Ambient Service on a Package Holiday

Project Aims

This project aims to discover how the use of a chatbot based mobile application together with a smart speaker skill can provide ambient customer service on a package holiday.

TUI is the focus stakeholder on account of their unique position of managing both an airline and hotels. I am not connected to or affiliated with TUI Group AG.

Problem

Can technology improve the customer experience on a traditional sun and sand get away?

Research

What is Ambient Service?

Imagine yourself on the first evening in a hotel room, looking at your bed it becomes apparent that the two pillows provided are not going to give a restful sleep. Thankfully, after a quick telephone call to the reception, an extra pillow is delivered to your door. One would hope that by now, you are a satisfied customer.

Fast forward another year, and you’re in a hotel provided by the same company but in a different city. Settling down for the night, you inspect the bedding arrangements and prepare yourself for a conversation with reception when you notice that on the bed, there’s not two, but three pillows are already waiting for you. 

The hotel chain has remembered your preference and made sure that you didn’t have to ask for it again, the service was calm and quiet, there was no interaction, it just happened. This is ambient service, providing help and assistance before you even think to ask.

Project Process - Design Thinking.

This project achieved its aims by using the respected method of design thinking. It’s defined, and flexible framework for learning about the customer, identifying their needs and then testing possible solutions is a logical option for undertaking a project such as this. 

State of the Art.

Of interest is the relatively recent development by the airline KLM to offer boarding passes through the app WhatsApp. Similarly, a hotel-casino in China has partnered with the instant messaging service WeChat to provide room controls such as lights and heating through their app.

The concept of offering help through such a chat service seems like a logical step in delivering an ambient service but is not without notable difficulties.  What if the user doesn’t subscribe to such a service? How much would your desired function be constrained by the interface of said app? Is it ethical to use a third party to handle the potentially sensitive data required to remember a customer's requests while in your care?

Smart speakers also assist in chat based interaction. Heavily marketed by the tech giants Amazon and Google, they have become an increasingly popular household item. They could provide an accessible alternative to customers with vision difficulties.

Value Proposition.

A value proposition was created to define the aims and scope of further research. The process revealed that customers who use technology to seek help in their lives at home would provide the best fit and that the app should provide a one-stop shop for both providing information and resolve enquiries.

Customer Interviews.

Three people who have had a package holiday were interviewed about their experience. Transcripts of the interviews we taken and then thematically coded to identify any trends. All seemed to back up the published research into why people chose package holidays, perceived value for money, precise holiday spend and ease of booking (one stop shop for flights and hotels). 

Interviewees felt that the free transfers were a pain point, while they liked the fact they were included, they didn’t feel they were an enjoyable part of the experience.

Hotel receptions were seen as a good place to get help, but the language barrier was a concern to some of those interviewed. One user remarked that when asking reception for a feather pillow “… you have to do bird impression”. Two of the three participants preferred to walk down to reception to make requests for their room rather than use the phone provided because they felt they would have to explain more to have their request fulfilled.

Asked if they used technology while on holiday, all said that they did. Mostly to keep in touch with friends and family back at home and find out information about the local area such as restaurant reviews, attractions and transport information. 

Experience Map.

An experience map was created from the current journey that customers undertake.  It highlights the pain points but also where TUI is most helpful. Notably, there’s a lot of queuing for help particularly in the early stages of the holiday such as airport arrivals and check-in. All of which could be eased with technology by providing information just before the event occurs.

Personas.

Four personas were created to synthesise the information gained in the empathise section. They cover the most populous demographics TUI currently serves, people aged over 50 and young families. 

Empathy Map.

Producing an empathy map provided insight into the experience of a person on a package holiday and the reasons for being there.  

Brainstorm.

Firstly, using the customer journey and testimony from interviewees, I mapped out the points of a holiday that people might seek assistance.

The five why’s technique was used to delve deeper into why people book a package holiday. Doing this backed up previous discoveries like needing a break from work and time with family.

