Use Case

Application of an E-Butler for Jumeirah Hotels & Resorts in Frankfurt

Jumeirah Hotels Use Case

A Use Case for the Application of an E-Butler System at Jumeirah Hotels & Resorts in Frankfurt in line with Deloitte Germany, Initiative D21 and ekipa – a Plattform for innovation Challenges.

Jumeriah Logo

Jumeirah Hotels & Resorts

In this case study we are analyzing the issue of the application of an E-Butler in the hospitality domain. For this we apply the framework provided by Müller & Andersen (2017): “Trade-offs in the digital ethics domain” in line with the D21 initiative to assess the various issues associated with deploying a digital butler at Jumeirah hotels & resorts. Risks in terms of privacy and customer service quality are the most serious ones for the luxury hospitality domain, as hotels and services from this domain are dependent on high quality and individualized services that support the guests with a high amount of empathy and a high standard code of conduct. Digital assistance systems may lead to service deterioration and quality decrease rendering guests unsatisfied and the thoroughly cultivated brand image destroyed. The early developments of those assistance systems have led to many luxury brands experimenting with digital initiatives such as LVMH opening up an incubator at Station F in Paris or luxury department stores embracing omni-channel strategies to target their progressive Asian clients.

How can we ensure ethical A.I.?

The case for digital assistance systems in hotels is still new and related to the phase of exploration, trial and error. The first part of this case study covers the analysis of ethical trade-offs when applying the E-Butler alongside four framework dimensions that depict the process behind the architecture of artificial intelligence. In the second part we include a stakeholder analysis and conclude with a discussion of the corporate digital responsibility which is followed by an outlook.

In line with the research request we cover the domains of (1) datafication, (2) automation, (3) networking and (4) human-machine interaction (see Illustration 1: Trade-offs and Digital Ethics, Appendix and the Supplement for a more detailed discussion).

Analysis

Bulletpoint blue

Datafication

The strongest arguments that favor the process of datafication in our analysis are: (1) a direct communication between the guest and hotel based on CRM-data allowing for high quality service, (2) subtle collection of customer data without impairing the hotel guest and (3) a holistic analysis of data and datafication of various guest’s properties. On the other hand, following respective disadvantages accrue: (1) abuse of personal data collected, (2) feeling of being spied on/monitored, (3) loosing of human-centricity due to a perceived data obsession and guest’s dehumanization.

ET Bulletpoints blue 2

Automation and algorithmic Decisions

Within the domain of automation, the strongest advantages are: (1) GDPR compliant access to data by machines, (2) circumvention of bad leisure time choices and (3) higher client satisfaction due to personalization. On the downside we were able to locate the following points: (1) quality of digital service might suffer from overly strict private settings, (2) machines make decisions about how guests are to spend time depriving them from self-determination and (3) creation of misaligned profiles.

Bulletpoints blue 3

Networking and Virtualization

Within the domain of networking and virtualization the advantageous points are: (1) other travel suppliers and vendors optimize their offering according to hotel bookings, (2) unified user experience and (3) a smart hotel that increases security (via IoT). On the contrary, the following arguments disclose the issues: (1) travel suppliers leverage the information against the guest by means of predictive pricing, (2) lack of control of data in networks, (3) complex data security and updates to prevent wrong analysis.

Bulletpoints blue 4

Human-Machine Interaction

Coming full circle of the analysis, the human-machine interaction raises some points: (1) correct intent and context-recognition by machine, (2) support of the hotel staff and the guests by E-Butler, (3) less hotel staff is needed which is good from the economic perspective, alternatively staff is able to provide more services to the guests. On the contrary we perceive the following pain points within the communication domain: (1) strong impact on opinion and triggering by machine, (2) ”impersonal” communication and (3) possibility to become addicted.

Stakeholder Analysis

By the implementation of an E-Butler the full-circle reaches from the hotel guest using it, the service personnel applying CRM knowledge, the Hotel IT securing the conversation, the society or public experiencing the effects of job rationalization and the law makers balancing all negative side effects such as job substitution and data security concerns with reskilling programs and data legislature. For the hotel guest it is important to have an easy and intuitive communication as well as a high service level including smart recommendations. For the service personnel the usage of the CRM data makes it easier to provide a high service level. The hotel owners and managers in turn can safe i.a. labor costs. The society will need reskilling while law makers will try to balance all negative side effects with legislature.

