contractually guaranteed. In contrast, Künsemöller and Karl (2014) argue that availability of on-demand instances is not guaranteed and that, depending on the criticalness of availability for customers, reserved goods might be preferable to on-demand goods. Accordingly, the degree of certainty of availability (contractually guaranteed or not) is not consistent between different on-demand services. Moreover, availability of on-demand services is not discussed by Chen and Wu (2013), Ma and Seidmann (2015) or Taylor (2018).

Compared with the inherent limitations associated with fixed resources that are owned, Weinman (2011, 2012) argues that, in the specific case of cloud computing, on-demand resource provisioning is characterized by nearly unlimited scalability for the customer. Resulting from this high degree of scalability, providing on-demand resources through the cloud ensures availability of exactly the right amount of resources at exactly the right time, providing “perfect capacity.” This perfect supply of resources in an environment of ever-changing demands requires elasticity to modify resource supply, as well as a high degree of granularity of supply (i.e. sufficiently fine-grained increments of resources). For example, if a customer uses 20 servers but suddenly needs 10 or 30, on-demand services are able to scale the provision of resources appropriately to the required amount for the required duration (Weinman, 2012). Similarly, Ma and Seidmann (2015) argue that the on-demand feature of SaaS enables customers to benefit from full scalability to handle possible demand fluctuations without risk. As such, the users of on-demand services are not negatively influenced by their fluctuating demand, and the service is characterized by a high degree of scalability. Chen and Wu (2013) and Künsemöller and Karl (2014) also make this point for cloud services. Although scalability seems to be an important aspect of on-demand services, it is only discussed in the context of IT services such as SaaS and Cloud computing, and not mentioned by Bai et al. (2018), Bratianu (2018), Ng et al. (1999), Taylor (2018), or Yao et al. (2005) for the on-demand service types they discuss (e.g. aviation and ride-hailing services).

Lastly, Weinman (2012), solely, argues that on-demand services should have a high degree of location independence, meaning that users should have access to the service ubiquitously and responsively, regardless of their location. Accordingly, the exact features of the on-demand services concept remain ambiguous in the literature.

2.4 The counterpart of on-demand services

Taylor (2018) argues that, owing to their continuous availability and immediate supply, on-demand services are in sharp contrast with scheduled services, where appointments for delivery are booked in advance. In support, Yao et al. (2005) argue that on-demand air transportation is a non-scheduled type of service, with random and unknown upfront demand, where passengers can directly fly anytime at request. Other examples of scheduled services and their on-demand counterparts are public transport vs on-demand ride-hailing services (e.g. Uber), movies at the cinema and on TV vs video on-demand services (e.g. Netflix), regular university lectures vs Massive Open On-line Courses, and non-acute hospital treatments vs emergency treatments. Scheduled services without clear on-demand counterparts, such as legal and consulting services, also exist.

2.5 Reflection and sub-questions

This paper is the first to review the growing literature on on-demand services. It links papers published in varied journals, supporting the creation of a body of literature currently scattered among different academic fields such as IT, business, engineering and economics.

Although the literature on the on-demand services concept is increasing, no single paper conceptualizes this service type or examines its defining characteristics. Nevertheless, the papers covered in this review enabled identification of various characteristics of the on-demand services concept (Table I). These characteristics, however, were mostly derived from modeling papers discussing specific economic or operational implications of on-demand service management (e.g. effects of pricing or capacity management) or conceptual papers/book chapters. These modeling papers often focused on a simplified and theorized service context, while the conceptual papers did not provide evidence for their arguments. Therefore, detailed empirical insights and evidence is lacking. Further, the papers generally focus on a single service context (e.g. ride-hailing, aviation or cloud services), embody different sets of the characteristics derived and do not clarify the interrelationships between characteristics. This makes it difficult to generalize findings to other contexts, and to establish a comprehensive list of key characteristics of on-demand services. Additionally, it is not clear how on-demand services differ conceptually from each other and how these differences can be made explicit to allow comparison. To address these shortcomings in the literature, this paper aims to answer the following sub-questions:

Answering these sub-questions requires empirical research covering a wide spectrum of on-demand services.

