MagicSurvey
My Role
Lead UX designer
Timeline & Status
6 weeks
Shipped in Q3 2024.
Team
1 product manager
1 designer (me)
2 software engineers
Tools
Figma, ProtoPie
Skills
Interaction design
Visual design
Design systems
Rapid prototyping
Usability testing
Handoff
growth in the monthly retention rate
increase in task success rate
improvement of the monthly churn rate
MagicSurvey is internal AI-Native Product for ByteDance's employees. With over 40K+ users, it uses text-based AI prompt to generate survey questionnaires, summaries, and interview outlines.
MagicSurvey's H1 2023 data shows the product suffers in low user retention rate and low task success rate, with significant user drop out at its homepage.
Low Retention Rate
-50%
More than 50% of the users dropped off after landing at the home papge.
Low Task Success Rate
-63%
More than 63% of the generated content were not exported and not utilized by the user.
At the homepage, users found it difficult to discover core functions, follow task completion steps, and write accurate AI prompts.
Redesign the homepage that allow users to clearly navigate, understand and choose functions in simple steps while receive assistance in AI prompt writing.
What isn't working for the current homepage?
Complicated task completion steps.
User are required to follow through 6 steps before completing the task, resulting in high drop off rate.
Template are not discoverable.
Templates hey are too hidden in the information architecture, not discoverable by the users.
Users tent to use the wrong feature and receive unsatisfactory results.
Lack of AI prompt guidance
Product does not provide enough assistance to show users what a "good" AI prompt is.
Enable users to clearly navigate core functions and easily choose different templates, while receiving guidance for the AI prompt input.
Highlighting the core functions while emphasizing on each function's templates, allowing users to quickly discover and choose.
Simplifying task completion process to 3 steps, reducing frictions in the homepage user flow.
Tailoring and populating an AI prompt input library with 30+ examples.
Creating a consistent visual identity and reusable components for the product's Design System.
Establish visual hierarchy based on functions and different usage sceneriors.
Default state
With AI prompt library