Stakeholderanalysis

Corporate Digital Responsibility

In line with a corporate digital responsibility Jumeirah Hotels and Resorts need to pay attention to the negative side-effects that the deployment of the E-Butler may create. Those being (1) job rationalization, (2) security threats, (3) short-term service decline, (4) ethical and social biases within the conversation of the E-Butler and (5) the GDPR legislation and the law risks that may entail costly penalties. Solutions can be found via (1) up- or reskilling of their employees via programs, (2) data security plans and qualified personnel, (3) well-advised project planning for the deployment of the E-Butler in new domains and an intuitive setup of the conversation with the E-Butler, (4) data training of the E-Butler to circumvent ethnically prejudiced statements and (5) a compliant setup. Paying attention to the ethical trade-offs mentioned above will sensitize the management of Jumeirah and bear the biggest rewards for the guests and the personnel from this technology such as (1) the improvement of processes and assisting the hotel staff via digital services and (2) the innovation of the overall experience of hotel customers via digital services for long-term success.

Business Application

Within this chapter we analyze the measures that have to be undertaken to reap the greatest benefits from the deployment of the E-Butler bearing in mind that luxury suppliers increasingly offer digital services and integrate those in their general strategy.

Bulletpoint blue

Hotel Staff

In order for the E-Butler to supplement the service offer optimally and not to be perceived as substitution for humans, the deployment of the E-Butler needs to be context driven and accompanied with picture analysis. This way round the Butler takes more and more the role of an empathetic person. Also, the hotel staff should be trained for the ethical application of the E-Butler, (i) on how to introduce the E-Butler to the clients, (ii) how to use the data gathered by the Butler and (iii) be trained for data security and GDPR.

ET Bulletpoints blue 2

Guests

Hotel guests should be prepared by (i) the hotel personnel for the application of the E-Butler via a sensitization regarding their privacy and the notification that they will be speaking to a digital assistant, (ii) it should always remain an option to talk to a real person with no differentiating pricing or additional cost. Otherwise guests will be incentivized to discriminate against human work. Hotel guests should be enabled to work with the E-Butler in an informed and sovereign way via a briefing on the possibilities of the E-Butler. The E-Butler could start with a very simple use-case of personalization so that the first interaction is a positive one e.g. the Butler asks the hotel guest whether the guest wants a coffee in a suitable moment to build positive emotions. The first interaction is the most important one as it determines the future adoptability of this technology strongly. Thereby, being a strong risk moment determining the success of implementation.

Outlook

All in all, we see that the application of the E-Butler brings about manifold positive effects such as (1) increase in service level, (2) increase in the overall customer satisfaction, (3) increase in competitiveness of the hotel, (4) seamless user experience for the guest and (5) upskilling opportunities for personnel due to coverage of repetitive tasks by the E-Butler. On the other hand, the society is afraid of (1) job losses, (2) security threats, (3) loss of empathy and emotional connection between people, (4) dependence on digital systems only and (5) being confronted with behavioral and social biases based on the data fed into the system. We have the impression that the deployment of this system will take place on a wide array across industries. We are now at the determining point on how this development will take place while a european interpretation of those systems requires knowledge of the different cultures and countries.

Framework Application

Bulletpoint blue

Across Domains

Within this part we introduce the framework proposed by Müller & Andersen (2017) to discuss the ethical trade-offs related to the application of E-Butlers or digital service assistants by summarizing the single trade-offs across the domains (1) datafication, (2) automation, (3) networking and virtualization and (4) human-machine interaction. 

Datafication

With datafication of any information in the hotel domain we see the following advantages: (i) a direct communication between guests and hotel based on CRM-data allowing for high quality service, (ii) the creation of a personal profile that matches preferences for a perfect hotel stay, (iii) subtle collection of customer data that does not impair the hotel guest, (iv) secure access to all personal data by guests and hotel personnel, (v) no necessity to fill out forms for personalization, (vi) easy data collection process that is subtle, (vii) collection of as much data as possible to completion, (viii) participation in the digital society, (ix) holistic analysis of data and datafication of various fields and (x) improvement of onboarding of employees based on CRM data collected.

