1 Introduction

The service revolution has started as technology rapidly becomes smarter and more powerful while getting smaller, lighter, and cheaper. This includes hardware such as physical robots, smart self-service technologies, drones, and wearable technologies, as well as software and systems such as natural language processing (NLP), image processing, cloud technologies, mobile technologies, geo-tagging, virtual and augmented reality, machine learning (ML), generative artificial intelligence (AI) such as ChatGPT, and the metaverse (Bornet et al., 2021; Dwivedi et al., 2023b; Mariani et al., 2022). Together, we expect that these technologies will transform virtually all service industries and will lead to improved customer experiences, service quality, and productivity all at the same time (Wirtz & Zeithaml, 2018).

Humanlike robots (e.g., Pepper and Nao), voice assistants (e.g., Amazon’s Alexa and Apple’s Siri) and chatbots (e.g., Cleo) have become part of our daily lives and are widely adopted in various service settings (Wirtz et al., 2018). For instance, the robot Spencer at Amsterdam Schiphol Airport helps passengers find the right departure gate while scanning boarding passes. Robots are used in hospitals to interact with patients and monitor their vital signs (Paluch et al., 2022). Due to the advancement of robotic technologies and AI, service robots are now able to perform various tasks autonomously without any or little support from human staff. These include routine and repetitive tasks (e.g., delivering room service, bringing luggage to guest rooms, and assisting customers online) as well as analytical tasks (e.g., robo advisors processing large amounts of data and making personalized investment recommendations).

When adopted in the organizational frontline, robots and AI-based services not only enhance the experience of the customers but also offer exceptional opportunities of economies of scale and scope to the service provider. Most of the costs of these new technologies incur in their development and their incremental costs are often close to zero, especially when it comes to virtual robots. For instance, Heathrow Airport in London uses a hologram-based humanoid service robot to offer travellers directions to check-in counters and gates and provide information about arrivals and departures. Since these holograms only require low-cost components such as a camera, microphone, speaker, and projector and do not need to occupy floor space, the airport could easily install holograms at any place where travellers might need assistance, even in the today generally neglected spaces such as car parks.

These examples illustrate the magnitude of the impact of the shift in the service sector toward robot and AI-powered services. In this opinion piece, we discuss the implications of the service revolution for service firms, their marketing, and their customers. Specifically, we examine how intelligent automation (IA) in the form of service robots, metaverse, wearable and smart self-service technologies, and generative AI impact service firms and customers by allowing companies to achieve customer satisfaction and cost-effectiveness while mitigating their potential risks (i.e., privacy, ethical and discrimination risks). Figure 1 provides an overview of the key points we make in this article.

Fig. 1
figure 1

The implications of the service revolution for service firms, their marketing, and their customers

Full size image

2 Cost-effective service excellence enabled by new technology

Intelligent automation (IA), service robots, and AI will enable unprecedented enhancements to the customer experience, service quality, and productivity. We believe that these technologies power a service revolution that enables service firms to achieve cost-effective service excellence (CESE) and therefore has the potential to increase our standard of living. As the industrial revolution did for manufacturing, this time the service revolution will radically transform services such as financial, communications, logistics, education, health care, and hospitality.

Service excellence is traditionally associated with a high level of customization and personal touch provided by service employees. With the advancement in digital service technologies, interactions with service providers have been dramatically changed and human service staff is frequently no longer at the centre of the service encounter. Service robots such as physical robots, chatbots, and virtual assistants are increasingly adopted by service providers to complement or replace human staff, and customers today may complete an entire service exchange by only interacting via technological interfaces. This is particularly true for information-type services where the robots’ ability to simplify, organise and scale information allows for more efficient services. For example, fintech and healthtech companies use automated platforms including phone, web, and apps where no human frontline staff is involved in providing service to customers. Mobile banks such as Monzo, and payment providers such as Venmo, allow customers to split payments among family and friends directly through their apps. GoodRx, a US-based healthcare company that operates a telemedicine platform, offers consumers the possibility to track prescriptions drug prices and manufacture coupons to access discounts on medications through its website and mobile app.

While today these types of automated services are not the majority yet, we expect they will quickly grow and become mainstream in the near future, and service after service will be productized (Wirtz et al., 2021) and automated. When this happens, the costs associated with these services will reduce significantly leading to more concentrated and less fragmented markets, such as the ones for sharing platforms (e.g., Airbnb), search engines (e.g., Google), and maps (e.g., Google Maps). Not only information-type services but also physical services will be disrupted by new technologies, which will bring opportunities for innovative business models, including in the B2B context. For example, shipping and mining industries have started to adopt autonomous systems to manage their operations (Wirtz & Kowalkowski, 2022).

