CONVERSATIONAL UI
Transforming Travel Experiences: Crafting a Conversational AI Travel Chatbot
Embark on a journey of redefining travel experiences with our Conversational AI-powered Travel Companion. This case study delves into the transformative impact of crafting a user-friendly virtual assistant that enhances every aspect of the user's travel journey.
Role
Timeline
Team
Conversational Designer
1 Month ( Feb 2024)
Sahith Uppala
Sindhura Arepalli
Core Responsibilities
Designing natural and engaging dialogues, creating adaptable conversational flows, integrating user preferences into personalized itineraries, and iterating based on user feedback.
Want to skim through case study in 1 Min?
Here is the quick summary of the entire case study
🤩What I did?
Crafted conversational dialogues for personalized trip planning and travel journaling, ensuring a seamless and engaging interaction.
🖋️What did I learn?
Iterative development based on user feedback is key; a balance of conciseness and informativeness is crucial in conversational design.
⁉️Why did I do it?
To create a user-friendly Travel Companion, enhancing the travel planning and journaling experiences through natural language interactions.
🤖What was the Challenge?
The challenge lies in striking the delicate balance between technological sophistication and human-centric design. Users don't want to interact with a robot; they crave a natural, conversational experience
INTRODUCTION
Travel Companion Chatbot is an innovative solution that revolutionizes the way users plan and document their travel journeys. Seamlessly blending text and voice interactions
the chatbot serves as a virtual travel assistant, guiding users through personalized trip planning and facilitating voice-based travel journaling. The goal was to enhance the user experience at every stage of the travel journey, from initial planning to on-the-go support.
PROBLEM
How might we make travel planning and journaling more personalized and user-friendly, overcoming current limitations and meeting the diverse needs of users?
Users often face challenges in planning personalized trips and maintaining travel journals, encountering friction in the process. Existing solutions lacked the natural flow and personal touch needed for an immersive user experience. Recognizing these pain points, our project aimed to address the gaps in user engagement, personalization, and accessibility within the realm of travel assistance.
INITIAL RESEARCH
How do we know if it's a problem?
1. Leveraging User Interviews :
We interviewed 10 people to know more about their experiences and pain point while planning a trip/vacation
1. Overwhelm with Options
60% of users expressed feeling overwhelmed by the multitude of choices during trip planning.
"There are so many options online; it's overwhelming. I end up spending hours just browsing without making a decision."
2. Lack of Personalization
50% of users highlighted the lack of personalized suggestions as a significant pain point.
"I wish there was a way to get recommendations based on my preferences. Everything seems one-size-fits-all."
3. Information Scattered Across Platforms
40% of users found the dispersion of travel information across platforms inconvenient.
"I have to check multiple websites for flights, hotels, and activities. It's time-consuming and frustrating."
4. Uncertain Budget Planning
80% of users struggled with uncertain budget planning due to unclear pricing structures.
"It's hard to plan when I don't have a clear idea of the total cost. Hidden fees and unclear pricing are major issues."
5. Lack of Interactive Assistance
70% of users expressed a desire for interactive assistance in the trip planning process.
"I wish there was someone or something interactive that understands my preferences and guides me through the process."
Who are our users?
Our potential users fall into the following categories
Busy Business Traveler
Alex, 35
Sales Executive
Always on the go, values efficiency, needs quick and seamless planning, prefers work-friendly accommodations.
Needs
-
Fast and reliable trip planning.
-
Access to key amenities like Wi-Fi and proximity to business locations.
-
Time-saving features for a hassle-free experience.
Adventure-Seeking Explorer
Bob, 28
Freelance Photographer
Loves exploration, seeks unique experiences, flexible with dates, interested in local cultures and offbeat destinations.
Needs
-
Personalized recommendations for unique activities and hidden gems.
-
Flexible date suggestions for budget optimization.
Family Vacation Planner
Chris, 40
Parent
Plans family vacations, considers budget constraints, needs family-friendly accommodations and activities.
