Designing an AI-powered Chatbot for OneCard: A User-focused Approach
Introducing AI into OneCard’s chatbot can revolutionize customer service.
Here I am going to give you the walkthrough of how I tackled the problem statement which was provided directly by OneCard, and also a part of our 10k Designers Capstone project.
Problem Statement 🙇
Being a mobile first credit card, we aim to resolve our user’s queries and disputes through our digital platform by providing an extensively curated and articulated FAQ library. However, this alone doesn’t solve the pain point at large. We intend to simplify the experience by providing accurate answers to the customer’s queries rather than expecting them to read and find the answers. Hence design a chatbot that identifies the user context and gives dynamic responses.
Why do we need a Chatbot?
To answer this let us take note of the following points:
- Unlike humans, Chatbots are available 24x7, providing assistance to customers or users at any time of the day or night.
- Unlike humans, chatbot can handle multiple conversations at once, making them an efficient tool for handling customer inquiries and providing support.
- Chatbots can be a cost-effective alternative to human customer service representatives, as they can handle a high volume of inquiries without the need for additional staffing.
- Chatbots can be multilingual.
- Chatbots can also answer FAQs.
- Chatbots can handle an increasing volume of inquiries and can be easily scaled up or down depending on the needs of the business.
Visualizing the Image for Chatbot
Chatbots are often seen as a “serious business” model. Why so serious? Why not having a little empathy, a small amount of humanity in our Chatbots?
Some humour and friendliness can increase the usage of the chatbots more, hence will give a good business. Hence a chatbot with a cute face would do the basics.
Scope of Chatbot in Onecard
Chatbots can be used for multiple purposes, while in case of OneCard, the scope of Chatbot is to answer to the financial queries asked by the consumers and try to solve the problem too if faced under similar context. Any other than the financial reasons, the chatbot will fail to comply with the result.
Design Process 🔁
Defining the Problem: As I started this project, I began by defining the problem that I wanted to solve. I wanted to create a chatbot that would enable users to access and manage their OneCard account in a more efficient and user-friendly manner.
User Research: Next, I conducted user research to understand the needs and pain points of OneCard users. I reached out to users through surveys and interviews to gather insights on how they interact with OneCard, what challenges they face, and what features they would like to see in a chatbot.
Designing the Chatbot: Based on the user research, I started designing the chatbot. I began by defining the chatbot’s purpose, its capabilities, and its limitations. I then developed a persona for the chatbot and created a user flow to map out the user journey.
Primary Research 🕵️♂️
I engaged with several individuals, including friends and members of the 10k designers C5 community, through various forms of communication to gain insights into their challenges. Additionally, I conducted interviews with a few individuals who are both users and non-users of credit cards to deepen my understanding of their unique perspectives and needs.
User Surveys and Interviews
This was the first step to defining the problem statement by understanding the user needs and views regarding chatbots.
From the surveys and interviews, most of the people complained about Chatbots not being able to show empathy or dynamic behavior. Some of them even complained about the bots not being able to understand their custom problems.
Takeaway from the Surveys and Interviews:
- Bots are not well trained/taught about the general problems which users face on the day to day basis.
- There is a limitation when it comes to age of using Chatbots. Users over 40 years of age prefer to speak to human executives rather than a chatbot as they feel like Chatbots won’t give them more dynamic response as a human does.
- Lack of empathy and unable to solve the problems creates distrust among users.
- Users tends to avoid non human interactions when it comes to monetary aspects.
- Users avoid waiting in queue when trying to talk to a customer executives
- Users want a one fixed location where they can view all their transactions made with their credit card.
Competitor Analysis 💻
For this part, I took some of the biggest shark currently in this sector to compare their Chatbot and Customer Support, and the following were my takeaway:
- Slice: Slice has a reliable chatbot that can handle most customer queries efficiently. Its customer support team is also knowledgeable and quick to respond to queries, which earns it a rating of 4 for both chatbot and customer support.
- Cred: Cred has an excellent chatbot that is highly intelligent and can handle complex queries with ease. Its customer support team is also highly responsive and knowledgeable, earning it a rating of 5 for both chatbot and customer support.
- Lazy Pay: Lazy Pay has a chatbot that can handle basic queries but struggles with more complex questions. Its customer support team is responsive but can take some time to resolve customer queries, earning it a rating of 3 for both chatbot and customer support.
- Bajaj Finserv: Bajaj Finserv has a reliable chatbot that can handle most queries efficiently. Its customer support team is also responsive and knowledgeable, earning it a rating of 4 for both chatbot and customer support.