3.1 Case selection and sampling strategy

On-demand services feature in various industries and, therefore, the paper develops knowledge and theory applicable to a wide range of organizations. The aim is to uncover common patterns of key service characteristics present across varying on-demand services and identify properties that differentiate the various on-demand services. Therefore, a “maximum variation” sampling strategy is used to select cases (Miles and Huberman, 1994). Additionally, a focus on only one industry or service type might not reveal the full spectrum of on-demand service characteristics.

The on-demand services concept is poorly conceptualized and defined in the literature, making the selection of suitable cases difficult. However, the literature suggests various exemplary on-demand services (e.g. Alemi et al., 2018; Bai et al., 2018; Ng et al., 1999; Taylor, 2018), which are used in searching for similar types of cases relevant to this study. Moreover, the characteristics derived from the literature (Table I) provided additional guidance in this search.

First, a list of potential on-demand services was created. Then, seven were selected, providing a manageable number of cases with sufficient variation, and considered satisfactory for this sampling strategy (Eisenhardt, 1989; Yin, 2009). To realize “maximum variation” in the sample set, on-demand services from various service sectors such as B2B, B2C and public services are selected. Additionally, the on-demand literature predominantly discusses IT-based services (e.g. cloud and SaaS services) and platform services (e.g. ride hailing). Therefore, also on-demand services relying less on IT and services that are not platform-based are selected for a more diverse perspective on on-demand services and higher generalizability.

The selected cases were subsequently contacted to discuss conducting a case study. Both practical (e.g. time required and access to relevant managers) and theoretical matters (e.g. whether the nature of the service was genuinely on-demand) were discussed. Two contacted companies (an on-demand music streaming service and parcel collection service) were unwilling to participate. These cases were replaced by alternatives from the list. Table II describes the selected cases.

In the study, the unit of analysis was the service being offered on-demand within each case company. This encompassed the entire company or only a service offered. In the latter instance, data collection focused on the particular on-demand service.

3.2 Data collection

Data were mainly collected via interviews with managers of case companies. To guide data collection, an interview protocol including a semi-structured questionnaire was developed (Appendix). The semi-structured questionnaire was ideal because it provided guidance and flexibility to focus on what was unique to each case, allowing the study of varying on-demand services.

Generally, there was theoretical underpinning for the questions in the protocol. Based on the literature review, potentially important constructs (e.g. on-demand services definition, service characteristics, and management practices) were specified a priori and included in the questionnaire. This allows these to be measured more accurately and provides firmer empirical grounding for the emergent theory when the constructs are significant (Eisenhardt, 1989). The questionnaire followed the funnel model (Voss et al., 2016), starting with general questions about the service studied and then focusing on the on-demand aspects related to the case (e.g. the on-demand service offered, associated management practices and customers’ perspectives). Once the initial interview protocol was completed, a pilot interview was held to test usability (Yin, 2009). This resulted in minor changes, mostly in the order and formulation of the questions, because it was experienced as “too theoretical” by the interviewee, which made some questions hard to answer. New insights into the on-demand services concept gained during data collection were added to the protocol for upcoming interviews. This constant updating of the protocol is a key aspect of the abductive approach to case research (Dubois and Gadde, 2002; Voss et al., 2016), but also advocated by Eisenhardt (1989) and Gioia et al. (2013).

In consultation with the contacted individuals from the companies, interviewees were selected based on their positions, experiences and knowledge, to create a complete and profound understanding of each case. Data were collected between October 2017 and March 2018. In total, 21 semi-structured interviews were held, two of them involving two interviewees each and two the same interviewee, such that 22 people were interviewed (Table II). All interviews were conducted by the first author. Multiple visits were required to interview all necessary people. Where a manager was interviewed twice, this was to obtain answers to additional questions. Each interview lasted between 35 and 90 min (on average 55 min). With permission from the interviewees, all but one of the interviews were recorded and fully transcribed. Interview notes and transcripts amounted to 235 pages of written documentation (136,801 words). To ensure the reliability of the data, all transcribed interviews were cross-checked with the interviewees.

Other sources of evidence were also gathered, including company presentations, (informal) group discussions and company documents (e.g. PowerPoints, flyers, service contracts, websites, service smart phone applications, news items and commercials of case companies emphasizing their on-demand features). These data directly relate to the cases and were used to triangulate the gathered interview data to mitigate biases of the interviewees and enhance reliability and validity (Eisenhardt, 1989; Voss et al., 2002; Yin, 2009).