On the other hand following respective disadvantages accrue: (i) collection of private data of the user, (ii) access control or control violation of the E-Butler service, (iii) abuse of personal data collected, (iv) limitation of the access of the data, (v) threat to disclose very private information, (vi) no transparency of what kind of data a user gives, (vii) feeling of being interrogated, (viii) tiresome process of data collection, (ix) storage of personal information and (x) loosing of human centricity via rationalization.

Automation and algorithmic Decisions

Within the domain of automation, we analyzed the following factors, the advantages being: (i) better decisions for the client due to personalized algorithms, (ii) GDPR compliant access to data by machines, (iii) circumvention of bad leisure time choices, (iv) guests can be notified if there is an emergency or a critical event, (v) filtering of data and pre-classification for smart personalization, (vi) higher client satisfaction due to personalization, (vii) capacity optimization in terms of hotel booking and stay based on algorithms, (viii) big data advantage to increase prediction precision, (ix) increase in service level and comfort and (x) automation in processes leading to efficiency increases.

On the downside we were able to locate the following points: (i) giving all private data to a third party, (ii) quality of digital service suffers with strict private settings, (iii) machines make choices how guests can spend time precluding them from self-determination, (iv) analysis of guilt-free people and incidences, (v) circumvention of freedom of will, (vi) incentivization to cross-sell to the client, (vii) creation of misaligned profiles, (viii) struggle in personalization, (ix) categorization of humans via machines and (x) lack of personalization in luxury.

Networking and Virtualization

Within the domain of networking and virtualization the advantageous points are: (i) data on booking and communication of preferences are in line with GDPR, (ii) other travel suppliers and vendors optimize their offering according to the booking with the hotel, (iii) unified user experience, (iv) smart hotel that increases security (via IoT), (v) improvement of client usage, (vi) APIs for WhatsApp to improve UX, (vii) control over data, (viii) encryption of client data, (ix) improvement in user experience and (x) improvement in personalization due to network effects.

On the contrary, the following arguments disclose the issues: (i) protection of data as big concern, (ii) travel suppliers leverage the information against the guest by means of predictive pricing, (iii) lack of control over data where it is spread to, (iv) privacy violation, (v) 3rd party monitoring, (vi) lack of control for data in networks, (vii) spreading of data across the network and the 3rd party entities, (viii) widespread usage and datafication of information, (ix) complex data security and updates to prevent wrong analysis and (x) feeling of followed by data and lack of re-interpretation of new needs.

Human-Machine Interaction

Coming full circle in the analysis the human-machine interaction presents the following points to be considered: (i) correct intent and context-recognition by machine, (ii) greater access to data enables the machine to work better, (iii) high positive impact on the planning of journeys and preferences, (iv) machine knows customer perfectly well, (v) support of the hotel staff and the guests by the E-Butler, (vi) via often interaction with digital assistant the conversation improves and reflects own needs, (vii) support of jobs and human dignity, (viii) perfect unified experience for customers, (ix) less hotel staff is needed which is good from the economic perspective and (x) hotel guests are more satisfied with the hotel experience.

On the contrary we perceive the following pain points within the communication domain: (i) giving up all information to be potentially misread by machine, (ii) loss of privacy, (iii) strong impact on opinion and triggering by machine, (iv) necessity to secure analytical knowledge from 3rd parties, (v) ”impersonal” communication, (vi) possibility to become addicted, (vii) decrease of human induced mistakes but also a human touch, (viii) individual needs in the handling as customer that are not coverable by a machine, (ix) loosing luxury hotel quality based on personal service and (x) long-term brand image damage if E-Butler does not work.

ET Bulletpoints blue 2

Across Trade-Offs

Within this part we introduce the framework proposed by Müller & Andersen (2017) to discuss the ethical trade-offs related to the application of E-Butlers or digital service assistants by analyzing the single trade-offs across the domains (1) datafication, (2) automation, (3) networking and virtualization and (4) human-machine interaction.

Access vs. Privacy

Within datafication, data on movement and any behaviorism are collected, which makes the customer transparent and prone to man in the middle attacks. Going further with automation violations in privacy can entail a wrong automatism within platforms regarding a person’s identity, followed by misled algorithmic decision-making. Furthermore, with the network character of platforms those violations can have serious effects if they are not detected. Misinformation is spread all over the networks leading to wrong outcomes in human-machine interactions that could be life-threatening. In the hospitality domain this could lead to food served to a customer to which the person is allergic.