- Ola Money: Ola Money’s chatbot struggles to understand and respond to customer queries accurately. Its customer support team is also slow to respond and often unable to resolve customer queries satisfactorily, earning it a rating of 2 for both chatbot and customer support.
User Personas 🧑
With all the surveys and user interviews, I manage to craft some User Personas to help myself in understanding what frustrations and goals they might have.
Empathy Map 🗺️
To identify the users thoughts and feelings, I created a common empathy map for all the User personas I had.
User Journey 🚀
Proposed Solutions 🔎
While keeping the User personas and their Frustrations and Goals into account, I cam up with the following solutions that might be beneficial for the AI-powered aspects for the Chatbot.
- AI-Powered Chatbot will answer all the problems of the user over chat and more complex ones can be taken by one human executive over WhatsApp
- Adding an external layer to the current algorithm where the users can see the trend of their Credit score and ways of increasing it
- A separate section of list of offers and rewards for using the OneCard credit card
Design Decisions 💭
Information Architecture
I laid down the IA which I actually wanted to design since the actual IA of the entire app already exists and we were asked to be constraint to one specific feature. And there can be a huge range of queries, so I took one specific query and created an IA around it.
Wireframe 📝
Keeping my IA into account, I created a basic lo-fi wireframe for how the chat flow would look like. And this is what I came up with.
Branding 🌈
Visual Designs 🎨
Typical Outcome of a Chatbot
Welcome Screen
In the welcome screen, users will have a pre-list of most used queries available for them. This way they can save time and get their queries answered in limited number of clicks.
Lost/Deactivation design flow
What if the users want to deactivate their card for a possible reason. How will the flow look like? Let’s see.
Checking Payment Schedule
The most used feature is about the user periodically checking the due date of his/her EMIs in order to keep their Credit Score in a good state by not missing out on any dates. A user might have loads of EMIs and sometimes the user can lost track of all, so how about having a chatbot who will display all the linked loans at one place?
There could be lots of visuals and FAQs which can be thought of while creating contents for a chatbot but for the sake of not keeping it long, I will restrict myself to the two core problems which user face on the regular basis.
Bot Persona 🧑
Along with the Visuals and stuffs, creating a persona for the bot which the user will interact with is one of the core fundamentals which should be kep in mind.
OneCard Bot
So while creating the chatbot I thought of two possible ways to interact with the OneCard Chatbot and those were:
- Free typing option
- Choices Selector available
With these two, the user could mostly get all their problems solved as long as it’s based on the very product.
Tone & Voice
How important is it for a bot to have a tone? Quite important right. Cause none of us would be entertained with a boring bot.
- Friendly
- Calm and understanding
- Informative
- Welcoming
Personality
We always end up judging someone and their personality. Oh we doooo! Everyone judges everyone, and that’s the fact, LOL!
This part is the most crucial in any chatbot you may find. Answering problems is the aim, but the way in which the user aims to return back by enjoying the process of guiding through should be the aim.
So as per research it was seen that most of the consumers either used chatbots out of fear or curiousity.
If fear’s the case, the chatbot should act very patient and comforting, with a small tint of humor along the way.
I felt since we are looking forward to solving users problems, having Gemini trait would be a go-to.
- Intelligent
- Charming
- Attractive
- Simple
Its like having Morgan Freeman with traits of Angelina Jolie and Johnny Depp.
Intents 🤔
Also before we end this case study, let us not forget how important intents are while creating a chatbot. I did some thorough research and this is what I have come up with as Intents for OneCard bot.
Some words/expressions to be kept in mind
Key Takeaways 🎓
Yayyy! Finally you’ve reached the end. This project showed me some new level of thinking process and how deep the research goes while coming up with possible solutions for a specific problems. There are already a lot of solved solutions but to come up with something new, brainstorming with fellow designers and product managers is a way to go. So to conclude, the key takeaways I got from this assignment were:
- We need to keep out language as simple as possible since fintech apps are very much sensitive for users, cause money is involved.
- No designs can be perfect, so it doesn’t mean that whatever we think of might be the best. There is always a room for a better solution.
- Iterate. Iterate.. Iterate….!!!! Yes! New possibilities unlock as we keep iterating.
- Users can be moody, hence their problems can be too. You might never know what moody problems they will come up with, unexpectedly.
It’s a wrap, baby!🧳
Please show appreciation by hitting on the clap. Your support will mean everything! Most of my credit goes to the 10kdesigners mentors who showed me the ropes and guided me through every its and bits.
Also what ideas would you have come up with around this problem statement? Do let me know in the comments. I would love to know your ideas and expand my thinking abilities! ❤
Feel free to reach out to me on:
