Meet Case
Conversation Design | Chatbot | Product Design
We listen to music on several platforms, therefore no one platform has enough data to provide relevant recommendations, making it more difficult to find new music.
We listen to music on several platforms, therefore no one platform has enough data to provide relevant recommendations, making it more difficult to find new music.
Team
Just me! 🤭
Just me! 🤭
Role
UX Writing, UX Research, Usability Tests, Flow Charts, Personas, Sample Scripts, Prototypes
UX Writing, UX Research, Usability Tests, Flow Charts, Personas, Sample Scripts, Prototypes
Impact
1. Recognized use of music to boost positive emotions and created an experience to enhance the effects.
2. Designed a prototype for discovering new music to counter lousy outcomes.
1. Recognized use of music to boost positive emotions and created an experience to enhance the effects.
2. Designed a prototype for discovering new music to counter lousy outcomes.
Tools
Figma, Miro, LucidChart, Google forms, Voiceflow
Figma, Miro, LucidChart, Google forms, Voiceflow
My survey focused on listening habits, how people find new music and if they'd be interested in chatting with a bot that curates music for them. There were 99 respondents in all!
First significant result is that 71% of them listen to music at home, and one-third of them listen to music for 3 hours or more every day. The survey's emotional component yielded some intriguing results. Following are the emotional motivations for listening to music (in order):
I also obtained information on the streaming platforms on which respondents have accounts vs which they actually use in order to have a better understanding of the popularity and usage. 80 percent of respondents have a Spotify account, and 63 percent utilize it on a regular basis. Something I found intriguing was that while 24 percent of them had accounts on Amazon Music and Soundcloud, just 5 percent of them utilize them on a daily basis!
68% of the respondants rely on their friends to recommend new music to them. Following were the responses when asked how difficult it was for them to discover new music.
68% of the respondants rely on their friends to recommend new music to them. Following were the responses when asked how difficult it was for them to discover new music.
The majority of the secondary research was done to learn how music was leveraged to boost optimistic emotions. I looked at the influence of COVID-19 in particular. For this project I needed to find out how this current situation changes the use of music and technology, also how people perceive its importance.
A habit has evolved, and users now spend their whole day at home scrolling on their phones. With such a surge in technology use and reliance on the internet. While this may appear to be a bad statistic, I would like to believe that this sense of dependency may be used constructively with a chatbot companion. Users might utilize their chatbot to get into a flow while avoiding or limiting the negative sensations of frustration, guilt, and boredom.
Say hello to music's first chatbot, created in 2001. GooglyMinotaur gave fans real-time information on Radiohead, hidden Radiohead trivia, played music on demand, and answered topics like "When is Radiohead playing in New York?" and "Where can I hear the new album?" He'd also challenge you to a round of hangman.
And then there were others...
Record Bird’s chatbot aims to keep keen music fans up-to-date on new releases, as they’re announced.
Part of the marketing campaign for the British band’s (Bastille) new album
Loudie- concert recommendations chatbot: a spin-off from the app of the same name
AudioShot is almost Shazam-like with its ability to identify songs recorded using your microphone
Dance star Hardwell’s official bot isn’t just for telling fans about new releases and live dates.
Rithm started life as a music messaging app, pivoted into a startup marrying music and messaging for clients in both of those industries.
FooBot- Fans can even type ‘Ride with Foo’ to connect their Uber account
Christina Milian- This is kinda like an app but inside Messenger
But mostly on Facebook messenger or kik...
The above- mentioned bots give a lot of info on music and artists. The data has been curated and structured so users can learn more about the artists. These bots solely offer chat sessions and are generally found on social media platforms. It's overwhelming to interact with various bots that are specific to each artist, give info, and receive recommendations.
There should be a one-on-one communication medium; it would be a perfect approach to obtain accurate info. The apps ask users too many questions in order to deliver better user-centred recommendations. Excessive information needs, on the other hand, may complicate the entire user flow and make it harder to use the app. Giving the option of providing easy-to-understand music information and a one stop app for all things music would be more helpful.
So far, my study has shown that people experience worry and stress and use music to cope. This feeling is heightened right now because the present scenario and being compelled to work from home and be isolated. Developing an encouraging, honest, and helpful chatbot that will ideally help decrease stress and foster a productive atmosphere.
Interaction goals
Guiding/Instructive
Users would need to be guided through the many options, therefore the interaction would not be simple
Users would need to be guided through the many options, therefore the interaction would not be simple
Fun/Smart Mouth
The bot would be like a friend, and if the conversation is not entertaining, the purpose of motivating users to do this in the first place is defeated.
The bot would be like a friend, and if the conversation is not entertaining, the purpose of motivating users to do this in the first place is defeated.
Context aware/Personalized
The user's experience will vary depending on their needs and preferred selections.
The user's experience will vary depending on their needs and preferred selections.
What power does each party have?
The user is often invited to choose an item from a list and is occasionally directly requested to input what they are looking for.
The user is often invited to choose an item from a list and is occasionally directly requested to input what they are looking for.
How intimate does this relationship need to be?
Mostly because the suggestions are built on understanding the user on some level and gathering feedback to constantly improve the response, the relationship is personal.
Mostly because the suggestions are built on understanding the user on some level and gathering feedback to constantly improve the response, the relationship is personal.
How will their relationship change over time?
It will change over time as the bot recognizes and evaluates previous entries for new suggestions.
It will change over time as the bot recognizes and evaluates previous entries for new suggestions.
Scripts were written here to show how the functionality of the concepts would be interpreted and to see how the discussion would go.
The purpose of this usability testing was to learn about the user's opinion regarding the experience, convenience of use, clarity, and likelihood of using/needing a music chat bot. I held two rounds of usability testing sessions with the participants to assess any functional workflow issues and to determine whether the initial script met the interaction goals.
