Although emotions are part of our everyday lives, they have largely been neglected in research on work until Hochschild’s seminal work on emotional labour (1983). Her study focused on how the emotions of service workers were formed and exploited by employers. According to Hochschild (1983), emotional labour is the effort to manage and display emotions in accordance with social expectations, norms or job requirements. It involves regulating one’s feelings to meet the emotional demands of a situation, often in service-oriented roles like customer service or caregiving. This can involve suppressing personal emotions and expressing those deemed appropriate for the context, which can be mentally and emotionally taxing. Later research has highlighted that emotional labour is not always explicitly demanded but can arise through implicit role expectations. Houben and Wüstner (2014), for example, show that even in technically oriented roles such as information and communication technology (ICT) support, workers engage in emotion regulation to maintain professional standards and manage client interactions. Their study suggests that role performance and emotional expression may be guided less by direct managerial instruction and more by internalised norms shaped by the organizational context.

Hochschild identifies two strategies that individuals use to manage their emotions in emotional labour: deep acting and surface acting. In surface acting, individuals modify their outward expressions of emotions without genuinely feeling those emotions internally. They may fake or display certain emotions required by their job or social role even if they do not authentically experience those feelings. For example, a customer service representative may smile and appear friendly to customers even when they feel frustrated or upset. Deep acting involves a more internal process where individuals try to genuinely experience the emotions they are expected to display. Instead of simply faking it, they may engage in cognitive or emotional processes to evoke the desired emotions. Interestingly, deep acting has been shown to be less emotionally draining than surface acting. Holman et al. (2008) argue that the increased effort involved in producing fake displays is associated with lower well-being.

Holman et al. (2008) also found that employees who experience high job demands (cf. Demerouti et al., 2001; Bakker and Demerouti, 2007), such as interpersonal job requirements, workload and unjust interactions, are more likely to experience the negative aspects of emotional labour, such as more negative emotions, greater surface acting and more faked emotional behaviour. On the other hand, employees who have access to large resources (cf. Demerouti et al., 2001; Bakker and Demerouti, 2007), such as control and social support at work, are more likely to experience the positive side of emotional labour, including positive emotions, deep acting and genuine displays.

A case study design was applied when exploring how the use of service robots impact FLEs’ emotional labour. Qualitative data were collected between January and June 2024 during off-season outside of weekends and holiday periods. The subject (cf. Thomas, 2011) of the study was an amusement park and the object (cf. Thomas, 2011) the interplay between robots and emotional labour. Multiple methods were used: semi-structured interviews, observations and document analysis. The study participants represent voices from FLEs, managers and HR officers. By mirroring information collected through different methods and from different informants and documents we reached a deeper understanding of the role robots play in frontline work in an amusement park (cf. Bowen, 2009).

The case is an amusement park in Sweden in which all together eight service robots assist restaurant staff in the park itself and in a hotel belonging to the park but located outside the gates. Three robots are active in two different restaurants within the park. The other five robots are active in a restaurant and a café located in the hotel. There are two types of robots in use: grey transport robots used to transport dirty dishes from the eating areas of the restaurants to the kitchen and white serving robots used to bring food from the kitchen to the tables. Robots bringing food do not serve the customers but position themselves where they have been instructed to by the waiting staff, so that the waiting staff can serve the food from the robot to the table. Both robot types have digital displays at the top of them which personnel use to instruct them where to go. The robots have blue blinking eyes when they are in motion and red crosses instead of eyes when they need to be charged. Both robot types have sensors that detect and respond to objects or people that are in the way of the robot. If this happens, the robot stops and tells the person or object to move. The white robots are able to speak more than the grey ones; they have small ears and can sing and purr like a cat. All robots have been given names by the serving staff (FLEs) who work in the restaurants and café. The FLEs serve food and remove dirty dishes together with the robots and interact with the customers, which are often families with children due to the location in or next to an amusement park.

Semi-structured interviews (Bryman, 2006) were used to get an in-depth understanding of how robots impact FLEs’ emotional labour. The interviews with employees circled around four areas: (1) demands/workload and influence, (2) social relationships, (3) learning and (4) organizational working environment. An example of a questions is: “In what way has the robot changed what you do at work and how you perform your tasks?” Interviews with HR were mainly about the organization’s work with the working environment, e.g., “How does your organization work with the organizational and social working environment?” Interviews with managers were mainly about why the robots were introduced and how they worked with working environmental tasks in relation to the robot. Examples of questions were: “What is the purpose of the robot?” and “How do you handle working environment issues?” The interviews contributed insights into how staff experienced working with robots.

