Case Study

Vail Resorts Digital Mountain Assistant

Project type: 

Chatbot

Year:

2016 – 2018

emma@2x 3.png
 

Project Overview

Vail Resorts approached us to create a digital assistant to help their customers maximize the amount of time they spend enjoying the resorts experience and mountain activities across nine mountain resorts in North America. R/GA was awarded to work on this project when we were actively developing our brand new Branded AI practice, and I was cast to help define the architecture of the conversation-based digital service for the client while defining the design process for ourselves.

     

    My Role in this Project

    • Defining workflow for a conversation-based digital service design
    • Field study and creating a report
    • Leading feature prioritization
    • Defining the scope of knowledge
    • Creating design framework for conversations
     

    KEY CHALLENGE

    How do we create one digital assistant that can act like a local across all nine mountain resorts?

    Because Vail Resorts owns a portfolio of mountain resorts across North America, we needed to make sure our digital assistant act like a local expert across nine mountains. To efficiently approach this problem, I designed a set of conversation templates that represent the brand personality and can be used to plug in mountain specific knowledge via web-hooks.

     

    STEP ONE

    Discovery

    We kicked off the discovery phase by visiting Vail and Breckenridge to conduct an experience audit. We asked one of the Vail resort employees to act as a digital assistant via Facebook Messenger while we were at the mountain resorts and asked him whenever we come across questions.

    Discovery phase deliverable (Tumblr Site)

    FireShot+Capture+11+-+Vail+Resorts+on+Mountain+Research_+-+https___vail-mountain-research.tum.png
    FireShot+Capture+10+-+Vail+Resorts+on+Mountain+Research_+-+https___vail-mountain-research.tum.png

    Because each of four R/GA member was documenting their experience on their own, at the end of 2-day research we had very scattered notes, photos, and screenshots. After seeing the number of materials we generated from the research, I decided to create a custom Tumblr page to consolidate everyone's experience into a chronological timeline. 

    See the Tumblr page 
    (password: rgainvail)

     

    STEP TWO

    Definition

    The Scope of Knowledge

    Based on the insights we uncovered from the discovery phase, we defined the scope of knowledge of the digital assistant. We used a mind-map building tool Coggle.it to visualize the tree of knowledge we needed to be able to answer to make our assistant feels smart.

    The initial draft of the mind map: It started off as an exhaustive list of topics.

    The initial draft of the mind map:
    It started off as an exhaustive list of topics.

    Mindmap after a few iterations: We were able to find a set of condensed topics that are relevant to each mountain.

    Mindmap after a few iterations:
    We were able to find a set of condensed topics that are relevant to each mountain.

     

    Conversation Template

    Next, I created a document that describes how to construct conversations for each major node that I defined within the mindmap. It included a list of invocation phrases, domain knowledge, content keys, and skeleton of the conversation to illustrate how the template can be applied to each mountain.

    Conversation Spec - Condition

    Conversation Spec - Condition

    Conversation Spec - Resort Transportation

    Conversation Spec - Resort Transportation

    Conversation Spec - Apres 01

    Conversation Spec - Apres 01

    Conversation Spec - Apres 02

    Conversation Spec - Apres 02

     

    STEP THREE

    Development

    Once what our system needed to know to support the experience became clear, I set up worksheets to capture all necessary information. Unfortunately, our client did not have a single point of reference that we could tap into for this process, I worked with Vail Resorts' business analysts to read through each mountain's digitals channels and developed knowledge base manually.

    Knowledge Base Document

    Knowledge Base Document

    Lastly, I converted the conversation spec document into copy worksheet.

    Copy worksheet - 01

    Copy worksheet - 01

    Copy worksheet - 02

    Copy worksheet - 02

     

    Conclusion

    This project gave me an excellent opportunity to reflect on what I have learned as an experience designer over the years and applied it to something that I have not done before. Each step of the project forced me to reach out to other disciplines and find out what information they need and how they want them to do their job better. 

    The process of not using wireframe or screen mock was a significant obstacle initially; however, it taught me how aligning priority of the content with user needs could fulfill the majority of user needs and provide a meaningful experience without relying solely on visual user interfaces.

     

    Thank you so much for reading.