Lyft / internal tool / data visualization
Marketplace Health_
Turning a single-chart query tool into one real-time analytics platform for Lyft’s leaders and analysts.
- My role
- Sole product designer
- Timeline
- 2020 → 2026
- Priority
- High
- Platform
- Internal web dashboard
- Reporting time
- 2h → 10m
- Weekly users
- 200+
My role and who I worked with
As the sole designer, I owned the experience end to end:
- User interviews, competitive analysis and usage-data review
- Information architecture and navigation for the whole product suite
- Interaction and visual design for charts, filters, views, search and the landing page
- A custom Health logo, plus onboarding, training documents and videos
- Kanchan GaurMEProduct Designer
- Bob NiscoStaff Engineer
- Petr OdehnalSenior Software Engineer
- Neille Illel, Anita LillieSupporting advisory
Marketplace/Central Market Management & AI/Decision Apps & Metrics Explorations
Summary
Lyft’s leadership and decision makers needed one place to monitor supply and demand and keep each market in balance.
Health started as a query tool that showed one metric on one chart at a time. Over six months I redesigned it into a multi-metric, real-time dashboard, and then helped fold it, the Executive Dashboard, Regional Insights and Source of Truth into one connected platform with shared navigation, global search and onboarding.
The problem and the goal
What wasn’t working
- Limited tool. Basic filtering, inconsistent metrics, no hierarchy in dimensions, little interactivity, and one chart or metric at a time.
- Fragmented workflow. Insights lived across Command Center, Mode and ad-hoc queries, so answers were slow.
- Hard to find. Later research showed Executive Dashboard users didn’t know Health existed, search needed exact metric names, and products had no way to move between each other.
Two distinct audiences
Directors, executives, VPs
- Need a quick pulse on business health
- Care about trends, summaries and quarterly targets
Analysts and data scientists
- Need flexible, powerful tools to investigate and compare trends
- Run queries, save views, export data
- Require accurate, consistent metrics they can trust
The goal
Transform Health into a seamless experience that balances power, flexibility and clarity, then streamline Lyft’s scattered tools into one unified, one-stop platform.
- Executive Dashboard
- Regional Insights
- Price Commander
- Market Management
- Reports
- Health
- Source of Truth
- Mode
My process
- 01ResearchUser interviews and competitive analysis
- 02IdeateBrainstorming and concept exploration
- 03DesignPrototyping and design systems
- 04TestFeedback sessions and iterations
- 05ShipDeliver and measure impact
Listening to users
I interviewed a Pricing Lead, an Analyst and a Supply Manager to understand their goals, what was and wasn’t working, how they built reports, and which tools they relied on. Their needs became a prioritized feature list:
- 1Faster filtering
- 2Multi-metric views
- 3Categorized metrics
- 4Layout customization
- 5Share reports
- 6Data export
- 7Anomaly detection and market events
Competitive analysis
I studied Grafana for its technical depth, Mode for its simplicity and Power BI for its interactivity, comparing how each handles adding metrics, metric settings and filters, legends and chart display.
Usage data and a second round of interviews
As the suite grew, I reviewed usage data (most visited dashboards, visits by tab, engagement by persona) to see what people actually used day to day, where they dropped off and where they got stuck. Follow-up interviews surfaced four themes:
Discoverability
Executive Dashboard users didn’t know Health existed.
Search
Two separate searches, and metrics couldn’t be found without exact names.
Navigation
Siloed products with no way to move between them.
Onboarding
No training and no guidance on where to start.
Information architecture
- Audited existing content for duplication and findability issues
- Grouped products by intent: Monitor, Explore, Source of Truth, Regional Insights
- Consolidated pages into clear buckets: Marketplace, Rider, Driver, Levers
- Built a scalable menu with consistent naming across all products, with high-traffic dashboards up front and low-usage items in secondary paths
How I solved it
The work happened in two phases: first rebuilding the Health dashboard itself, then connecting it with the rest of Lyft’s market tools.
Phase 1: Rebuilding the Health dashboard
Saved and shared views
- Save a setup and pick up where you left off, without rebuilding charts
- Share a view with a link
- A warning alert flags an edited view and offers to reset or save the changes
Multi-metric renderings
Clear line color schemes keep metrics stable and readable across groupings and lookback scenarios, so multi-metric charts stay consistent even when split by dozens of regions.
Global filters
Global and region-specific filters made it simple to focus on one business area. Tooltips were refined to tell timestamps apart in cross-granularity settings.
Customizable chart grids
Drag-and-drop chart resizing and a reorganized metric selection layout gave people control over their workspace and quicker access to advanced configuration.
Market events and anomaly detection
I designed Market Events so the Revenue Operations team could see every factor affecting a market in an interactive Gantt chart alongside key metrics:
- Price changes
- Incentives
- Severe weather
- Competition
It made it easier to spot trends, confirm specific actions and plan for upcoming changes.
A custom Health logo
I created a mark for Health, iterating with the team to land on something that felt precise and alive. It became a badge of the tool’s evolution, sized from a 16px favicon to a 96px desktop shortcut.
Phase 2: One platform for the whole suite
Leadership relied on summary tools while analysts lived in Health. I explored four directions for a shared homepage before landing on the final structure.
Four homepage concepts
- Guided hub: a welcoming home that explains each product with direct calls to action
- Unified highlights: top metrics, saved views and top regions in one overview
- Questionnaire: a short onboarding flow that generates a personalized home
- Build your own: drag metrics onto a canvas and save custom dashboards
Navigation and menus
The new information architecture became the menus themselves. Dashboards opens a single panel grouped by business area (Business Overview, Reports, Quarterly Targets, Marketplace, Levers, Rider, Driver and Media & Lyft Business), so people browse by topic instead of hunting through separate tools. Explore puts My Views and Popular Views up front, with a shortcut to create a new view, so returning users go straight back to their own work and new users can start from what others already rely on.
Cross-linking products
People can now jump from the Executive Dashboard to Explore, from Explore to Source of Truth, and back again, without losing context.
Robust global search
One search finds dashboards, pages, chart titles, metrics, views, regions, analyst insights and metric groups, with tags and categories, so no one needs an exact metric name.
Onboarding, training documents and videos
Outcome
Reporting time reduced
Weekly active users
Scattered tools consolidated into four products
- Easier market insights: real-time views for quick scanning, with clear paths into trends and Lyft-wide performance
- A connected ecosystem: seamless navigation across products with a consistent structure
- Better findability: global search that locates metrics even without exact names
- Closed the training gap: announcements, guided onboarding, docs and videos to help people ramp faster
- More trust: unified branding and consistent design patterns across the experience
Learnings
Ranking dashboards by real traffic settled navigation debates faster than opinions and kept high-value tools up front.
Leaders want a pulse, analysts want depth. Grouping products by intent (Monitor versus Explore) served both without compromise.
Consistent names across products, and renaming Health to Explore, did as much for findability as any new screen.
Shipping the dashboard in stages over six months let the team learn from each release before taking on the suite.
As long as you are honest and sincere, and follow the process, nothing will go wrong.