3.3 Data coding and analysis

Data analysis involved within-case and cross-case analyses (Eisenhardt, 1989). The within-case analysis enabled in-depth understanding of individual cases, in their specific contexts, concerning the characteristics of the on-demand services concept, and corresponding implications for service design and management. The cross-case analysis enabled identification of common patterns and differences between cases.

Following the abductive approach to case research (Dubois and Gadde, 2002; Voss et al., 2016), data analysis and coding was supported by preliminary insights into the characteristics of on-demand services gained from the literature review and data collection. Importantly, however, the coding process was not constrained by these preliminary insights. As argued by Voss et al. (2016), in abductive research, a template with preliminary concepts can be used to analyze data; however, the template should be adaptable to new findings and therefore not be leading.

Specifically, data analysis started with open coding of the interviews (Strauss and Corbin, 1998). To be as unconstrained and open as possible, this round of open coding was done without having the preliminary insights into the on-demand service characteristics at hand. The emerged codes were grouped according to the attributes of the on-demand service offer; management practices and properties of the on-demand service provider; and the implications for, and properties of, the on-demand service customers. All interviews were coded by the lead author. Additionally, half of the interviews were independently coded by the other three researchers. Then, the outcomes of the coding from all researchers were compared and differences were discussed until a consensus regarding the appropriate code was reached.

The open coding was followed by axial coding (Strauss and Corbin, 1998) focusing on the attributes of on-demand service offers, as this is the scope of the research. Axial coding enabled linking of the codes from the open coding with the preliminary insights into the characteristics of on-demand services, in line with the abductive approach to case research (Dubois and Gadde, 2002; Voss et al., 2016). Specifically, the preliminary insights into on-demand service characteristics were compared with the concepts from the open coding to translate and cluster the obtained concepts into higher level theoretically related themes. Then, related themes were grouped into aggregate dimensions representing the key-characteristics of on-demand services and their associated parameters (Figure 1). Finally, the data were analyzed again in a more deductive manner with the updated framework at hand to search for additional supporting quotes in the data and identify relationships between the key characteristics. Table III shows how the different concepts that emerged relate to the data (i.e. the transcribed interview quotes).

Axial coding was performed by the lead author. Subsequently, all steps in the coding process were discussed in workshops with the other three researchers, who were on purpose not initially involved in this coding process. Differing insights were discussed until consensus was reached.

4.1 Key characteristics of on-demand service offers

The results characterize on-demand services as being highly available, responsive and scalable. These three interrelated and defining characteristics are visualized in Figure 1 and described below. The findings are supported by evidence from the individual cases, summarized in Table III.

4.2 Interrelatedness of the key characteristics

Availability affects both responsiveness and scalability. The degree of scalability (in terms of quantity) depends on availability because service offerings cannot be scaled up beyond the resources available. To illustrate, the fire brigade cannot provide more fire engines than the numbers available, what also applies to the water supply, video on-demand, energy flexibility and tradesmen case. In contrast, in the alarm room and roadside assistance cases, the limited amount of resources available has no impact on their scalability (which is in terms of duration) because their clients/customers use just one “unit” of resource (i.e. one officer answering the emergency call or one mechanic in a tow truck with tools). Hence, in cases where customers consume just one unit and scalability thus only applies to duration of supply, limited availability only impacts response times and not scalability.

The response time is also often dependent on service availability. When resources are limited or temporarily unavailable, this will not necessarily result in a failure to deliver, but may increase response time. The roadside assistance case illustrates this, as interviewee three argues: “An associated mobile mechanic company has a contract with us in which they guarantee to always accept a customer request in their region. However, in busy times, response times before a customer is helped may be longer.” An exception is the Energy flexibility case, as interviewee 2 argues: “Placing a flexibility bid is a sort of a contract that you have to deliver on if requested.” Interviewee 1 further argues: “As a flexibility supplier, you need to constantly monitor the actual available flexibility when a bid is accepted and update the power system operator accordingly.” Hence, increased response times due to limited or temporarily unavailability do not occur.

