Lyft Express Drive Analytics - Business analytics dashboard for rideshare operations
Business AnalyticsRideshare Operations

LYFT EXPRESS DRIVE DASHBOARD

A comprehensive business analytics dashboard tracking the health of Lyft's Express Drive Program, analyzing regional fleet distribution, driver activity patterns, financial performance, and lifetime value across NYC, Chicago, and Philadelphia markets.

VIEW LIVE DASHBOARD →

Interactive dashboard built with Mode Analytics (a business intelligence platform) - click to explore the data yourself

PROJECT OVERVIEW:

What is Express Drive? Lyft's Express Drive Program is a vehicle rental service that allows people without personal cars to rent vehicles and drive for Lyft. Think of it as "car rental meets rideshare" - drivers pay a weekly rental fee (~$255) to access a vehicle, then earn money by giving rides to passengers.

This dashboard analyzes the business health of this program across three major markets (NYC, Chicago, and Philadelphia), answering questions like: Are drivers making money? Is the fleet being used efficiently? How valuable are these drivers to Lyft long-term?

Platform:

Mode Analytics

Data Sources:

Lyft Operations

Markets:

NYC, CHI, PHI

Drivers:

9,874 unique

THE BUSINESS CHALLENGE:

Why does this matter? Lyft has thousands of Express Drive vehicles and drivers across multiple cities. Without proper data analysis, it's impossible to know if the program is profitable, if drivers are happy, or where improvements are needed. This project analyzes real operational data to answer critical business questions:

  • •How is fleet utilization performing across NYC, Chicago, and Philadelphia markets?
  • •Are there data quality issues affecting driver-vehicle relationship tracking?
  • •What are the true financial outcomes for drivers after rental costs?
  • •How can we calculate lifetime value to optimize driver acquisition and retention?

The challenge? Express Drive data comes from multiple computer systems that don't always talk to each other perfectly. Driver reservations, vehicle assignments, and earnings data all live in different databases. This project involved cleaning and connecting this messy data to create reliable business insights that Lyft can actually use.

ANALYTICAL APPROACH:

Rather than providing surface-level metrics, this dashboard takes a comprehensive approach to Express Drive program analysis with four key focus areas:

1

REGIONAL FLEET DISTRIBUTION

Analyze vehicle distribution across NYC, Chicago, and Philadelphia markets with data quality checks for regional mapping accuracy.

2

DRIVER-VEHICLE ACTIVITY

Track daily fleet activity with sophisticated data cleaning to ensure accurate 1:1 driver-vehicle relationships and identify data integrity issues.

3

FINANCIAL PERFORMANCE

Calculate true driver earnings by deducting prorated rental costs ($255/week) from platform revenue to understand real financial outcomes.

4

LIFETIME VALUE MODELING

Build churn-based LTV model incorporating platform commission, Express Drive haircut, and retention assumptions to calculate driver value.

The following sections detail each analysis area with specific methodologies, findings, and business insights derived from the Express Drive program data.

DATA STRUCTURE & METADATA:

Dataset Metadata View - Mode Analytics metadata view showing dataset structure, column descriptions, and data types

Mode Analytics metadata view showing dataset structure, column descriptions, and data types

💡

DESIGN DECISION: Metadata Visibility

I included the metadata section prominently because it allows users to quickly understand the data structure, column meanings, and data sources without diving into technical documentation. This transparency builds trust and helps stakeholders understand exactly what metrics are being tracked and how they're defined.

KEY VISUALIZATIONS & INSIGHTS:

🗺️ REGIONAL FLEET DISTRIBUTION

Regional Metrics Dashboard - Screenshot showing vehicle distribution pie chart and data table

KEY FINDINGS

  • • NYC dominates with 4,311 vehicles (49.2%)
  • • Chicago has 2,489 vehicles (28.4%)
  • • Philadelphia operates 1,784 vehicles (20.4%)
  • • 465 vehicles lack proper region mapping

DATA INTEGRITY PROCESS

  • • Validated 1:1 vehicle_id mapping by comparing distinct IDs to total rows
  • • Standardized inconsistent region naming conventions
  • • Created separate bucket for missing region data
  • • Flagged vehicles for downstream data cleanup efforts
🔍

ANALYTICAL APPROACH: Regional Data Standardization

Started by validating vehicle_id integrity, comparing distinct IDs to total rows to identify duplicates. Discovered inconsistent region naming conventions and standardized them into NYC, CHI, and PHI buckets. Created a separate category for vehicles with missing region data - this became a valuable data quality metric for future cleanup efforts and highlighted potential gaps in the data collection process.

