- Digital Products & Software
From Critical Bugs to Market-Ready: Upvrd’s Technical Overhaul
From Critical Bugs to Market-Ready: Upvrd’s Technical Overhaul
Case Study - Digital Product Development - Mobile App Engineering - India
179
Bugs Resolved in Phase 1
<1s
API Response (was 54.8s)
0%
Recommendation Engine Errors
0%
Journey Module Error Rate
3
Structured Delivery Phases
0
Critical Bugs Remaining
01: The Challenge
A Strong Vision, a Broken Foundation
Upvrd's architecture was sophisticated by design - three interlocking systems managing personalised journeys, a multi-currency gamification economy, and a content recommendation engine. That same complexity made its initial instability compound rapidly. By the time engineering support was engaged, failures were not isolated to one area; they ran across authentication, core features, data tracking, and API performance simultaneously.
Before
179 bugs across core user flows. No new user could acquire or engage with the platform.
OTP verification entirely non-functional. All new sign-ups blocked at the first step.
Book data API averaging 54.8 seconds per response. App froze and became unresponsive on load.
Recommendation engine failing 90% of the time. Personalised homepage completely unreliable.
User progress and coin balances not tracking in real time. Core gamification broken.
After Auriga IT
All 179 bugs resolved. Every critical and high-severity defect cleared before phase progression.
OTP verification and authentication flow fully restored. New users enter the platform.
API response under one second. The book data endpoint no longer causes app freezes.
Recommendation engine error rate at zero. Homepage renders content on every load.
Real-time progress tracking and coin balance updates working across the gamification system.
01
179 Bugs Across Core Flows
Critical user flows were completely broken. The OTP verification flow was non-functional, blocking every new sign-up before they entered the platform.
02
Core Features Inoperable
Global search, user profile management, and "My Highlights" failed to return results or save data. Key parts of the user experience were completely unavailable.
03
54.8-Second API Response
The book data API averaged 54.8 seconds per response. A separate 90% error rate on the recommendation engine made the personalised homepage completely unreliable.
04
Broken Gamification Tracking
User progress and in-app currency balances were not updating in real time. The three-currency economy - Wisdom, Compassion, Energy - could not be trusted, breaking the core engagement loop.
02: The Solution
A Three-Phase Recovery Strategy
The recovery was structured deliberately - stabilise the platform first, then re-architect for performance, then build the processes to keep it that way. Each phase had a clear entry condition and a measurable exit threshold before the next began.
1
Triage and Stabilization - Stopping the Bleeding First
With 179 bugs distributed across authentication, core features, and data integrity, the priority was eliminating every blocker preventing users from entering the platform at all. The team triaged by severity and worked through the backlog systematically - authentication first, then the features a new user encounters immediately.
Authentication Repair
OTP verification and sign-up flow restored - new users can enter the platform
Feature Restoration
Global search, profile management, and "My Highlights" returned to full functionality
Gamification Fixes
Real-time progress tracking and coin balance updates working across all three currencies
API Patching
Broken endpoint calls repaired to stabilise the core user journey end-to-end
upvrd.app / engineering / phase-1-triage-stabilization
Resolved
Phase 1 - 179 bugs triaged and resolved, authentication flow restored, core features operational
Phase 1 proof points: OTP repair | Authentication flow | Real-time coin tracking | Core feature restoration | 179-bug backlog cleared
2
Performance Re-architecture - Fixing What Made the App Unusable
With baseline stability in place, the focus shifted to the two performance failures that would have prevented Upvrd from retaining any user it successfully acquired. A 54.8-second API response and a 90% recommendation engine failure rate required architectural changes, not patches.
API Pagination
Book data API re-architected with pagination - response from 54.8 seconds to under one second
Query Optimisation
Database queries restructured to return paginated result sets instead of full datasets per call
Fallback Mechanisms
Recommendation engine rebuilt with error handling and content fallback - homepage always renders
Journey Module Repair
25% journey module error rate resolved - primary learning experience runs without interruption
upvrd.app / engineering / phase-2-performance-rearchitecture
Optimised
Phase 2 - API response from 54.8s to under 1s, recommendation engine error rate from 90% to 0%
Phase 2 proof points: API pagination | Query optimisation | Recommendation engine fallback | Journey module stability
3
Process Refinement - Keeping the Platform Stable Post-Launch
Fixing existing defects is only half the work. The third phase addressed the development and QA practices that allowed the initial backlog to accumulate - introducing the controls needed to keep the platform stable as new features are added after launch.
