Upload Address Data
Upload a CSV file with delivery addresses. The file can include lat/lon coordinates, or raw street addresses (we'll geocode them for free).
Addresses need geocoding (converting to lat/lon). This uses the free US Census Bureau API.
How This Analysis Works
Geographic Clustering
Addresses are grouped into geographic "zones" using K-means clustering (~50 addresses per zone). This mirrors how real delivery territories work.
Road Distance Matrix
For each zone, we compute actual driving distances between every pair of addresses using OSRM (real road network data from OpenStreetMap).
Route Optimization (VRP)
Google OR-Tools solves the Vehicle Routing Problem — finding the shortest route through all addresses while respecting the max hours per driver.
Scenario Comparison
We repeat the entire analysis for different subscriber counts (100%, 50%, 30%, etc.) to prove that costs don't scale linearly with subscribers.
Configure Analysis
Adjust the simulation parameters below. Defaults are optimized for newspaper delivery in Las Vegas.
Select which subscriber levels to simulate
Haversine: Straight-line distance multiplied by 1.35x road factor. Fast but approximate. Good for quick testing.
Analyzing Routes...
Running VRP optimization for each subscriber scenario. This may take a few minutes.
Initializing...
Key Findings
Distance vs. Subscriber Count
How total driving miles change as subscriber count drops
| Population | Addresses | Miles | % of Full | Cost Ratio |
|---|
Distance vs. Population
Actual distance % vs. what it would be if costs scaled linearly
Time-of-Day Comparison
Full population: night delivery vs afternoon delivery
| Scenario | Drive Hrs | Total Hrs | Drivers |
|---|
Night vs. Afternoon
Breakdown of driving time and stop time by delivery window
Delivery Address Map
Color-coded by delivery zone (cluster). Each color represents one geographic group.
Export Reports:
Methodology & Technical Details
Routing Engine
OSRM (Open Source Routing Machine) using OpenStreetMap road network data. Provides actual driving distances and durations based on real road geometry, speed limits, and turn restrictions.
Route Optimization
Google OR-Tools Vehicle Routing Problem (VRP) solver with time window constraints. Same algorithm family used by Amazon, UPS, and FedEx. Each vehicle (driver) is limited to the configured shift duration.
Geographic Clustering
K-means clustering partitions addresses into delivery zones of ~50 addresses each. This mirrors real-world delivery territory management and keeps computation within API limits.
Time-of-Day Traffic Model
Road-class-specific speed profiles: Residential streets 90% of free-flow at 2 PM, Arterials (Flamingo, Sahara, etc.) 65%, Highways (I-15, US-95) 72%. Based on Nevada DOT traffic monitoring data and FHWA HPMS standards.
Delivery Stop Model
30 seconds per delivery stop (configurable). This accounts for vehicle deceleration, newspaper delivery, and re-acceleration typical of residential newspaper delivery by vehicle.
Driver Count Calculation
VRP solver automatically determines minimum drivers needed. Each driver's total route time (driving + stops) is constrained to the configured delivery window. This provides an operationally accurate driver count.