Summary
Built and stabilized a reservation system for global luxury fashion brand “D” by keeping a failing system running while delivering a new system within 3 months, then operating it reliably for 3 years under high traffic.
Overview
Key Information
- Duration : Nov 2022 – Dec 2025
- Industry : Retail / Luxury Fashion
- Type : System Redesign and Rebuild
- Team : 8 (PM 1, Planner 2, Designer 2, Developer 3)
Role
- Project Manager (70%)
Scope
- Service architecture design
- Reservation logic design
- Infra and tech stack decisions
- Operation policy
- Client communication
Background
- The client, global luxury fashion brand D, runs a reservation-based store experience.
- At the time of contract, the service was already live on a freelancer-built system.
- The system frequently crashed during reservation openings.
- Reservations could not be paused, but a full rebuild required at least 3 months.
Problem Definition
User Perspective
- Frequent failures during reservation opening
- Loss of trust and drop-off due to delay and errors
Business Perspective
- Continuous failures while service could not be stopped
- Brand experience damage and increasing VOC
Technical Perspective
- No consideration for concurrent traffic
- Infrastructure incapable of preventing failures
- Full rebuild not feasible in a short time
Strategy
UX Strategy
- Split approach between maintaining operations and rebuilding
Structural Strategy
Business Strategy
- Maintain existing system under temporary conditions
- Prioritize “working” over “fast”
- Increase infra cost temporarily to ensure stability
- Set clear goal for new system delivery within 3 months
- Design for long-term cost efficiency
- Separate short-term workaround from long-term architecture
Execution
Stabilizing the legacy system
- Expanded infra and applied load balancing to prevent crashes
- Allowed latency but ensured successful reservations
- Added a simple progress bar on the front-end
- Made waiting feel intentional
- Reduced user drop-off during delays
- Managed incidents and client communication
Building the new system
- Rebuilt on cloud infrastructure
- Redesigned reservation logic for high concurrency
- Simplified UX for fast booking
Admin dashboard
- Designed based on real store operations
- Real-time reservation monitoring
- Daily visitor tracking
- Centralized control
Transition
- Switched systems without stopping service
- Optimized infra cost after transition
- Delivered clear improvement perception to client
Results
Quantitative Impact
- Maintained service during 3-month transition without downtime
- Stable operation for 3 years after rebuild
- 99.99% availability
- Supported hundreds to thousands of concurrent users
- Less than 1 incident per year
Qualitative Impact
- Significant reduction in reservation-related VOC
- Improved infrastructure efficiency
- Evolved into a data-driven operation system
Key Learnings
- Maintaining a failing system can be harder than building a new one
- “Working” can be more important than “perfect”
- UX can compensate perceived performance
- PM’s role includes managing the transition, not just delivery