Enterprise Survey Logic & Analytics Reimagined
The 2-Hour Survey Tax
Skill Constellation
Primary
Supporting
Emerging
John Deere's 250+ products all need user feedback. But building a single survey took ~120 minutes of back-and-forth between a Product Manager and an engineer — filing tickets, reviewing builds, requesting changes. For an organisation that needs feedback constantly, this wasn't just inefficient. It was a bottleneck slowing down product decisions across the entire company.
I was asked to fix it. Before touching a pixel, I set three baseline metrics with the product owner so we'd know whether we actually succeeded:
Time to build a survey
User satisfaction (CSAT)
Survey completion rate
Metric-Driven Design
Set explicit success criteria with the product owner before starting design. Baseline metrics became the benchmark for every design decision.
Evidence: 3 measurable targets defined pre-design; all exceeded post-launch.
How UX Research Reshaped the Entire Workflow
I started with 45-minute semi-structured interviews across product owners, engineers, and UX researchers — everyone who touched a survey. The finding that changed the entire project came from observing users building branching logic: they had to save their question, switch to a separate "Logic Rules" tab, manually look up question IDs, and write conditional rules by hand. This cognitively heavy process led to a 60% drop-off rate when building complex surveys.
Contextual User Research
Conducted 45-min semi-structured interviews AND observed real survey-building sessions. Observation uncovered the branching logic pain point that interview questions alone missed.
Evidence: 60% drop-off rate at branching logic identified through task observation.
The Solution: I fundamentally restructured the mental model. By introducing an Inline Branching UI directly on the Question Card, users could define rules contextually without ever leaving the builder canvas.
Fragmented cognitive load: Users had to memorize question IDs and navigate across multiple disconnected tabs to link logic paths.
Contextual execution: Branching rules are defined directly inline with the question, dropping cognitive load entirely.
Inline Logic Design
Replaced disconnected tab-switching workflow with contextual inline branching directly on the question card, eliminating the need to memorize question IDs.
Evidence: Hesitation around branching dropped to zero in Round 2 testing.
This single design decision eliminated the biggest usability barrier. Round 2 testing confirmed that hesitation around branching dropped to zero — users no longer had to hold question IDs in their head while switching between disconnected tabs.
Designing a Self-Serve Platform
Research revealed four core capabilities the platform needed: a drag-and-drop question builder, inline branching logic, real-time preview with John Deere branding, and an analytics dashboard that eliminated the manual CSV-to-Excel pipeline entirely.
Solving the Analytics Dead Zone
The old process: export a CSV, open Excel, build a chart, share it in an email. By the time insights reached decision makers, the data was stale. I designed an expandable data table — expanding a survey row reveals interactive analytics right where PMs track their surveys. No separate reports page. No context switch.
Analytics-in-Context Design
Eliminated the CSV→Excel→Email pipeline by embedding interactive analytics directly within the survey management table.
Evidence: Insights accessible in one click, not a multi-step export workflow.
Every Baseline Target Was Exceeded
The pilot launched with a 50-person PM cohort across four product lines, tracked over 4 weeks.
Survey build time dropped from ~120 minutes to just 48 minutes.
Survey completion jumped from 50% to 62.5% due to better logic and branding.
Platform satisfaction score hit 4.4 / 5, easily beating the target.
Completely eliminated engineering dependency for routine surveys.
I can spin up a survey in under five minutes — no coding needed.
— Product Manager, Pilot Participant
Enterprise UX & Self-Serve Design
Transformed a tool requiring engineering support into a fully self-serve platform, eliminating dependency entirely.
Evidence: 100% self-serve adoption; pilot success led to funding for advanced features.
What Came Next
The pilot's success led directly to funding for Multi-Language Support, Advanced Cross-Tabulation Analytics, a permanent Template Library, and full CRM integrations. A tool meant to solve a 2-hour bottleneck became the foundation for how John Deere listens to its users.