Designing a Survey Execution App for Real-World Data Collection
Reduced setup time and improved data accuracy by optimizing workflows, standardizing UI, and introducing access-based automation.
60%
Setup Time Reduced
Automation reduced setup steps from 6+ to 3, cutting time and improving agent efficiency.
+40%
Task completion speed
-50%
Input error rate
+90%
UI consistency score
role
Lead Product Designer
platform
Tablet & Mobile
duration
6 Months · Q4 2025
team
1 Designer · Dev Team
company
Karnival
Context
This app serves field teams responsible for on-ground data collection across distributed locations. Surveys are configured in a centralized backend system by administrators, then executed by field agents in fast-paced, real-world environments — retail stores, warehouses, and public spaces.
Backend System
Administrators define survey logic, branching rules, access policies, and field-level validations.
- Survey logic
- Access rules
- Field validations
- Branching conditions
Mobile & Tablet App
Field agents run surveys in real-world environments with minimal setup and maximum speed.
- Task execution
- Data capture
- Offline support
- Location context
Problem
Multi-step setup
Agents completed several manual steps before starting any task
Repetitive selections
Same context had to be re-entered for every survey session
Inconsistent UI
Similar interactions behaved differently across the app
Slow input interactions
Input fields caused friction and slowed data entry
Reliability issues
App instability in low-connectivity environments disrupted work
Core problem — Increased setup time and reduced efficiency in field operations, directly impacting data quality and agent productivity.
Research
74%
of setup time was spent on steps that could be automated based on login context
3×
more input errors occurred when field agents were interrupted mid-session
60%
of support tickets were caused by inconsistent UI behavior across similar flows
Reducing friction in setup has greater impact than adding new features
Before & After
6+ steps, manual setup, error-prone
3 steps, context-aware automation
Design Decisions
Agents spent significant time re-entering context already known from login
Implemented workflow automation using login data to pre-populate access context
Reduces agent control in edge cases — mitigated with override options
Different teams had built similar components with inconsistent behavior
Established a UI standardization system with a shared, documented component set
Initial cost of migration from legacy components to new system
Default inputs were too small for gloved or outdoor use
Optimized all inputs for touch — larger targets, reduced steps, smart defaults
Denser information per screen reduced in favor of usability
Generic UI didn't reflect the agent's location or survey context
Built context-aware components that adapt to access scope
More complex rendering logic requires thorough testing per config combination
Iteration Log
Auto-setup failed for multi-region agents
Added role-based override screen for edge cases
0 escalations after rollout
Input fields too small for outdoor gloved use
Increased touch target area and spacing on all form elements
Reduced mis-taps in field testing
Inconsistent loading states confused users
Standardized progress and skeleton patterns across all screens
Reduced confusion and support tickets
Want to go deeper?
I can walk through research artifacts, the Figma file, or the PRD in a live conversation.