Needs
-
Comprehensive budget planning tools.
-
Family-friendly accommodation suggestions.
-
Itineraries with a mix of adult and child-friendly activities.
2. Domain Study
We decided to spend some time understanding about how trips/vacations usually get planned and what type of tools and websites are used
-
Overview: Users commonly use travel platforms like Expedia, TripAdvisor, and Booking.com. These platforms offer a range of features, including flight and hotel bookings, activity recommendations, and user reviews.
-
How They Work:
-
Flight Booking: Users input departure and arrival locations and dates, browse available flights, select preferred options, and make bookings.
-
Hotel Booking: Users enter their destination and dates, view available accommodations, filter based on preferences, and book.
-
Activity Recommendations: Platforms suggest activities based on user preferences, ratings, and location.
-
-
Key Features:
-
User reviews and ratings for informed decision-making.
-
Package deals bundling flights, hotels, and activities.
-
Integration with maps for location awareness.
-
Intial Flow
Challenges:
-
Complexity: Users navigate multiple tabs for flights, hotels, and activities.
-
Information Overload: Copious options lead to decision fatigue.
-
Back & Forth: Frequent switching between platforms for comprehensive planning.
IDEATION
Defining the Ideal Bot
An ideal chatbot should seamlessly blend efficiency with a human touch, offering a user-friendly experience.
The chatbot should be adept at guiding users through tasks, providing relevant information, and adapting to diverse conversational styles. A well-designed chatbot anticipates user needs, offering proactive suggestions and demonstrating empathy in its responses.
Human Conversation
=
Human-like conversation by bot learning emotional Intelligence
1. Natural Conversation Framework
This step involves creating conversation flows for different aspects. Initially we focused on 2 major tasks that we want the CUI to perform
1. Help with personalized trip planning 2. Help user maintain a travel journal by giving voice-based input.
Task 1: Help with personalized trip planning
Task 2: Help user maintain a travel journal by giving voice-based input.
Intent: Travel - Recommendations
Task 1: Help with personalized trip planning
Intent: Travel-Planning
Entities: Destination, Budget, Interests, Accommodation, Dates
Context: The user initiates a request for personalized trip planning.
Response: The system acknowledges the request and starts asking about destination preferences.
Base action pair: It establishes the primary user actions that trigger the system's responses, forming the foundation for the conversation.
Expanding pair: Expanding pairs allow for flexibility by introducing variations in user input and system responses, making the conversation more dynamic and accommodating.
Classes organize actions: Classes group related actions together, making it easier to manage the flow of conversation and handle different aspects of the user's request distinctly.
Travel-Planning Expansions: Expansions add depth to the conversation by providing additional information, suggestions, or options, enhancing the overall user experience.
2. Utterance Framing
Utterances are crafted to handle different ways users might express their intent. Variations consider synonyms, different sentence structures, and user preferences.
By defining the intent and entities, the system can understand the user's primary goal (planning a trip to Paris) and respond accordingly. Crafting varied utterances ensures the system's robustness in recognizing user input diversity, making the conversation more natural and accommodating.
Sample Conversation
BUILD
Building an Ideal Bot Personality
Three different personalities were defined according to the four primary tone-of-voice dimensions.
Joe
Casual, Serious, Respectful
Joe has a friendly and relaxed tone, making users feel at ease. Joe adopts a serious tone when dealing with important matters. Respect is key for Joe. He addresses users politely, acknowledges their preferences, and values their opinions. Joe provides information straightforwardly, focusing on clarity and precision.
May
Formal, Funny, Respectful
May maintains a formal tone, ensuring a professional and polished interaction. She follows established norms of communication. May inject humor appropriately into conversation. Respect is fundamental for May. May expresses enthusiasm, making interactions lively and engaging. She conveys excitement about assisting users
Eva
Formal, Serious, Respectful
Eva adopts a formal tone, emphasizing professionalism in all interactions. Eva maintains a serious tone when dealing with important topics, conveying a sense of gravity and importance. Respect is a cornerstone for Eva. She treats users with courtesy, values their opinions. Eva communicates in a matter-of-fact manner, providing clear and concise information without unnecessary embellishments
Sample Conversation of Chatbot - 1 (Joe)
User: Hi Joe! I need help planning a trip.