Personalization vs. Privacy

Within the digital domain, customers release behaviorisms or data that links back to their psychometric profile. Whether they do it consciously or sub-consciously, data is released that gives corporates insights on a granular level about the customers. The related analyses can be of a causal nature or a non-causal one depending on the information provided. As not every search or utterance is related to us personally, the analyzed recommendation may not be correct as the intent behind the question may be not be related to the person requiring a specific service. With wrong analyses the whole machine learning system behind an E-Butler becomes un-useful for the user and the firm deploying it. So, the risk entailed in personalization when it comes to a hospitality case is quite high. Instead of serving the customer, the digital assistant can be misleading which requires human moderation to prevent. Especially in the luxury domain this becomes an important aspect where customers are used to exquisite service and a high degree of empathy and privacy.

Usability vs. Proportion in Decision making

An E-Butler can increase the usability of services for a hotel guest and create an immersive experience. However, on the expense of individual decision making. Using assistance systems to make decisions can lead to unlearning of own decision making which could have good and bad effects. With digital assistants being inert and adverse to flexibility, changes in preferences can create a problem for the E-Butler host.

Security vs. Privacy

Governments or security institutions can help to spot and analyze personal data that may be related to fraud or criminal activities. Within the hospitality sector this may lead to predictions about the guests to arrive. Algorithms can then form prejudiced decisions based on big data evaluations that are socio-economically impacted. In hospitality this may entail the mistreatment of guests based on these biases leaving them unsatisfied or even insulted. Also, hospitality is known and valued for its neutrality which stands in contrast with the biases related to socio-economic evaluations.

Client usage vs. Privacy

For the guests in hospitality, the planning of the visit is important, the check-in and the navigation during the stay as well. For this, applications on the mobile phone are handy such as E-Butlers. They can point to suitable options and activities for the consumer. While these services if applied correctly are useful, they can lead to isolation of the customer to only this channel, to obtain information about the stay. This in turn can lead to the decrease of brand identity associated with excellent and personal service of hotels. 

Client satisfaction vs. Freedom of will

With personalization in digital and the adaptation of services, client satisfaction can increase and lead to higher sales for the company offering personalized services. However, at the expense of a free will. Also, digital assistants can influence the interaction with clients and the general, public opinion. Brands deploying any kinds of bots need therefore the insurance that those will act in a specific, ethical and PR-regulated way. Otherwise, the application of assistance systems or E-Butlers can be at the expense of a good reputation.

Authors

Marwane El Kharbili, Ph.D. - Associate Director Digital Strategy at Gartner

Sofia Picture

Sofia Trojanowska - Founder and Digital Consultant at Singularity.Design

Patryk Dumicz - Senior Consultant at Capture Europe

Sources

  • Christian Montag & Peter Walla | Monika Koller (Reviewing Editor) (2016): Carpe diem instead of losing your social mind: Beyond digital addiction and why we all suffer from digital overuse, Cogent Psychology, 3:1, DOI: 1080/23311908.2016.1157281
  • Deloitte (2019): Fashion & Luxury Private Equity and Investors Survey 2019 Global report.
  • Goodman, B. and Seth Flaxman (2016): EU regulations on algorithmic decision making and a “right to explanation”. CoRR, abs/1606.08813, 2016
  • La Maison de Startups (2019): https://lamaisondesstartups.lvmh.com/en/, accessed on 1.12.2019 12:35.
  • Marco Túlio Ribeiro, Sameer Singh, Carlos Guestrin: “Why Should I Trust You?” (2016): Explaining the Prediction of Any Classifier. KDD 2016: 1135-1144.
  • McKinsey & Company (2019): China Luxury Report 2019 – How young Chinese consumers are reshaping global luxury. McKinsey Greater China’s Apparel, Fashion and Luxury Group.
  • Mnih, V. Kavukcuoglu, L. Silver, D., Rusu, A. A., Veness, J., Bellemare, M.G., … &Petersen, S. (2015): Human-level control through deep reinforcement learning. Nature, 518(7549), 529-533.
  • Müller, Lena-Sophie & Andersen , Nicolai (2017): Denkimpuls Digital Ethik: Warum wir uns mit digitaler Ethik beschäftigen sollten – Ein Denkmuster, Arbeitsgruppe Ethik | Stand: 06. November 2017 | Version: 01.