Many service jobs are characterized by repetitive tasks, low discretion, and little or no empowerment (Wirtz & Jerger, 2017). For such roles, robots may be a better service delivery channel for all key stakeholders. Service robots can increase productivity, improve efficiency, and ensure standardization of services while lowering operational costs and freeing employee time for more fulfilling and value-adding roles. Moreover, since robots are not ‘human’, they do not experience fatigue and emotional labour even in the most mundane and repetitive jobs.

If these are scenarios where service robots and intelligent automation will substitute human work, there are also types of services where human–robot collaboration will be the best option to provide cost-effective service excellence. Here, firms may use human employees for more professional service roles, while technology provides analytical support and takes care of the more repetitive and mechanical tasks. For example, professional roles such as lawyers require a range of soft skills and capabilities including negotiation, persuasive skills, flexibility, as well as social-emotional abilities, which cannot be easily performed by technology. In these cases, AI can be used to assist lawyers in the preparation of case notes and thereby increase the productivity of the firm at lower costs.

3 Consumer responses to new technologies

Although many customers already use various types of technologies including voice assistants (e.g., Alexa and Siri), chatbots, and virtual friends (e.g., Replika) every day, research on consumer reactions to intelligent automation, AI-based services and service robots offer mixed empirical evidence on whether individuals tend to accept or resist these technologies. A first stream of studies provides support for the so-called ‘algorithm aversion’ phenomenon (Dietvorst et al., 2015) and shows that customers prefer human and personal services in a variety of settings including medical, financial, and organizational contexts. Algorithm aversion happens because people believe that new technologies are not able to capture their individual unique circumstances (Longoni et al., 2019), cannot provide explanations for their decisions and outcomes (Önkal et al., 2009), and are generally considered less fair and trustworthy when used in selection decisions (Lee, 2018).

On the other hand, an increasing number of studies document the so-called ‘algorithm appreciation’ tendency and show situations where individuals accept and prefer more automated and AI-based services than human services (Logg et al., 2019). For example, people rely more on automated recommendation systems when they believe that the technology has greater expertise than themselves regarding the context at hand (Banker & Khetani, 2019).

Taken together, these two streams of research provide contrasting answers to the question of whether consumers favour or avoid these technologies. However, when looking at how customers interact with new technologies, it is unusual to find people consistently accepting or resisting AI-based services. More relevant for services are the situational elements (i.e., contextual, personal, and technological) that can turn individuals’ avoidance into acceptance in specific service contexts.

A first example is provided by utilitarian services. People trust, accept, and prefer more automated technologies when they provide instrumental, functional, and practical utilities (Longoni & Cian, 2022). This preference is driven by the lay belief that AI technologies are more competent than humans in evaluating utilitarian attributes because they are rational and logical whereas humans are associated with emotions and feelings. That is, people would trust more an AI recommendation about a business trip (i.e., utilitarian) rather than a holiday (i.e., hedonic). Similarly, new technologies are viewed as more effective and are preferred when they offer services that serve objective tasks (Castelo et al., 2019) and involve measurable and quantifiable outcomes (e.g., analysing data, scheduling events, and predicting the weather). In these contexts, people generally believe that AI technologies are better equipped with the cognitive and rational abilities necessary to perform these tasks (Castelo et al., 2019). For instance, recent and advanced versions of service robots can be found in professional services such as investment advice where they provide customers with financial recommendations based on sophisticated data analysis.

People also appear to prefer interacting with human-like technology as it is perceived as more competent (Belanche et al., 2021), trustworthy, and enjoyable (van Pinxteren et al., 2019). Humanoid robots, for instance, increase perceptions of service quality, visit intentions, and willingness to pay for a service (Yoganathan et al., 2021). Similarly, virtual chatbots that are able to imitate natural human-to-human conversations result in more positive evaluations of the service experience (Hildebrand & Bergner, 2021). However, research pointed out that when robots are ‘too much’ humanlike, they can result in negative emotions (Mende et al., 2019) and decrease consumer attitudes toward the technology and the service (Kim et al., 2019).

Finally, service robots and AI services are also preferred in settings where consumers may be exposed to judgments and opinions by service employees such as in potentially embarrassing encounters (Pitardi et al., 2022). This preference is driven by customer beliefs that robots are unable to form opinions and make moral and social judgments. Thus, customers feel more comfortable interacting in potentially embarrassing service encounters with a robot rather than a human employee (Pitardi et al., 2022). This may explain the proliferation of fitness apps such as Under Armour that utilize AI to offer personalized health programmes. These services are beneficial to customers because they feel more at ease receiving dietary and physical activity recommendations from a non-judgmental AI than from a fellow human.

