Leena AI
Unified Analytics Dashboard
Leena AI's analytics lived across separate modules HR, bot performance, tickets, agent metrics forcing teams to jump between dashboards to answer one question. I owned the end-to-end redesign: research, IA, design, and handoff turning eight fragmented dashboards into one cohesive system.
Role
Lead Product Designer
Timeline
2 Months
team
1 Designers & 2 Engineers
Scope
Analytics, UX
Final Results
Time to insights: ~5 days to Real-time access
Bot improvement cycles: Monthly to Weekly
Bot improvement cycles: Monthly to Weekly
Bot improvement cycles: Monthly to Weekly
Why this mattered
Every team at Leena AI was looking at the same product through a different dashboard, and none of those dashboards agreed with each other. A meaningful share of users struggled to find what they actually needed.
Many stakeholders had resorted to pulling their own numbers into spreadsheets because the tool didn't answer their question. And there was no built-in way to compare metrics across reports or act on what they showed.
Eight dashboards, eight audiences, and nothing tying them together.
The problem
Area | What was broken |
|---|---|
Navigation | Felt like switching to a different product every time |
Filters | Didn't persist across dashboards, breaking the decision flow |
Benchmarks | No way to compare data, so numbers had no context |
Trust | The same view served HR, Support Ops, and Leadership equally poorly |
This wasn't a visual polish problem. Eight separate dashboards had been built for eight audiences, with no shared vocabulary connecting any of them.
What research showed
Persona | Core job | What research showed |
|---|---|---|
HR Admin / VP | Org health at a glance | Wanted a single overview, not module hopping |
Support Ops | Tickets and SLA tracking | Wanted a role-specific view, not a shared one |
Analytics / BI | Bot and session performance | Wanted cross-audience benchmarks, not raw numbers |
The real problem wasn't which data to show. It was that the same tool meant something different to each role, and the product never explained the difference.
The principle that shaped every screen
One system, eight dashboards, not eight separate products. Two rules carried through every decision:
Modular. Every metric is a card built from the same underlying components, not siloed per-dashboard logic. Consistency came from the parts, not from copying layouts.
Comparable. Every metric shows a range, a benchmark, or a trend by default, because context is what turns a number into an insight.
That gave each of the eight dashboards its own purpose without losing a shared visual language:
Overview
A high-level snapshot of new, engaged, and churned users, the anchor view everything else builds from.
User Metrics
Behavior, acquisition, and demographic segmentation, aimed at grounding feature decisions in real user segments.
Bot Metrics
Response quality, thumbs up/down feedback, response time by channel, with usage heatmaps for quick diagnosis.
Session Metrics
Session time, interaction rate, and session count, with clear separation between sessions and queries to isolate drop-off.
Queries & Interactions
What users are asking and how, focused on mining intent to keep improving bot flows.
Ticket Volume Trend
Backlog, status, channel, and priority, visible in real time so agents get allocated before queues build.
Case Management Performance
Resolution efficiency and SLA adherence, with color-coded, multi-channel compliance tracking.
Agent Performance
Individual performance across the full lifecycle, built for scanning large datasets quickly.
Testing before trusting the design
Ran co-design workshops with PM and engineering to validate the widget approach against the real data pipeline before any screens got built in Figma, and stress-tested filters against the messiest real-world data available rather than clean samples.
Testing led to a simpler default state and built-in benchmarks, so no number showed up without something to compare it against.
Results
Metric | Before | After |
|---|---|---|
Time to insight | 5+ days | Same day |
Bot improvement cycle | Monthly | Weekly |
SLA compliance visibility | Manual / Excel | Dashboard report |
Feature request prioritization | Reactive | Data-driven |
Looking back
One system doesn't mean one identical experience. HR and Support Ops needed different KPIs, and that was fine as long as they shared the same visual language underneath. The color-coded performance cues turned out to matter more than expected, people could tell if a number was good or bad without needing a legend to explain it.