The observations contributed to a first-hand understanding of interactions between staff, customers and robots on “the stage”. The observers took a passive role (cf. Baker, 2006), meaning that they were present on the stage but did not interact with the FLEs. This approach is positive in the sense that the researcher is able to watch what is happening while being detached from those observed, but it is negative in the sense that it does not allow the observer to hear all conversations or ask questions about what is happening and why. To reduce the risk of influencing participants’ behaviour, we adopted a passive stance during observations. This approach was chosen to avoid disrupting the natural onstage performance (cf. Goffman, 1990 [1959]) and to capture how interactions and emotional expressions unfolded in practice. The observations focused on what happened on the stage, for example: How does the interaction between the robot, the customer and the employee affect the work process? Are they disturbed or interrupted? What expression of emotions (joy, frustration, anger) could be seen in customer meetings?

As we were not allowed by the management to collect data during peak periods, our observations did not include the experiences of crowding told by the interviewees.

Document analysis complemented other data collection methods used in this qualitative case study (Bowen, 2009) and was used to understand the formal values supposed to guide the employees. The strength of using documents in our study is that it uncovers important formal contextual factors that we believe guide the FLEs, not least the interaction with the service robots and the customers (cf. Bowen, 2009).

In total eight interviews were conducted, with four men and four women, between January and May 2024. Six of the interviews were with waiting staff in three different restaurants or cafés, including one restaurant manager (man), three head waiters (two men and one woman) and two waitresses. The waitresses were young and had no previous working experience in restaurant work before they started a year earlier. The head waiters and restaurant managers had between 11 and 25 years of experience of working in restaurants. Two of the interviews were with staff at central management level of the amusement park, including one safety and health strategist working in the HR department (woman) and one operations manager of the restaurants (man). The latter was responsible for leasing all robots and is also the one who took the initiative to introduce robots in the first place.

Observations performed in the restaurants were carried out in May and June 2024. During the May observation two researchers observed worker–customer–robot interactions in both the hotel and park restaurants for approximately 60 minutes. During the June observation another researcher spent 195 minutes in the hotel restaurant. By observing the interactions only in off-season, we only witnessed unproblematic situations. However, this was compensated by detailed descriptions of stressful experiences during peak periods in the interviews.

A company document named “Our Vision and Compass”, presenting the mission, vision, business idea, goals and values of the park, was also analysed.

Thematic analysis was conducted using the method inspired by Clarke and Braun (2017). This process involved identifying, analysing and reporting themes within the qualitative data to provide a nuanced understanding of the phenomenon under study. Our analytical approach was further informed by abductive reasoning (Thompson, 2022), in which we moved iteratively between empirical observations and theoretical concepts. Rather than being guided from the outset by a specific theoretical framework, the analytical process was exploratory and open-ended, allowing for unexpected patterns and ideas to emerge from the data.

Initial readings of interview transcripts, observational notes and organizational documents opened for discussions among the authors around recurring motifs related to role-play and emotional management in employee–robot–customer interactions. While emotional labour theory (Hochschild, 1983) offered an initial lens, it was through the evolving analysis that we found Goffman’s dramaturgical perspective (1990 [1959]) to offer a particularly generative vocabulary for understanding how employees engaged in behaviours resembling theatrical performances.

This perspective helped us conceptualize the service setting as a “stage”, with employees performing emotional labour not only towards customers but also in relation to the robots, which were integrated into the service experience. Concepts such as frontstage, backstage, scripts and role-play became central to our interpretation of the data. The idea of “theatricalisation” (Blackwell-Pal, 2020) thus emerged as an overarching theme, encompassing both the performative dimension of employee behaviour and the symbolic role played by the robots.

In this way the analysis was not a straightforward application of predetermined theory but rather a process in which theoretical framing emerged through engagement with the empirical material. This iterative interplay between theory and data allowed us to identify four key themes: the script and the stage, the act, the actors and humans left on the stage. For simplicity and reasons related to anonymity the amusement park is simply referred to as “the Park”.