🚗 DRIVER & FLEET ACTIVITY ANALYSIS

Fleet Activity Time Series - Screenshot of vehicle/driver count over time with cleaning logic
🔧

ANALYTICAL APPROACH: Solving the 1:1 Relationship Problem

Expected vs. Reality: Initially assumed that daily vehicle-driver relationships should be 1:1 (100 vehicles = 100 active drivers). However, raw data showed discrepancies with more drivers than vehicles on most days.

Root Cause Analysis: Discovered vehicles with multiple drivers during overlapping periods, likely due to reservation system issues where canceled or shortened reservations weren't updated retroactively. These represent data cleanliness issues rather than actual usage patterns.

Solution: Implemented "first reservation" filtering logic to clean the data. Created conflict flags for vehicles with multiple drivers and drivers with multiple vehicles. This produced the clean 1:1 relationship expected and turned data discrepancies into actionable data quality metrics.

RAW METRICS

  • • 6,467 avg daily vehicles
  • • 7,770 avg daily drivers
  • • 1,303 vehicle-driver variance

CLEANED METRICS

  • • 6,448 avg daily vehicles/drivers
  • • Perfect 1:1 relationship achieved
  • • First reservation rule applied

DATA QUALITY FLAGS

  • • 1,159 vehicles w/ multiple drivers
  • • 4 drivers w/ multiple vehicles
  • • Overlap detection implemented

💰 DRIVER FINANCIAL PERFORMANCE

Driver Financials Dashboard - Screenshot of revenue, rental costs, and earnings analysis

FINANCIAL METRICS

Average Revenue$822.92
Rental Cost (Prorated)$242.68
Net Earnings$580.24

KEY METHODOLOGY

  • ▲Filtered earnings to match active reservation periods
  • ●Prorated $255 weekly rental cost by days reserved
  • ▼Removed null earnings and invalid driver IDs
  • ♦Covered Express Drive drivers only
💰

ANALYTICAL APPROACH: Financial Performance Calculation

Initial Assumptions: Earnings data appeared to be raw amounts before deductions (only 1 negative value and few zeros found). Weekly rental cost of $255 sourced from research showing costs between $240-$270.

Data Validation: Matched earnings to active reservation periods to ensure accuracy. Created Express Drive flag by checking earnings against reservation windows - confirmed our sample only covered Express Drive drivers.

Cost Logic: Since earnings didn't include rental costs, manually deducted them by prorating weekly $255 cost based on actual reservation days. Critical for accurate net earnings since not all reservations last full weeks.

📈 DRIVER LIFETIME VALUE MODELING

LTV Analysis Dashboard - Screenshot of churn modeling and lifetime value calculations
💡

ANALYTICAL APPROACH: First-Time LTV Modeling

Data Limitations & Learning: Limited dataset made LTV calculation challenging, but this was my first time working through an LTV problem and found it genuinely enjoyable. Researched various LTV approaches and chose a simple, transparent model that still captured the core business value.

Revenue Model: Applied 17% Express Drive haircut (from research) to driver earnings, then took 20% commission on post-haircut revenue. Formula: Weekly Profit = (Driver Earnings × (1 - 0.17)) × 0.20

Churn Assumptions: Used 7.5% weekly churn rate from industry benchmarks since dataset lacked clear churn signals. Expected active weeks = 1 / churn rate = 13.33 weeks average driver lifetime.