Responsiveness also impacts the availability of a service. The faster a service can respond to a request (i.e. response time B), the sooner it is available for consumption. To illustrate, internet services such as video on-demand are generally able to supply instantaneously upon request, making the service directly available for consumption. However, with a limited internet connection, response times increase and customers can have difficulties in streaming. This results in long-lasting or frequent loading times in which the service is unavailable for consumption. As Interviewee 2 argues: “The moment a customer presses the play button, the streaming servers send small four-second chunks of the film to the device of the customer in real time and start playing. […] A sound internet connection, however, is required for a seamless experience.” Another example is the roadside assistance case: when a mechanic heading toward a customer is stuck in a traffic jam, the response time increases and therewith the time the service is not available for consumption.

4.3 Variety among on-demand services

Although all three characteristics are present in every studied case, there is variety among the services (Table III), both within the key characteristics (i.e. variation in the degree of availability, scalability, and responsiveness) and beyond.

4.4 Customer perspective on on-demand services

The studied cases revealed three main reasons customers request and value on-demand services: first, the nature of the demand for on-demand services is generally unpredictable/impulsive and fluctuating. As Interviewee 1 from the roadside assistance service argues: “The individual demand is hard to predict as it is linked to vehicle breakdowns, which happen randomly.” Interviewee 1 from the video on-demand service similarly argues: “We have no clue upfront what videos our customers are going to watch, they will not inform us in advance. Therefore, we need to keep a large number of videos continuously available in our digital store, ready to stream to potential customers.” This unpredictable and fluctuating nature of service demand is present in all seven cases.

Second, customers of on-demand services value immediacy of demand fulfilment. As Interviewee 1 (roadside assistance service) observed: “It is important for our customers that they can use our service at the moment it is required, because a car breakdown is always an inconvenient event. […] Therefore, when the customer needs it, the service should be available for consumption.” Similarly, Interviewee two (tradesmen service) added: “As a service provider, we reason from the perspective of the customer, having a problem that needs to be solved as quickly as possible by a specialized tradesman. Therefore, the customer can propose a convenient timeslot, and we will contact and contract a tradesman for that time as fast as possible.” This immediacy is also present in all studied cases.

The third reason customers value on-demand services is the desire to be independent and in control of service consumption. As Interviewee 2 from the video on-demand service explained: “For the customer, an important feature of an on-demand service is its guaranteed availability. As a customer, you want to be in control, meaning that whenever you want to watch a movie, it should be possible to immediately start watching one using the video service.” Interviewee 1 from the same provider added: “The customer no longer likes conventional TV, they want to watch TV programs or films at the moment they want to. […] People are busy, resulting in an increasing need to be in control of their own time use, and do not have the time to follow prescheduled broadcasts. […] Accordingly, our customer experience is not the fact that we have movies available, but that customers can watch when and how they like. That means being able to choose the movie they want to see, and to start, stop, and pause whenever wanted.” This customer control was only present in the video on-demand, energy flexibility and roadside assistance services.

4.5 Provider perspective on offering on-demand services

The cases studied revealed several management practices that enable highly available, scalable and responsive services. These are briefly presented in Table IV, and the implications are discussed in Section 5.3.

5.1 Preliminary

The characteristics found in the case studies show differences and commonalities with the ones initially found in the literature review. Three characteristics (i.e. availability, scalability and responsiveness) were found both in the literature and the studied cases. While their interrelationships were unclear in the literature, the case studies clearly enlightened this (Figure 1). Further, while “location-independent supply” and “granularity of supply” both emerged from the literature as standalone characteristics, they are labeled in the final conceptualization as parameters of availability and scalability, respectively. The specification of the key characteristics with parameters is an insight that emerged from the case studies. Further, while the literature often links usage-based pricing to on-demand services, the cases studied showed various pricing logics. Lastly, where “highly standardized supply” emerged as a characteristic in the literature, the case study learned that offering standardized services is more a means to supply on-demand, than an aim in offering services on-demand.