The major purpose of the initial prototype was to evaluate user perceptions about Case's personality and test the functionality concepts. To create expectations, participants were told they would be not be interacting directly with a conversational interface and just doing a walk through instead. At the same time as the participants were going through it, I was explaining my ideas to them. So that, they may still communicate any issues or queries to the assistance. It was completed using my first sample script, and then I adjusted the script and flow based on the testing results.
Following iteration, the second prototype was used for the second round of testing. The Voiceflow Prototype was tested with a large number of people. The testing sessions were held via Zoom. The following instructions were given to the participants for engaging with the prototype: Following each scenario, participants were asked follow-up questions on what worked, what did not work, what areas of improvement exist, and their personality perception.
The findings revealed that, in general, people believe this music assistant is funny and easy to use, and that the phrasing and dialog are generally simple to comprehend. However, the bot is still inefficient, causing people to become confused in specific areas.
The major purpose of the initial prototype was to evaluate user perceptions about Case's personality and test the functionality concepts. To create expectations, participants were told they would be not be interacting directly with a conversational interface and just doing a walk through instead. At the same time as the participants were going through it, I was explaining my ideas to them. So that, they may still communicate any issues or queries to the assistance. It was completed using my first sample script, and then I adjusted the script and flow based on the testing results.
Following iteration, the second prototype was used for the second round of testing. The Voiceflow Prototype was tested with a large number of people. The testing sessions were held via Zoom. The following instructions were given to the participants for engaging with the prototype: Following each scenario, participants were asked follow-up questions on what worked, what did not work, what areas of improvement exist, and their personality perception.
The findings revealed that, in general, people believe this music assistant is funny and easy to use, and that the phrasing and dialog are generally simple to comprehend. However, the bot is still inefficient, causing people to become confused in specific areas.
Find a balance between maintaining the personality and the quality of information. The fallback intents were difficult for users to understand. I was focusing on maintaining the personality and that dialogues seemed vague and the lesson about being an extra smart mouth was learned.
🔧 To reduce the confusion I created various specific intents that would prompt the user to reply to the question that was asked initially and made it sounds like a part of the conversation itself based on the personality feedback I received during testing.
Add a prompt to confirm with the users if they would like to continue/ be asked more question. Since the feedback indicated that customers were losing patience to continue if they were unable to find what they were looking for.
🔧 Instead of asking a series of questions and then recommending a playlist, I chose to check with the user and then recommend a playlist based on what the bot knows so far. This prompt will emphasize the recommendation's user-centered context.
The option of choosing none. Some participants felt compelled to choose an artist based on the possibilities provided initially, and I didn't want to establish a dark design pattern inside the bot even though it was just for the prototype. It was becoming quite time demanding to add options within the prototype if they answered no or none.
🔧 It was simpler to add this functionality with the genres question since I wanted to introduce the notion of users suggesting a genre they are already familiar with so that it may be remembered for the next time.
To improve efficiency, I changed the flow diagram based on the input I received. In the initial flow, users were not given a warning as to whether they should continue answering the questions or exit.
🔧 So I decided to ask the user if they wanted a playlist or to keep answering questions and offer each stage individually.
I believe that in the future, as the chat bot will obtain a more personalized route and the interaction model will become more intricate. Though it will be difficult to build utterances for all purposes, conversational developers can use methods to teach the bot to identify keywords more correctly and therefore understand and remember their preferences.
🔧 To reduce the confusion I created various specific intents that would prompt the user to reply to the question that was asked initially and made it sounds like a part of the conversation itself based on the personality feedback I received during testing.
Add a prompt to confirm with the users if they would like to continue/ be asked more question. Since the feedback indicated that customers were losing patience to continue if they were unable to find what they were looking for.
🔧 Instead of asking a series of questions and then recommending a playlist, I chose to check with the user and then recommend a playlist based on what the bot knows so far. This prompt will emphasize the recommendation's user-centered context.
The option of choosing none. Some participants felt compelled to choose an artist based on the possibilities provided initially, and I didn't want to establish a dark design pattern inside the bot even though it was just for the prototype. It was becoming quite time demanding to add options within the prototype if they answered no or none.
🔧 It was simpler to add this functionality with the genres question since I wanted to introduce the notion of users suggesting a genre they are already familiar with so that it may be remembered for the next time.
To improve efficiency, I changed the flow diagram based on the input I received. In the initial flow, users were not given a warning as to whether they should continue answering the questions or exit.
🔧 So I decided to ask the user if they wanted a playlist or to keep answering questions and offer each stage individually.
I believe that in the future, as the chat bot will obtain a more personalized route and the interaction model will become more intricate. Though it will be difficult to build utterances for all purposes, conversational developers can use methods to teach the bot to identify keywords more correctly and therefore understand and remember their preferences.
Reflection✨
Throughout the process, I asked myself the following questions:
How can I incorporate all my previous learnings in this project?
What should it say about me? How can I be brave and take more risks?
Can I make something unexpected? Can I keep learning?
Can I venture out into an unknown territory?
This motivated out of my personal desire for a friend like Case. Throughout the course of this project, I learned that I was not alone in believing that there must be a better and more comprehensive approach to find new music. Also, what problems users faced and what type of information they most require. Case, I feel, would be a useful and entertaining source of information. As an introvert who prefers text to speaking on the phone, I opted to focus on the chat features first. If I had more time, I would have worked on creating a multi-modal conversation for this.
Building on all of the features that I think it should have, the ability to suggest a playlist based on mood rather than just happy or sad, more broader terms like happy (joyful, pleased, playful, optimistic, amused, proud) angry (frustration, outrage, powerless, annoyance, bitter). My next step in broadening my expertise as a user experience designer will be to learn more about the multi-modal conversation’s limitless potential.