Redesigning an internal ideas platform to cut submission time by 90%, reduce operational costs by 50%, and bring AI triage into production, while growing user satisfaction as the platform scaled 400x.
Reframed an internal ideas tool from idea intake into a scalable execution system.
Delivered a 2-phase rollout: Frontline experience (100 hard users), then operational unification + automation (scaled to 40K users over 3 months).
Impact: faster submission, higher adoption, lower processing cost, and improved decision velocity.
The platform captured ideas but did not support execution.
The redesign required aligning teams with different priorities and constraints.
Speed vs Structure
Visibility & governance
Launch pressure & cost constraints
How this shaped the strategy: we sequenced decisions to increase adoption first, then added structure through automation-improving governance without raising operational complexity.
Instead of expanding features, the strategy focused on sequencing change:
This led to a phased redesign:
The key principle:
Reduce effort before adding complexity
My role & Influence
Scope: Sole designer on the project. Led UX + product design across Phase 1–2, from operating model to rollout, in a cross-functional team with Product Manager, Tech Lead, Full-stack Engineers, and Operational stakeholders.
Strategy & success metrics
Operating model & system design
Execution & rollout
Collaboration
Partnered closely with Product, Engineering, and Operational stakeholders to scale the system without increasing friction or governance risk.
Throughout first and second quarter
What we changed:
What we changed:
The platform allowed employees to submit ideas, but it did not effectively support execution.
Sequence adoption before structure
We prioritized information hierarchy and surfaced decision-critical fields first.
Trade-off: more structure improves evaluation but may reduce speed.
Decision: balance clarity with fast submission.
Natural language input vs upfront forms
Users think in problems and opportunities, not categories.
We allowed natural-language input first and deferred categorization to later stages.
Trade-off: increased downstream processing, but higher adoption and better input quality.
Progressive rollout to manage risk
Rollout sequenced in 3 stages:
100 hard users (month 1) → 20K users (month 2) → 40K users (month 3)
Trade-off: there was pressure for a full launch. We chose a phased rollout instead, accepting a slower release to protect adoption quality, reduce operational disruption, and validate the system at each scale threshold before expanding.
The platform had two phases solution:
Impact: Reduced friction, improved participation, and increased journey completion.
Impact: Reduced manual triage effort and improved evaluation consistency.
What were our outcomes in production:
Adoption & User Behavior
Scaled from 100 → 40K users across a 3-month controlled rollout. System Usability Scale was 92 points which means the new userflow is excellent, and way easier to create a new idea, validated through moderated usability testing.
Results:
-90%
reduction in idea submission
98%
user satisfaction - target was 70%
Signal:
Stronger adoption and improved journey completion.
Operational Efficiency & Decision Throughput
70%
Projected reduction in reading time per idea via AI summarization (classification in production). AI summarization to reduce reading effort
reduced 80% idea filtering time
Signal:
Faster decision cycles and lower operational processing cost.
Business Impact & ROI
50%
reduction in recurring platform cost
annual operational cost reduced 50%
Signal:
Higher operational efficiency and improved ROI without sacrificing governance or performance.
AI was applied where it reduces measurable operational cost, not as a feature, but as a cost and throughput lever.
In production:
FinOps rationale:
Next measurement opportunity:
Structural shift: from idea intake to an execution-ready operating model.
Increased adoption without adding friction.
Reduced processing cost through a unified evaluation workflow.
Improved decision clarity across teams without increasing system complexity.
Now (0–4 weeks) → Next (1–2 months) → Later (quarter+)
Improve edge-case operational flows (Now)
Strengthen decision traceability (who/why/when) (Next)
Reduce feedback latency from capture → review (Next)
Expand AI summarization for evaluations (Later)
Add adoption + throughput dashboards (Later)
