Reframing a feature launch to improve existing UX
Sole Product Designer
Cross-functional strategy
7 weeks
Apr - May 2025
1 CTO
1 PM
3 Engineers
tl;dr.
Data-driven Design
I identified system bottlenecks that'd harm the launch of a new feature.
Stakeholder Management
I mitigated risk by recommending improvements to the existing system even if it meant delaying the feature roadmap.
Adapting to the Org
I designed under lean resources and technical constraints, acknowledging design changes would occur through small increments.
overview.
Context
Curious Cardinals (CC) matches mentors with K-12 students for academic and passion-based mentorships. Mentors use the web-based platform to manage learning progress, communication with parents, and more.
The Problem
Curious Cardinals wanted to launch Progress Updates—a new task type to improve mentor-guardian communication. However, the existing task system lacked structure to support it.
What I Did
I redesigned the post-session task flow for 650+ mentors, to reduce completion delays. I then designed a new feature within this improved system.
Product & Business Impact
1️⃣
Higher customer satisfaction: Created a new touchpoint for proactive guardian fulfillment, differentiating CC's competitors who rely on reactive, email-on-demand models
2️⃣
Improved operational efficiency: Reduced mentor support needs through clearer, enforced workflows
3️⃣
Enhanced data foundation: Complete session records enable future ML personalization and matching improvements
4️⃣
Unlocked new revenue: Progress Updates enable future upsell opportunities for mentor-guardian interactions
Key Outcomes
Redesigned components for the task management system
Redesigned task cards
Redesigned Attendance task
Redesigned Session Notes task
A complete user flow for Progress Updates that will serve as the catalyst for new revenue opportunities
Progress Update task: what was designed and handed off
research.
TL;DR
My audit revealed bottlenecks that'd harm the launch of Progress Updates, which depended on timely Attendance and Session Notes.
What I Found
Mentors took 10+ days to complete session notes, meaning stale information and paying customers kept in the dark.
13% of sessions skipped the AI notetaker, meaning mentors wrote manual notes, which adds days to completion time
Root Cause Analysis
I identified the root cause: Attendance and Session Notes were decoupled, causing lags in post-session task completion.
Currently, Attendance and Session Notes tasks are separate tasks with separate entry points
Event
Mentor Action
Curious Cardinals Platform (System)
Mentors completed Attendance immediately but delayed Session Notes indefinitely.
The Risk
Launching Progress Updates on this weak foundation would:
⌛️ Feed guardians outdated session data
🔀 Create non-chronological progress reports
⏰ Add another overdue task mentors would ignore, only adding clutter to the task management system
design.
TL;DR
I tackled the problem in 2 steps.
1️⃣
Couple Attendance and Session Notes to enforce completion
Success metrics: increased completion rates, decreased time-to-complete
2️⃣
Design Progress Updates as a 3-step task (Attendance → Session Notes → Progress Update)
Success metrics: decreased guardian follow-up, successful feature launch
1️⃣ couple Attendance and Session Notes.
The Problem
Attendance lived in a modal. Session Notes lived in a drawer. The jarring transition let mentors close Attendance and ignore Session Notes, reinforcing the behavior.
Left: Attendance modal + Right: Session Notes drawer = UI nightmare
The Solution
I merged both into a single 2-step modal.
I swapped the order: Duration is only applicable if the user selects "Attended."
During the design phase, another opportunity to gather data emerged: mentor’s satisfaction with Sessions overtime.
I ran into an Eng. constraint. The intention of labeled steps was to reinforce how the 2 decoupled tasks were joined into 1 modal. But using an existing component was higher priority.
I introduced an alert component and set a pattern for its usage, modified after the Chalkra design system, which was being used by CC very loosely.
Although I advised against this, I removed correction and undo functionality as a result of an Eng. constraint: data is stored and uneditable after step 1.
Task Card Redesign
I then redesigned task cards to show multi-step status and enforce chronological completion. Incomplete prerequisites? Locked task.
Explorations
I flagged the benefits and tradeoffs of the merge, especially with Engineering constraints in mind.
Benefits
Reduces decision fatigue: locked sequence removes ambiguity
Faster task completion: less context-switching between different UI patterns
Increases data quality: better foundation for automation
Tradeoffs
Less flexibility for users: correcting errors is harder
Greater lift for Eng: the system needs to regularly run a query on transcript processing status
Result
I resolved the bottlenecks with the locked task card.
With safeguards, the unhappy path no longer created issues in the platform operations
2️⃣ design Progress Updates.
The Requirement
Progress Updates needed to surface Milestones and Artifacts from recent sessions—making it a 3-step task vs. the 2-step post-session flow.
How the 3-step post-session task compares against the 2-step post-session task
The Solution
I explored how to introduce Milestones/Artifacts as Step 3 without overloading the modal.
Explorations
Engineering Constraints
The team required reusing existing Milestone/Artifact components from elsewhere in the platform. I adapted my designs to work within this constraint while maintaining visual consistency.
All 3 steps put together
Collaboration with Engineering in Notion via mocks and comments
outcomes.
Multi-Step Post Session Task System
Locked task cards enforce chronological completion
Clear visual status for each step
Reduced UI friction between Attendance and Session Notes
Progress Update
3-step modal extends the post-session flow
Surfaces relevant Milestones/Artifacts for guardian communication
reflections.
Challenge Assumptions
CC asked for a new feature. I stepped back to look at the big picture and delivered a system redesign that made the feature possible.
Prioritize—Ruthlessly
I moved with high velocity, thinking at a systems level, because while priorities are constantly shifting at a startup, there's still all interconnected.
Design for the Long Game
I evaluated solutions based on their adaptivity and scalability. Incomplete PRDs and tons of ambiguity is an opportunity, not an obstacle.