Joe: Hey there! Absolutely, I'm here to help. Where are you thinking of heading?
User: I'm not sure yet, maybe somewhere with mountains.
Joe: Cool choice! Mountains offer a great escape. Any specific activities you're into?
User: Hiking and maybe some photography.
Joe: Nice combo! I'll find destinations with stunning trails and picturesque spots. What's your budget range?
User: I'd say mid-range, not too fancy but not camping either.
Joe: Got it. Mid-range it is! What dates are you considering for this adventure?
User: Flexible on dates, just want a memorable experience.
Joe: Sweet! Flexibility opens up exciting possibilities. Any accommodation preferences?
User: Something cozy, like a cabin or a small lodge.
Joe: Cozy it is! I'll put together a tailored itinerary with mountain hikes, photogenic spots, and comfy stays. Sound good?
User: Perfect! Thanks, Joe.
Joe: No problemo! If you have more preferences or any changes, just let me know. Happy travels!
Similarly, sample conversations of chatbot 2 & 3 were created and tested out.
TESTING
Testing out Chatbot versions
Wizard of Oz testing was done on a existing chatting platform like whatsapp, As the goal was to only understand the tonality preferences
5 users were asked to assess the chatbot on a 5 point scale regarding
1. Professionalism - 5 Points
2. Trustworthiness - 5 Points
3. Willingness to assist - 5 Points
Chatbot 2 came out to be on the top with a score of 13/15
The Final tone hence decided: Formal, Funny, Respectful & Enthusiastic
WIREFRAMING
Prototyping in Voiceflow
Voiceflow is a highly valuable tool in the creation of Conversational User Interfaces (CUIs). It offers an intuitive and user-friendly platform that enables designers and developers to build, prototype, and iterate voice and chat-based applications seamlessly.
1. Onboarding
2. Error Handling
3. Feedback
Onboarding
Onboarding in CUI is a critical phase that shapes the user's initial experience, fosters engagement, and sets the tone for the entire interaction. It acts as a bridge between the user and the chatbot, ensuring a positive and user-friendly start to the conversational journey.
Onboarding script for two different scenarios
Error Handling
Users interact through natural language, which introduces complexity as users might provide unexpected inputs or encounter issues during the conversation. Proper error handling helps maintain a smooth and engaging interaction, preventing frustration and confusion.
Examples of error inputs
1. Misunderstood Intent
2. Out-of-Scope Requests
3. Ambiguous Requests
4. Repeated or Long Queries
Bot doesn't have related information
Bot doesn’t understand the information
Feedback
Feedback serves as a valuable source of insights into user experiences, helping identify areas of improvement, uncovering potential issues, and gauging user preferences. In the dynamic realm of natural language interactions, user feedback aids in refining language models, understanding colloquialisms, and adapting to diverse user communication styles. It also provides an avenue for users to express their needs, allowing designers and developers to iteratively enhance the CUI
CONCLUSION
What did I learn?
Throughout the project, the iterative design process played a pivotal role. Feedback from users provided invaluable insights into pain points and preferences during travel planning. This iterative approach not only refined the chatbot's functionality but also highlighted the dynamic nature of user interactions.
One notable learning was the importance of user personas and the diverse expectations users bring to a travel companion chatbot. From meticulous trip planners to spontaneous adventurers, tailoring the bot's responses to cater to varied preferences is crucial.
Biggest Challenge
The biggest challenge in working on a chatbot project lies in striking the delicate balance between technological sophistication and human-centric design. Building an effective chatbot is not just about coding the right algorithms; it's about understanding the nuances of human communication, emotions, and context. Users don't want to interact with a robot; they crave a natural, conversational experience.