An essential aspect of understanding the experiences and interactions of FLEs in the service sector with robots is to comprehend the context. Drawing on Goffman’s dramaturgical perspective, we refer to this framework as “the stage”. The stage encompasses the local conditions for the drama, which include not only the material aspects such as decor and props but also the behavioural expectations set by the “producers” and “directors” of the drama, namely the managers and leadership.

A central theme in the literature on emotional labour is the necessity for service workers to either suppress or emphasize certain emotions in alignment with social expectations, norms or job requirements (e.g., Holyfield, 1999; Van Dijk et al., 2011). Our empirical findings indicate that interactions with robots elicit both irritation and joy among service workers. Albeit generally positive to the robots, the service workers must suppress any irritation they experience to maintain professionalism and meet the expectations of their role. This duality underscores the complex emotional regulation required in the service sector, particularly in environments where technology and human labour intersect. Furthermore, the empirical material provides several examples of what Blackwell-Pal (2020) terms the theatricalisation of labour. In line with Holyfield’s (1999) concept of the expectation to embody the (amusement) product, FLEs assist in bringing the robots to life. Part of this involved naming all the robots.

We have named all the robots because it’s fun, and the children love them. […] Adults also appreciate the robots, often taking pictures and videos, finding it amusing because it is still quite uncommon in Sweden today. (Waiter 4)

Another aspect of humanising the robots was attributing human characteristics to them. For instance, if a robot was undergoing maintenance or charging, it was described to the customer that the robot was “not feeling well today”.

We often talk to the robots by name when guests are nearby or with the staff. For example, ‘Robin is not feeling well today. He has a problem. Put him on charge.’ That’s sort of how we talk. (Waiter 1)

In addition to attributing personal characteristics to the robots, the material also reveals more creative ways of bringing the robots to life. In this specific case, a zoomorphic robot had been given cat-like attributes:

If you pat it on the ears, it might say, ‘Oh, that tickles’ or ‘Don’t touch me’. Children believe that… When I talk to the children and they ask about it, ‘What is this?’ I usually say, ‘Well, there’s a little cat living inside that controls the robot.’ They find it very exciting. So, it becomes almost more real… We talk about the robots as if they are humans […]. (Waiter 5)

In this context, it becomes interesting to question whether this type of role-playing should be seen as “deep acting” (Hochschild, 1983), meaning that the service workers fundamentally see the robot as something more than a machine, or if this process of zoomorphising or anthropomorphising should be seen as “surface acting” (Hochschild, 1983), i.e., a performance to enhance the customer experience. There are different views on this in the empirical material. One example of surface acting is how some FLEs describe the employee-robot interaction purely as a performance. Although they name the robots and talk about them as individuals, they still see them as robots:

So, it’s fun to talk about ‘here comes Mark and over there is Laura and then comes Mimmi.’ So, we have this in the communication with the guest. But it’s not that I see them as anything other than that they should transport our dirty dishes from point A to point B [laughs]. (Waiter 4)

Another example of surface acting is the following description of how the robots are merely machines that are brought to life solely through FLE acting:

We talk about the robots as if they were humans though […] They have names, so we usually say ‘where is she?’ Things like that. ‘Where has she gone?’ Or ‘is she working?’ So, it becomes more real in a way if we… what do you say, talk about them as if they were alive. […] Yes, they are just machines, but still, we talk about them as if they were something alive. (Waiter 5)

Deep acting is exemplified by employees showing signs of seeing the robot as something more than a machine outside direct customer interaction. Among the affirmative responses to questions about whether they see the robot more as a colleague than a machine, some talk about the robots as if they have personality traits, such as being daft, quiet or even amusing.