There are situations, however, when human and personal services are preferred over automated technologies. People believe that these technologies lack the affective and emotional abilities associated with more subjective tasks (e.g., writing news articles or composing music) which makes them less preferred in such contexts (Castelo et al., 2019). People also prefer humans in service contexts that have higher symbolic value and are strongly associated with self-expression motives (Granulo et al., 2021). That is, robot and AI-provided services are more consistent, uniform, and standardized, and therefore viewed as less suitable when customers want to express their uniqueness as individuals. Similar responses have been documented in automated services (e.g., assisted bikes, automated cars) when such services are strongly linked to customers’ identity (Leung et al., 2018). ‘Proud bikers’ and individuals who consider biking as part of their identity, will enjoy more performing this activity themselves, resisting the automation offered by intelligent machines. Similarly, individuals who consider themselves bakers will avoid adopting automated baking products as they think automation replaces skills essential to the baker’s identity.

In sum, the literature shows that, the deployment of intelligent automation and robots in the frontline generally receives positive reactions from consumers. Exceptions, however, may include highly hedonic services and services that are strongly connected to the consumer’s identity.

4 Mitigating potential risks of intelligent automation

Concerns over ethical, fairness, and privacy issues may hinder the adoption of technologies (Pitardi & Marriott, 2021; Wirtz et al., 2023a). These concerns and how service firms can mitigate them is discussed in this section.

To perform at its best, new technology and AI depend heavily on the access to large volumes of data, algorithms, and machine learning (Puntoni et al., 2021; Zuboff, 2015). This results in the collection and integration of vast amounts of different types of data ranging from individual data (e.g., profile and transaction data, social media footprints), direct data (e.g., clickstream data and photos provided and uploaded by the customer), inferred data (e.g., data not provided by customers but inferred from user data such as credit scores), private data (e.g., transactions and profile data), and public data (e.g., weather, maps, locations, and reviews).

The accumulation of all these data types carries important ethical, fairness, and privacy risks for customers and society, and is viewed with increasing concern (Lobschat et al., 2021; Wirtz et al., 2023a). Issues and problems related to privacy, data breach, and data manipulation have become increasingly common and often cause customers vulnerabilities. Especially in the context of AI and automated technologies that heavily rely on the access to private and non-private data, it has become imperative to find ways of holding firms’ decisions and behaviours accountable to moral and ethical considerations.

Three relevant areas in which firms should balance their own interests with those of the consumers are privacy, ethics, and potential bias (or fairness) (Davenport et al., 2020; Wirtz et al., 2023a). Privacy violations include issues such as undisclosed commercialization of data, data breaches, and identity theft (Martin, 2015), as well as highly personalized advertising, sneaky online tracking, and ubiquitous surveillance (e.g., Martin & Murphy, 2017; Quach et al., 2022). Ethical issues involve threats such as coercive data disclosure, dehumanization, social deprivation, disempowerment, and social engineering (Belk, 2021; Breidbach & Maglio, 2020; Čaić et al., 2018; Wirtz et al., 2023a). Bias and fairness issues include algorithmic biases and potential discrimination (Someh et al., 2019).

In these areas, organizations need to mitigate the concerns that arise from the use and adoption of new technologies by embracing corporate digital responsibility (CDR; Lobschat et al., 2021; Wirtz et al., 2023a). Norms, shared values, and actionable guidelines for a responsible use of technology and data need be developed and adopted. For example, organizations may decide to regulate which data to capture (e.g., avoid biometrics or social media accounts), how to use it (e.g., creating indices and scores to support impactful decision making such as loan approval), and which data to delete from their systems.

Notwithstanding this discussion, companies often engage in poor CDR practices. This is mainly due to challenging trade-offs they have to face when choosing between good CDR practices and organizational objectives. Wirtz et al. (2023a) suggest that the choice between a more or less responsible approach is guided by a ‘CDR calculus’ where organizations weigh the benefits and the costs associated with CDR. The potential costs associated with a responsible management of digital data include the costs of developing and operating good CDR such as the costs of creating a CDR culture and effective CDR governance, opportunity costs of a reduced customer experience (e.g., using less data may result in less personalization, customization, and convenience), costs of lost incremental revenues (e.g., less effective upselling and cross-selling), and untranslated cost savings (e.g., less effective supply chain optimization). However, engaging in good CDR practices may offer important benefits for organizations. For instance, it may help organizations being considered a "good organization" and improve their brand image and become a competitive advantage to develop brand equity, consumer trust, and loyalty. Furthermore, it can mitigate potential legal, reputational, and regulatory risks (Mueller, 2022; Wirtz et al., 2023a).

