ZhongZhong Chatbot
2024
Project Duration
2 weeks
My Roles
UX Designer
Visual Designer
My Skills
Comparative Analysis
Ideation & Brainstorming
Team Collaboration
Reminder: the original project is developed in Simplified Chinese. All design assets presented below are translated to English for presentation purpose.
Project Overview
About ZhongAn Insurance
ZhongAn Online P&C Insurance Co., Ltd. is an insurtech company in Shanghai, China. It is the first and largest insurance company in the country to offer its products through the Internet. By 2023, ZhongAn has served more than 5 million customers, and has issued about 57.4 billion insurance policies in total.
Project Background
In 2024, ZhongAn Insurance has planned to launch an AI-powered chatbot specifically on the "Book Service" section on the "policy details" page in its app. It will help users with inquiries related to additional services included in their insurance policy. By July 2024, the product team has launched an alpha version of the chatbot for internal testing.
Alpha Version of the ZhongAn Chatbot
The Problem & The Goal
The alpha version of the ZhongAn chatbot faces a few problems:
The conversational tone is too plain and robotic, lacking emotional empathy with users.
The lack of interactive components hinders users from interacting with the chatbot smoothly.
The UI looks sketchy and unprofessional, losing user trust and confidence.
A word from the manager
Project Timeline
Project Outcome
250+
Users per day, on average
↑15%
Satisfaction rate, compared to the internal competitor
54%
Engagement rate on prompt shortcuts and interactive buttons
User Data
The basic user demographics of the ZhongAn app are provided to offer an overview of the chatbot’s target user.
- 1/4 of users are between age 46 - 60.
- Majority of the users access the app through the Andriod platform.
- Almost all users use the app in Chinese.
The chatbot design will need to be accessible across various user groups, especially with the elderly:
- Content needs to be unambiguous and easy to understand
- Large font size and high color contrast to increase readability
- Avoid complicated interaction and hand gesture
- Multimodal approach to reduce cognitive load
Comparative Analysis
I have only 2 weeks for the project so there is no time for a complete user research. Alternatively, the comparative analysis would help inform the improvement of ZhongAn’s chatbot by investigating the effective design patterns and user interaction from its competitors.
Comparative Matrix
Some key findings that would inspire ZhongAn chatbot, after evaluating various competitors' chatbots through comparative matrix:
- Most competitors used a representative character or avatar for their chatbots
- Almost all competitors provided suggested prompts that exempt users from typing repeatedly
- Alipay and C-Trip have loading animation while generating responses
.png)
Ideation Process
User Flow Diagram
The “suggested prompt” interactive element will allow users to enter prompts with a simple click without repeatedly typing. To maximize the smoothness and efficiency of such a feature, I plan to make suggested prompts to appear in three different stages:
- The user initially enters the chat room
- The prompt is too broad (like “insurance”) and the chatbot asks the user to be more specific
- The chatbot asks whether they have further questions
I create a diagram to better understand the user flow of the chatbot with the presence of suggested prompts.
Rapid Wireframes + Prototyping
I brainstormed design solutions that might better present the users with suggested prompts, as well as other actions the chatbot prompts the user to do. Different color choices are also tested to ensure the color contrast is accessible to most users.
Discussion with the Programming Team
Before moving to designing high-fidelity prototypes, I met with the programming team to discuss about some limitations on the functionality of the chatbot based on the current coding setup.
Design Process
The Avatar - ZhongZhong
The chatbot will have an avatar called ZhongZhong - an animated robot charater who communicates with users in a friendly, respectful, and interesting way. The use of ZhongZhong will make interactions more approachable, engaging, and emotionally relatable.
The animation of ZhongZhong conveys intelligence and interactivity, helping users feel more connected.
Normal State
Loading State
The Design of Conversational Tone
Using a friendly and empathetic tone — like adding affirming language and light emojis — makes the chatbot feel more approachable and human. This approach makes the user feel more supported and emotionally connected with the chatbot, encouraging continued engagement.
The Design Solution
The high-fidelity design solution fulfills user needs and wants in various aspects.
- Empathetic & Approachable - expresses emotions in a friendly tone, improving connection with users
- Efficient - prompt shortcuts and quick start button made user interaction easier
- Accessible - Readable texts and simple user interaction
- Lively - Subtle animations makes the bot feel more alive
- Professional - Modern and minimalist UI design builds trust
Heuristic Evaluations
As an alternative to usability tests, this chatbot-specific heuristic evaluation is a less time-consuming way to identify usability issues and assess the overall user experience.
Key improvement recommendations derived from the heuristic evaluation:
- Avoid professional terms to communicate with the user more easily
- Run user testing to see whether the quick start buttons cause distraction
- Provide advice on writing prompts such as "try to type in keywords or a specific product or service name", informing the users how to fix an error when prompting
Project Outcome
After 1 month the launch of the ZhongZhong chatbot, the usage data are as follows:
250+
Users per day, on average
↑15%
Satisfaction rate, compared to the internal competitor
54%
Engagement rate on prompt shortcuts and interactive buttons
Conclusion
Key Takeaways
Through this project, I learned how to carry out UX design in a professional setting while working within tight time and budget constraints. I was able to extract useful insights from limited data and turn them into reliable design decisions. I also learned the importance of collaborating with developers while overcoming the challenges between design and implementation to ensure a smooth and effective outcome.