By brainstorming, I was able to clarify what needed to be designed, chatbots and smart speakers being the preferred choice because they’re both available now and recognised by the many customers.

Storyboards.

Storyboards provide a way of visualising small parts of the customer journey and how possible solutions generated might help them achieve their goals.

Design

 The Chatbot.

The chatbot forms the basis of the interactions between the company and customer. Both push and pull assistance would be performed in this function. The idea is that customers would be on-boarded when either booking a holiday through the app or downloading it to check in for their flight and receive boarding passes.

The design follows the TUI brand guidelines for colour, typography (TUI 2018). 

Design Issue - How casual is casual?

Typically, service representatives of an airline or hotel would formally address a customer. Would this apply to a chatbot too?

To test how what the chatbots tone of voice should be, a low fidelity chat prototype was created with two scenarios, one with a formal tone of voice and the second being casual. Participants were asked to imagine themselves in a hotel and then ask for an item to be delivered.

In both scenarios, the participants responded well to both styles of interaction, with no apparent preference to either it was decided that customers should be asked how they would like to be addressed, tailoring subsequent responses.

A persona was created to provide a reference for the tone of voice used by the chatbot. The general idea was that it should be someone warm and friendly that was also knowledgeable and efficient.

Design Issue - Chat Flow.

Inspired by the storyboards a basic script was written to plan for what kind of conversations would take place in the early stages of a holiday. These were: Checking in, airport arrivals, transfers, asking for information, requesting an item and booking a taxi while away from the hotel. 

Aside from cards during the design of the script, the conversation could be directed using buttons and auto-suggestions. By doing this, we can reduce the number of utterances that the system would need to know.

Alexa Skill.

The work produced so far on the chatbot would benefit the development of the skill as they’re both chat based functions. However, we won’t be able to use auto-suggestions or buttons to direct users to the correct path. Therefore, the different ways people may ask for help or utterances must be studied. 

Design Issue - Planning for Utterances.

The format for interaction is displayed below. 

The intent is the goal of the interaction, what does the user want to achieve.

An utterance is how the user may ask for help. 

The slot value is the specific information we need to fulfil the request. Here is an example to illustrate how this would work with our project.

[ Alexa wake word], [ask launch] [TUI skill] [for a utterance] [razor slot value]

The slot value of razor would activate the following intent.

[Housekeeping deliver to room 365 intent] [one razor slot value]

Now, it would be great if Alexa worked this quickly, but at the time of writing, she’s not quite this good at recognising intents from third-party developers.

To combat this, when the intent isn’t clear we may ask for clarification like this.

[I can order intent verification] [a razor slot value] for you, [is that correct? user prompt]

Planning for all utterances is outside of the scope of this project, a prototype was built with the best case scenario in mind. This section mainly serves as a guide to how Alexa would integrate with the service plan and app.

Test.

Situational Test.

To test the chat function, Skype was chosen as a medium for interaction with users. This would allow the tester to respond to and learn from requests that were not yet programmed into a chatbot.

A basic script for the scenario was created, taking into account the tone of voice of the chatbot persona Emma. Also, each user would be asked whether they preferred to be called Sir or Madam, Mr or Ms <last name> or <first name>. If a user selected <first name>, then emojis would be added to the conversation.

Secondly, a Skype account was created and branded with TUI insignia to immerse the user and provide a visual reference throughout the interaction. At the start of each test, participants were given a written as well as a verbal explanation of the test this was followed by five tasks.

Conclusion

Chat based interaction has been identified as a viable method for assisting customers on holiday. With no human interaction, the fear of asking is removed which could allow a company to reach those who would typically suffer in silence. 

The benefits of this service go beyond only providing help to the customer. By analysing the data of these small interactions, TUI would be able to examine and quickly adapt the product it offers to meet the needs of a changing market. With preferences saved for subsequent trips, customer loyalty could be increased.