5.2 Theoretical contributions

This study explores the key characteristics of on-demand services and contributes valuable empirical insights into this service type. This section discusses the main theoretical contributions by answering the following research questions:

To date, the service literature had not studied the characteristics of on-demand services. Based on analysis of the literature on on-demand services and empirical evidence collected from multiple cases in different service contexts, a key finding and contribution of this paper is that on-demand services can now be described using three interrelated characteristics. Based on these three key characteristics, an idealized on-demand service is now defined as “a ubiquitously available service able to instantaneously supply the exact required amount of resources for the exact required duration, at the moment they are requested by the customer.” This definition is an advance in that it is based on characteristics determined via an empirical study covering a wide spectrum of services. This gives it firmer theoretical grounding and generalizability than other existing definitions that have predominantly focused on a single service context (generally IT). Furthermore, now that the characteristics of on-demand services are known and a comprehensive definition has been proposed, on-demand services can be better compared and contrasted with other not on-demand services such as scheduled services.

In answering RQ1b, this paper additionally contributes the development of an on-demand service continuum as a basis for providing detailed insights into the variety among different on-demand services (Figure 2). As shown, on-demand services vary in the extent of their availability, responsiveness, and scalability. Therefore, some on-demand services are closer than others to the idealized on-demand definition. Therefore, an on-demand service continuum is proposed, specified for each of these three characteristics. At one end, there is a perfect on-demand service offering that performs optimally; however, this is almost impossible to achieve in practice, especially for all three characteristics at the same time. Farther from this end of the continuum are “on-demand” services that only partly satisfy the all-encompassing definition and perform less well in terms of one or more of the on-demand characteristics. No matter how far deviated from the optimal, a service that pursues being on-demand can always be labeled as on-demand when the three characteristics are present to some degree. This, however, might result in a poor quality on-demand service offering. This on-demand service continuum is a novel concept in the service literature and clarifies how various on-demand services might differ from each other within the conceptual boundaries of the three defining characteristics. It should be noted, however, that the variations in on-demand services for different service types must be considered in context. Comparisons between on-demand services are only meaningful when considering services with similar norms in terms of availability, responsiveness and scalability.

5.3 Customer perspective: the need for on-demand services

Insights into the customer reasons to opt for on-demand services contribute to a better understanding of the specific characteristics of on-demand services, their main presence and the value on-demand services add over other services. The results showed three reasons for customers to request and value on-demand services over others. One reason is the rather unpredictable and fluctuating nature of demand for on-demand services, which makes it difficult for service providers to anticipate demand early. Weinman (2011, 2012) similarly argues that when demand is unpredictable and highly volatile, on-demand provisioning generates clear additional value relative to a fixed resources strategy. Moreover, he argues that when demand is constant or predictable, on-demand provisioning has no added value. In these situations, service provisioning can be scheduled.

The results also show that customers value the ability to consume a service as soon as it is needed. Taylor (2018) and Bai et al. (2018) similarly assert that on-demand service customers are sensitive to delays and desire immediate service provision after experiencing a need. This immediacy arises from various reasons such as emergency (e.g. fire or car breakdown services), business reasons (economic impact of waiting), or comfort and convenience (watching films, using the internet). Bai et al. (2018) further argue that the increasing on-demand services also amplifies the expectations and demands of customers for immediate service supply. Realizing this immediacy requires availability and responsiveness of services.

Lastly, customers value being independent and in control of service consumption. The latter is especially evident in consumer services and reflects the trend of switching from buying and owning goods to using goods through on-demand offerings (Lawson et al., 2016). Although customers do not necessarily want to own goods, they still desire the same immediate access and control over use as owning them would enable (Bratianu, 2018). Customers want to independently determine when and what service to consume and how much to spend financially. This requires services that are available and scalable to be able to always fulfill the exact needs with a usage-based pricing.

5.4 Provider perspective: on-demand service management

On-demand services have a unique combination of characteristics with specific implications for service management. This section discusses the key challenge on-demand companies face and potential management practices to offer highly available, scalable and responsive services. The key challenge on-demand companies face is finding the best trade-offs in the degrees of being on-demand (i.e. in terms of availability, responsiveness, and scalability) vs the associated costs and standardization of service offerings. Interviewee 2 of the fire brigade case argues: “The targeted response time for this region is 15 minutes. […] The response times are set by societal expectations and political decisions, and have an impact on the size and coverage plan of fire departments. […] Response times can be reduced, however, at a cost.”