5 Research opportunities

As this is an opinion piece, we take the liberty to raise a few of broad research streams we find particularly exciting. First, generative AI (e.g., ChatGPT) is getting closer to achieving artificial general intelligence, especially when rules are superimposed (e.g., a firm’s current chat bot knowledge base) and memory is added (e.g., a firm’s generative AI remembers interactions with a particular customer and is connected to that customer’s CRM data). This allows front line technologies such as chatbots, digital agents, and service robots to understand and learn virtually any intellectual tasks just as frontline employees can, which makes today’s chatbots look primitive in comparison. We expect that generative AI will become a common customer interface that will provide service levels and interactions closer to what today’s human service employees provide. For example, Soul Machines, a developer of what they termed ‘digital people’, designs their virtual agents to deliver their clients the desired brand image, positioning, and customer preferences (e.g., age, ethnic group, and gender). For customers it will therefore be largely indistinguishable whether they are being served by a digital technology or a human frontline employee (Dwivedi et al., 2023a).

It would be interesting to explore how customers would respond to such technologies and if they care at all if both channels deliver service reliably. Our prediction is that customers are unlikely to care and may even prefer technology to be the interface if it delivers on customer expectations. One might even expect that the instant availably (i.e., no waiting for a customer contact agent to become available), 24/7 service, and in one’s preferred language will be preferred by most customers over having to dial into a traditional call centre. This is akin to today where the vast majority of customers prefers to withdraw money from ATMs rather than going into a branch and deal with human tellers. Further research on consumer responses to frontline technologies with general intelligence seems a fascinating and wide-open area for research.

Second, the implications and applications of the metaverse in service delivery seems a promising area for investigation with a myriad of potential use cases. For example, mirror worlds could be used in showing customers what to expect (e.g., what a theme park, a hotel or a holiday destination looks like). Virtual real‐time experiences can mirror live events such as rock concerts, soccer matches and F1 races. This may lead to new business models where events are offered at a premium subscription fee where soccer fans no longer watch the World Cup on TV but join a virtual mirror world to experience it first‐hand (virtually at least) from a preferred seat in the stadium (Dwivedi et al., 2023a).

Even keynote addresses, lectures, and skills training could be offered on a global basis with a full and immersive experience at close to zero marginal costs. For example, teaching languages could be powered by generative AI and digital people in a metaverse. French classes could be held in realistic settings such as restaurants, shopping, and business meetings. Such learning will be highly immersive. Furthermore, unprecedented levels of personalization are likely to be achieved where the system picks up use (or lack of) suitable vocabulary, identifies grammar issues, and it can even pick up pronunciation and specifically train based on the specific issues of a learner. Such training is likely to be much more effective than today’s online and classroom environments. Such innovations offer have the potential to bring transformational customer experiences at negligible marginal cost (Wirtz et al., 2023b).

Recent work suggests that we should explore new concepts such as immersive time which is defined as “the conscious, deliberate, and dedicated time during which consumers escape the real world by using a headset and other accessories to continually engage in the metaverse” (Mogaji et al., 2023). We concur as service research has yet to examine the potential the metaverse can play in service delivery, how it can be integrated with current delivery channels and tools (e.g., email, phone, account updates, robo advisors), and how consumers will respond and interact with service provided on the metaverse.

Third, the adoption of automated technology holds exciting promises but presents also potential downsides mainly associated with their misuse and unintended consequences in terms of ethical, fairness (biases), and privacy violations. Discrimination issues arising from the adoption of algorithms as decision-makers in a variety of settings (e.g., decisions loan approvals, loan amounts, and interest rates) have been widely discussed in public media. As more and more services will be delivered by digital agents, additional negative and unintended consequences such as dehumanization, social isolation, and addiction may arise. For instance, the growing adoption of AI well-being applications has also been associated with mental health harm for users and a raising emotional dependence on technology (Laestadius et al., 2022). More research is needed to understand how such risks may be alleviated. Given rapid technology advancements, it seems also fruitful to examine how these very same technologies can be designed to reduce rather than cause social and ethical issues. For instance, since AI and machine learning technologies are already able to detect fake content and deep-fake pictures online, exploring how they can be designed to overcome discrimination rather than causing it seem fruitful avenues of future studies.

Finally, and related to the previous point, the CDR calculus appears as a key underlying reason for companies to adopt poor CDR practices (Wirtz et al, 2023a). Yet, little research has focused on this topic (Mihale-Wilson et al., 2021). More studies are needed to explore how service firms can manage the trade-off between their organizational goals and good CDR practices, and further develop their CDR commitment. Previous research in service research on cost-effective service excellence (Wirtz & Zeithaml, 2018) and in the management literature on organizational ambidexterity (e.g., Raisch & Birkinshaw, 2008) has already demonstrated how firms can deal with conflicting objectives. We believe that research empirically examining the benefits and associated costs of CDR not only within service firms but in their broader service ecosystems may identify potential avenues that can help companies achieve their objectives while developing a strong CDR culture.

In closing, this opinion piece deals with the future of services and their delivery, and we hope that we managed to excite our readers to join the growing academic community that works on these topics.