QA Checkpoints
Structured review gates at each development milestone - defects caught before reaching production
Regression Testing
Protocols introduced to prevent resolved bugs from resurfacing in future release cycles
Deployment Standards
Code review and release standards defined to maintain the performance gains from Phase 2
Technical Documentation
Recommendation engine and currency economy logic documented for future development teams
Phase 3 proof points: QA checkpoints | Regression testing | Deployment standards | Technical documentation
03: What Was Built
Capabilities Delivered
OTP and Authentication Repair
Sign-up flow fully restored - new users can create accounts and enter the platform
API Pagination and Query Optimisation
Book data API redesigned to return paginated results - response from 54.8s to under 1s
Recommendation Engine Rebuild
Error handling and content fallback logic added - homepage renders on every load
Real-Time Progress and Currency Tracking
Wisdom, Compassion, and Energy balances update in real time across the gamification system
Core Feature Restoration
Global search, profile management, and "My Highlights" returned to full functionality
Journey Module Stability
25% error rate on the primary learning experience resolved across all journey flows
QA Process and Regression Testing
Structured checkpoints and regression protocols introduced for ongoing release stability
Deployment Standards and Code Review
Release and review frameworks defined to sustain performance gains from re-architecture
Technical Documentation
Recommendation engine and currency system logic documented for future development teams
04: Business Impact
From Technically Paralysed to Market-Ready
Every metric tracked during the engagement showed a complete reversal. The application that could not acquire or retain a single user now operates within normal performance parameters across all previously failing systems.
<1s
API Response
(from 54.8s)
(from 54.8s)
0%
Recommendation
Engine Errors
Engine Errors
0%
Journey Module
Error Rate
Error Rate
0
Critical Bugs
Remaining
Remaining
| Metric | Before | After |
|---|---|---|
| Book Data API Response Time | ~55 seconds | Under 1 second |
| Recommendation Engine Error Rate | 90% | 0% |
| Journey Module Error Rate | 25% | 0% |
| Critical and High-Severity Bugs | 120+ open | 0 |
| User Sign-up Flow | Non-functional | Fully operational |
| Core Feature Availability | Multiple inoperable | All features restored |
05: Technology and Engagement
Engagement Overview
Mobile Application (iOS / Android)
Backend API Layer
Database Query Optimisation
API Pagination
Recommendation Engine
Regression Testing
QA Checkpoint Framework
Technical Documentation
A
Platform
Mobile application on iOS and Android with a cloud-hosted backend API layer powering the recommendation engine, gamification system, and content delivery.
B
Engagement Type
Application support, performance engineering, and QA process establishment across a three-phase structured delivery model.
C
Delivery Model
Severity-first triage progressing through three milestone-gated phases. Each phase required measurable exit criteria before the next began.
06: What This Proves
For Any App Team Carrying Pre-Launch Technical Debt
Consumer apps with layered architectures - gamification economies, personalisation engines, multi-role user flows - carry compounding risk when stability is deferred. A single broken authentication flow does not just block sign-ups; it disconnects every downstream system that assumes a valid user exists. This engagement demonstrates that a structured, severity-first recovery can move a product from pre-launch paralysis to market readiness without discarding the architecture that gives it its differentiation. Platforms with similar complexity can be stabilised and performance-tuned without a rebuild - provided the intervention is methodical and performance targets are defined before re-architecture begins. A severity audit before launch costs significantly less than addressing the same failures after public release, when acquisition and retention data are already affected.
07: Frequently Asked Questions
Questions About This Engagement
How did Auriga IT prioritise which bugs to fix first?
The triage process ranked defects by user impact and system dependency. Authentication and sign-up failures were addressed first because they blocked access to every other part of the platform. Once new users could enter the app, the team moved to features they would encounter immediately - search, profile, and progress tracking - before addressing lower-severity edge-case issues.
What caused the 54.8-second API response time?
The root cause was unoptimised database queries returning unpaginated result sets. The book data API fetched the full dataset on each request rather than returning results in pages. Re-architecting the query logic and implementing pagination brought response time to under one second without changes to the underlying data model.
Was the recommendation engine rebuilt from scratch?
No. The re-architecture focused on adding error handling and content fallback logic to the existing rules-based engine rather than replacing it. When the engine previously hit an edge case or a content gap, it returned an error. After the intervention, it falls back to a default content set, ensuring the personalised homepage always renders regardless of edge conditions.
How long did the full engagement take?
The three phases were executed sequentially with clear entry and exit conditions for each. Phase 1 triage and stabilisation proceeded until all critical and high-severity bugs were resolved. Phase 2 performance re-architecture followed once the baseline was stable. Phase 3 process refinement ran in parallel with the tail end of Phase 2 to avoid delaying the overall timeline.
Is this engagement relevant if our product is not yet in the market?
Yes. Pre-launch products with accumulated technical debt or inherited codebases from third-party development often carry the same risk profile as Upvrd did. A severity audit and structured triage before launch costs significantly less than addressing the same issues after public release, when user acquisition and retention data are already affected by the failures.
Can Auriga IT support ongoing maintenance after the engagement ends?
Yes. Auriga IT offers continued application support and maintenance engagements for teams that need ongoing engineering capacity. The documentation and QA frameworks delivered in Phase 3 are designed to reduce that dependency over time as the client's own team absorbs the process standards introduced during the engagement.
Dealing With a Similar Problem?
If your application has accumulated technical debt, performance bottlenecks, or a defect backlog delaying launch - one conversation to map where the process is breaking and what a structured intervention changes.
Talk to the Team
Industry
BSFI
Services
Enterprise IT Transformation & Automation
Product Development
Product Development
Related content
Auriga: Leveling Up for Enterprise Growth!
Auriga’s journey began in 2010 crafting products for India’s [...]