The results showed that specific service management practices are commonly used to provide services on-demand and deal with trade-offs. One practice, applied in all cases, is to use standardized service processes. The alarm room and fire brigade both follow highly standardized protocols, while the other service offerings are highly automated through IT. This use of standardized service processes enables high availability and responsiveness. However, one drawback is that adopting standardized service processes limits the ability to customize, resulting in a poor fit with consumer needs (Ma and Seidmann, 2015). Customization, nonetheless, is mainly realized via scalability (quantity/duration) allowing customers to consume exactly what they need.

To achieve responsiveness, it is essential to either have a robust and dedicated network (e.g. internet, telephone or water pipes) that enables access to one or a few central locations of the service provider, or to have multiple distributed smaller locations closer to customers to deal with response time issues. An innovative way to be close to the customer and reach them responsively without investing heavily on key resources is to adopt a multi-sided platform model, as the roadside assistance and tradesmen services do. As Bai et al. (2018) argue, to meet fluctuating customer demands anytime and anywhere, it is economically attractive for on-demand service providers to adopt a multi-sided platform model and use many independent providers to quickly fulfill customer demands. However, the cases studied also showed that using a network of independent providers can be challenging because the participation and availability of these providers is inherently difficult to manage. Another inventive – and less costly – practice to deal with response time issues is to improve customers’ “waiting” experience, for example by informing customers of the expected waiting time and location of approaching service providers, as in the roadside assistance case. This creates less uncertainty for the customer and enables them to do other things while waiting.

Another frequent observation was that most providers have built ample capacity into their service systems, allowing them to easily scale up or down their service output to guarantee availability and support responsiveness and scalability. Although ample capacity seems common in on-demand services (Ma and Seidmann, 2015; Ng et al., 1999), on-demand service providers inevitably need to optimize their service infrastructure capacity to reduce costs to remain competitive. In addressing this, different approaches are used in the cases studied. Some of the cases use a shared pool of common resources with other parties, somewhat based on the principle of “statistical multiplexing” in which capacity is shared by multiple parties. This is made possible by demand variations over time, enabling more efficient use of available capacity (Künsemöller and Karl, 2014; Weinman, 2011). Both the alarm room and the fire brigade cases share key resources (alarm room operators and fire engines) with other alarm rooms and neighboring fire stations during local peak demand. However, a drawback of using a shared resources pool is that a high degree of standardization is required, leading to commonality in supply (Weinman, 2012). Other cases used externally owned on-demand capacity or resources as input for their own on-demand service provision. The cases adopted different approaches in acquiring external on-demand capacity. The video on-demand service hires on-demand available computing power, which is managed by a professional supplier, to stream movies. The roadside assistance, energy flexibility, and tradesmen matchmaking service acquire externally owned on-demand capacity by using the multi-sided platform model. Both forms, however, create a kind of “on-demand supply chain or network.” Establishing such on-demand supply chains and using techniques such as statistical multiplexing can increase efficiency. However, it also increases dependencies and potential risks of being unable to supply, as recognized by Künsemöller and Karl (2014). This is especially apparent in the cases where resources acquired can also be used by other parties (e.g. the roadside assistance and tradesmen services) or have other purposes (the electrical vehicles in the energy flexibility services). Coordination between suppliers in the chain and adoption of appropriate risk management strategies is therefore important (Andreassen et al., 2018).

5.5 Managerial implications

Beyond its theoretical contributions, this study also has practical implications for on-demand service management. The detailed characterization developed has practical relevance as it provides a basis for detailed understanding of a specific service’s peculiarities, which will prove valuable in developing and managing related on-demand services. Not every on-demand service needs to fully achieve the ideal position on the continuum. Becoming on-demand (i.e. moving to the right of the continuum) is hard and requires careful balancing of trade-offs with costs and customization. Hence, on-demand service providers should understand, in terms of the on-demand characteristics, their customers’ needs and their competitors’ offerings in relation to these needs. The proposed continuum can be used to operationalize and measure this. Subsequently, having specified the on-demand requirements for each characteristic of the continuum, appropriate targets can be set. Accordingly, practitioners can use the on-demand service continuum as a management tool to steer their service offerings toward the on-demand requirements of potential customers.