The Problem
- Business: Support tickets up 15% QoQ; only 34% feature adoption on a 2M-user IT ops platform; executive pressure for AI without production risk.
- Users: 12+ clicks to triage incidents; 40% of session time spent navigating; AI suggestions ignored due to lack of context and confidence signals.
- Constraints: Phased rollout required — no downtime, no full platform rewrite.
The Solution
- Strategy: Fix top 5 workflows first (validated by support data), then layer AI at decision points — not a generic chatbot.
- Design: Workflow-first IA, progressive disclosure, inline AI with confidence scores and expandable rationale; 12 patterns added to design system.
- Validation: 16 usability sessions, A/B test with 2,400 users over 6 weeks, WCAG 2.1 AA audit.
Executive Summary
Enterprise IT administrators were struggling with a fragmented dashboard that required 12+ clicks to complete common troubleshooting workflows. Support tickets were rising, feature adoption was stagnant, and executive stakeholders questioned the platform's ROI.
I led a strategic redesign grounded in user research and journey mapping — introducing tiered information architecture, contextual AI recommendations, and workflow-optimized layouts. The result: measurable productivity gains, reduced support burden, and renewed stakeholder confidence in the product's strategic value.
Business Problem
The platform served 2M+ enterprise users managing critical IT infrastructure. Despite significant engineering investment, key metrics were declining:
- Task completion times 2x industry benchmarks for comparable SaaS products
- Support ticket volume increasing 15% quarter-over-quarter
- Only 34% of shipped features showing meaningful adoption
- Executive pressure to demonstrate AI differentiation in a competitive market
User Research
Methods: Contextual inquiry (12 sessions), support ticket analysis (500+ tickets), analytics review, stakeholder interviews with 8 PMs and engineering leads.
Participants: IT administrators, operations managers, and tier-2 support engineers across enterprise accounts in North America and EMEA.
Key finding: Users weren't failing because features were missing — they were failing because critical information was buried under layers of navigation, and AI capabilities lacked contextual relevance.
Discovery Insights
Insight 01
80% of daily tasks involved the same 5 workflows — yet the UI treated all features with equal visual weight.
Insight 02
Users spent 40% of session time navigating between modules instead of acting on data.
Insight 03
AI recommendations were generic — users ignored them because they lacked situational context and confidence indicators.
Insight 04
Power users created personal workarounds (spreadsheets, bookmarks) — signaling unmet product needs.
Journey Mapping
Mapped the end-to-end journey for the top 3 workflows: incident triage, capacity planning, and compliance reporting. Identified 14 friction points across awareness, navigation, comprehension, and action stages.
Critical moment: The "triage decision point" — where administrators evaluate alert severity and determine response — had 6 unnecessary steps and no AI assistance despite available data.
Opportunity Areas
- Workflow-first IA: Restructure navigation around tasks, not feature modules
- Contextual AI layer: Surface recommendations at decision points with confidence scoring
- Progressive disclosure: Show summary → detail → action in a single viewport flow
- Personalization: Allow role-based dashboard configurations for different admin personas
Product Strategy
Partnered with PM and engineering leadership to define a phased approach:
- Phase 1: Redesign core triage workflow (highest support volume)
- Phase 2: Introduce contextual AI recommendations with human-in-the-loop controls
- Phase 3: Expand personalized dashboard configurations across user roles
Defined success metrics upfront: task completion time, support ticket reduction, feature adoption rate, and AI recommendation acceptance rate.
Design Exploration
Explored 4 concept directions through low and high-fidelity prototypes — testing with 8 users per round. The winning direction combined a workflow-centric sidebar, card-based alert prioritization, and inline AI suggestions with expandable rationale panels.
Contributed 12 new patterns to the enterprise design system — ensuring consistency across adjacent products while accelerating future development.
Validation
- Moderated usability testing with 16 enterprise administrators (3 rounds)
- A/B test on triage workflow with 2,400 active users over 6 weeks
- Accessibility audit achieving WCAG 2.1 AA compliance
- Stakeholder review with VP Product and engineering directors
Final Solution
Delivered a redesigned dashboard with workflow-first navigation, AI-assisted triage with transparent confidence scores, progressive disclosure patterns, and role-based personalization — all built on the enterprise design system for cross-product consistency.