💰 REVENUE MODEL BREAKDOWN

Weekly Profit Formula
Weekly Profit = Weekly Earnings × (1 - Express Drive %) × Commission %
Average Weekly Lyft Profit = $106.79
Expected Active Weeks Formula
Expected Active Weeks = 1 ÷ Churn Rate
Expected Active Weeks = 1 ÷ 0.075 = 13.33 weeks

📊 LIFETIME VALUE CALCULATION

LTV = Weekly Profit × Expected Active Weeks
LTV = $106.79 × (1 ÷ 0.075)
LTV = $106.79 × 13.33 = $1,423.82
Weekly Profit
$106.79
Lyft's profit per driver
Churn Rate
7.5%
Weekly (industry benchmark)
Expected Weeks
13.33
1 ÷ churn rate

LTV COMPONENTS

  • • 9,874 unique drivers analyzed
  • • $106.79 avg weekly Lyft profit
  • • 13.33 expected active weeks
  • • $1,423.82 average driver LTV

REVENUE MODEL

  • • 17% Express Drive haircut applied first
  • • 20% Lyft commission on post-haircut revenue
  • • Formula: (Earnings × 0.83) × 0.20
  • • 7.5% weekly churn rate assumed

FINAL RESULTS

Average Driver LTV
$1,423.82
Total value to Lyft per driver

TECHNICAL IMPLEMENTATION:

🔗

SQL QUERIES & DATA ANALYSIS

All SQL queries, data transformations, and technical implementation details are available in the live Mode Analytics dashboard.

🔍View Full Mode Analytics Dashboard

BUSINESS IMPACT & KEY FINDINGS:

🎯

BUSINESS IMPACT: Strategic Driver Program Insights

This analysis provided critical insights for optimizing Lyft's Express Drive program across multiple markets. Key findings included identifying data quality issues affecting 1,159 vehicles, calculating true driver profitability after rental costs, and building an LTV model showing $1,423.82 average driver value. These insights directly informed driver acquisition strategies and operational improvements.

🔍 DATA QUALITY INSIGHTS

  • ⚠️
    1,159 vehicles with multiple drivers detected
    Indicates reservation system overlap issues
  • 📍
    465 vehicles missing regional mapping
    Flagged for data cleanup and standardization
  • ✅
    Data cleaning achieved 1:1 driver-vehicle ratio
    From 7,770 drivers to 6,448 clean relationships

💰 FINANCIAL PERFORMANCE

  • 💵
    $580.24 average net driver earnings
    After $242.68 prorated rental costs
  • 📈
    $1,423.82 average driver lifetime value
    Based on 13.33 week retention model
  • 🗺️
    NYC dominates with 49.2% of fleet
    4,311 vehicles across 3 major markets

LESSONS LEARNED & FUTURE IMPROVEMENTS:

🔄 DATA QUALITY CHALLENGES

Inconsistent data collection from different sources required significant cleaning and normalization. Future iterations would benefit from standardized data ingestion pipelines and automated quality checks to reduce manual preprocessing time.

Key Issues to Address:

  • • Follow up on vehicles w/ no region mapping
  • • Investigate data feeding the vehicle reservation table
  • • Implement lookback period improvements
  • • Clarify earnings calculation methodology

📊 VISUALIZATION EFFECTIVENESS

Mode Analytics' business intelligence platform proved invaluable for rapid data exploration and visualization. The ability to iterate quickly on SQL queries and generate comprehensive charts revealed insights that would have been missed in traditional static reporting approaches.

Analysis Improvements:

  • • More thorough LTV model development
  • • Ability to verify key assumptions
  • • Enhanced (XD) driver insights & average lifespan
  • • Larger timeframe for trend analysis
🚀

PLANNED ENHANCEMENTS

Next steps include developing more sophisticated modeling approaches, expanding data coverage, and implementing automated monitoring systems for key business metrics.

Model Enhancements:
  • • Predictive modeling for driver retention
  • • Regional performance optimization
  • • Enhanced LTV calculations with cohort analysis
Data Expansion:
  • • Tie more granular regional data
  • • Integrate additional driver behavioral metrics
  • • Implement real-time data quality monitoring

TECHNICAL STACK:

Platform & Tools

  • • Mode Analytics (Primary Platform)
  • • SQL (Data Analysis & Queries)
  • • Statistical Analysis

Data Sources & Visualization

  • • Lyft Express Public Mode Data
  • • Lyft Express Drive FAQ