The first Mimmi, and the others are a bit… shall we say, not as smart as Mimmi, they are a bit quieter. […] So, it’s Mimmi. Quite an entertaining robot. (Waiter 1)

As mentioned, emotional labour also involves highlighting and amplifying certain specific emotions. In the case of the Park, the goal is to offer amusement to its customers. The Park’s mission statement reads that it is specifically about offering joy together to the customers as part of the amusement: “The park exists so that people can experience joy together. The park’s guests are united by a longing to experience joy together.” In the work of contributing to this experience for the customers, spreading joy consequently became part of the service staff’s duties, as reported by this informant:

Obviously, the environment in the restaurant contributes to guests finding things a bit more enjoyable as soon as they enter, due to the presence of robots. This makes the guests’ experiences entirely different, which means that they also largely carry this with them when they approach the staff to order something. (Manager 1)

In addition to spreading joy independently of the human employees, the robots were described as an integral part of the interaction with the customer:

Yes, but I think it becomes an easier task when the robot comes out. You get a starting point to communicate with the guests. You get a free conversation topic: ‘Hey, here comes Mimmi’ and ‘Mimmi does this’. (Waiter 3)

From this perspective, robots contributed to alleviating the emotional labour of service workers. They achieved this by enhancing guest satisfaction and facilitating communication. By creating a more enjoyable experience for guests, robots helped to reduce the emotional strain on service workers, allowing them to focus on providing high quality service without the added pressure of managing negative emotions. However, the interaction between staff and robots also depended on the context. In this study, the interaction was more intense in the restaurant and café of the hotel than in the restaurant in the park, where customers ordered and picked up their own food via a telephone app. Consequently, interaction between the staff and the customers was less frequent – including interaction with the robots.

It is interesting to note that our findings reveal no evidence suggesting that this performative behaviour was an explicit expectation from management. In a sense, this “drama” operated without a director. Instead, it appeared that expectations for such behaviour were shaped by the context of the amusement park and the broader, implicitly formulated expectations placed on service workers to entertain guests. This absence of a “director” to coordinate and guide the performance is further examined in the discussion section of this article. The emotional regulation required by the FLEs combined with the theatricalisation of labour (bring the robot to life) created a more complex work situation than if there were no robots.

Observation: The robot comes rolling in from the kitchen, loaded with food, into the dining area. The children follow the robot as it passes by. Adults point it out to their children as it approaches. The robot continues and stops at its final station, the landing spot. The staff are not involved beyond unloading the food and carrying it to the guest. Later, the staff loads the robot with dishes and as they press a button, the robot leaves the dining area and moves through the doors to the kitchen. The robot is now out of sight, and the employees remain in the restaurant.

Departing from the two central concepts in Goffman’s dramaturgical perspective, frontstage and backstage, we ask how the relationship between these two dimensions of the workplace changes after the entry of service robots. When taking a step back and zooming out from the moments of interaction in order to consider the set-up of the scene (Goffman’s frontstage), it can be noted that the automatic access of the staff to the backstage decreases because they no longer regularly leave the stage to enter the dishwashing/kitchen area. Consequently, not only has the direct interaction between staff and guests changed (whether the staff should play along or not), but also the scene itself – where the robot had its own “role” in relation to the guests, thereby affecting role-playing as well as work practices. One respondent highlighted that the work had become more focused on being with the guests, i.e., on stage.

You avoid that part a bit now [going to the kitchen], because now you don’t have to stop what you’re doing until the food is in front of you, more or less. So, then you have more time to be out where it happens or help more guests around the bar or café. And the same goes for the dishwashing robots, you don’t have to go away all the time and leave it [the dishes] somewhere, but you place it here and then it solves itself. (Waiter 3)

The more informal and collegial interactions backstage with co-workers were particularly important for the person working in the café, who often worked alone. In the restaurants, there were always at least two front-stage workers present at the same time.

So that it [the food] can get out of the kitchen and be served to the guest as quickly as possible… Yes, it’s just that tasks are taken away from you in a way, even though it can sometimes make things easier. But also, when you enter the kitchen, you end up talking to more colleagues, and that can also be pleasant. It’s the daily human contact. (Waiter 5)

By relieving service workers of tasks such as carrying dishes to the kitchen, some of the heavy and strenuous physical work was alleviated. At the same time, it emerged that the work in some sense placed even more demands on emotional labour due to the robot’s entry and, for some, reduced social interaction with colleagues.

While social robots in the restaurant have alleviated some of the emotional labour from the FLEs, while adding other types of emotional labour demands, it could also be argued to have taken over the role as the main actor, where the FLEs now plays the supporting role in the theatricalisation of work. As the star of the performance the robot is periodically withdrawn from the stage for maintenance and recharging, while frontline employees remain continuously present in the customer-facing environment. This ongoing visibility may limit opportunities for informal recovery and backstage interaction, potentially reinforcing the emotional demands placed on staff between each “act”.