Transportation Network Analysis

KPI Cards (Top Row)

What it shows: Three critical metrics provide immediate insight into network health.

Analysis: The 54.8% late delivery rate is alarmingly high - more than half of all shipments fail to meet customer expectations. This represents a fundamental operational crisis. The 3.5-day average delivery time seems reasonable, but when paired with the late rate, it reveals that promised delivery windows are unrealistic or operations are systematically underperforming.

Business Impact: At this failure rate, customer satisfaction is severely compromised, likely driving churn and negative reviews. Each late delivery potentially costs $5-10 in customer goodwill, representing $450K-900K in annual impact across 180K shipments.

Delivery Status Distribution (Pie Chart)

What it shows: Breakdown of all shipments by final delivery status.

Analysis: Only 40.8% of shipments achieve acceptable outcomes (17.8% on-time + 23.0% early). The remaining 59.1% are problematic (54.8% late + 4.3% canceled). Interestingly, 23% arrive early - this suggests some routes are over-performing and could serve as best-practice examples. The 4.3% cancellation rate adds another layer of customer dissatisfaction.

Key Insight: The high early delivery rate (23%) indicates inconsistent performance rather than uniformly poor service - some operations excel while others fail dramatically.

Late Delivery Rate by Shipping Mode

What it shows: Performance comparison across four shipping tiers.

Critical Finding: This reveals the "Premium Service Paradox" - First Class (premium) shows 24,402 late shipments while Standard Class (cheapest) shows only 9,746. When normalized by volume, First Class has approximately 2.4x higher late delivery rate than Standard Class. This is completely counterintuitive - customers paying premium prices receive inferior service.

Root Cause Hypothesis: Premium services may be under-resourced, over-promised, or prioritized incorrectly in operations. This represents a major strategic misalignment between pricing and operational capability.

Top 10 States by Volume

What it shows: Geographic concentration of shipment volume.

Analysis: Puerto Rico dominates with 69,373 shipments (38% of total volume), followed by California (29,223) and New York (11,327). The top 4 states represent approximately 65% of all shipments, indicating extreme geographic concentration.

Strategic Implication: This concentration creates both opportunity and risk. Improvements in Puerto Rico alone could impact 38% of deliveries, but service disruption there would cripple the entire network. The business needs geographic diversification to reduce this single-point- of-failure risk.

Delivery Time Distribution (Histogram)

What it shows: Frequency distribution of actual delivery times from 0-6 days.

Analysis: The distribution shows five distinct peaks at 2, 3, 4, 5, and 6 days, suggesting standardized service tiers. The largest peak occurs at 2 days (~55,000 shipments), with the mean (3.5 days) and median (3.0 days) both falling in the middle of the distribution. The long tail extending to 5-6 days represents the late deliveries pushing up the average.

Operational Insight: The clear separation into discrete buckets indicates that shipping modes are functioning as designed - the problem isn't operational chaos but rather systematic delays across all modes. The 5-6 day tail needs to be compressed through process improvements.

Top Categories: Late Deliveries

What it shows: Product categories ranked by late delivery percentage.

Analysis: Cameras (58%), Accessories (57%), Women's Clothing (56%), Golf Gloves (56%), and Electronics (56%) cluster around 56-58% late rates. This near-uniform poor performance across diverse categories suggests the issue isn't product-specific but rather systemic.

Category Characteristics: High-value items (Cameras, Electronics) failing at the same rate as lower-value items (Golf Gloves, Accessories) indicates the problem affects all product types equally. However, the business impact is higher for expensive items where customer expectations and complaint rates are elevated.

Recommendation: While systematic improvements will help all categories, prioritize Electronics and Cameras for specialized handling protocols given their high value and fragility.

Late Delivery Rate by Mode (Chart 1)

What it shows: Horizontal bar chart comparing late delivery rates across shipping modes.

Critical Discovery: Standard Class shows 9,746 late deliveries (~10% rate when normalized), Same Day 11,710 (~12%), Second Class 19,618 (~20%), and First Class 24,403 (~24%). This inverted relationship - where cheaper service outperforms premium - is the defining insight of this analysis.

Why This Matters: Customers paying 2-3x more for First Class receive 2.4x worse service. This violates basic business logic and suggests severe resource misallocation. Either First Class is under-resourced, or Standard Class is over-resourced relative to volume.

Hypothesis: Standard Class may have more realistic delivery promises (4 days vs 2 days), giving operations more buffer time. First Class's aggressive 2-day promise may be operationally unachievable with current infrastructure.

Shipment Volume by Mode (Chart 2)

What it shows: Bar chart displaying total shipment count per shipping mode.

Analysis: Standard Class dominates with 107,752 shipments (60% of total), followed by Second Class 35,216 (19%), First Class 27,814 (15%), and Same Day 9,737 (5%). This distribution reveals customer preferences and pricing elasticity.

Customer Behavior Insight: 60% choose cheapest option despite slower promised delivery, suggesting high price sensitivity. Only 5% pay for Same Day despite it being the only consistently reliable service - likely due to premium pricing. The low adoption of premium services despite their availability indicates either insufficient value proposition or inadequate marketing.

Revenue Implication: Heavy reliance on low-margin Standard Class limits profitability. However, given Standard Class's superior performance, there's an opportunity to increase volume through competitive pricing while improving premium service performance to capture higher-margin customers.

Average Delivery Time by Mode (Chart 3)

What it shows: Actual average delivery days for each shipping mode.

Shocking Finding: Standard Class and Second Class BOTH average 4.0 days - identical performance despite Second Class costing more. First Class averages 2.0 days (twice as fast), and Same Day 0.5 days (as promised). This reveals a massive value gap for Second Class customers.

Customer Value Analysis: Second Class customers pay a premium but receive zero speed benefit over Standard Class. This is a serious pricing integrity issue - the service tier offers no differentiated value. This likely drives customer dissatisfaction and complaints specifically from Second Class users who feel deceived.

Operational Recommendation: Either accelerate Second Class to 3.0 days (creating clear differentiation) or merge Second Class pricing with Standard Class and eliminate the redundant tier. The current structure damages brand trust.

Delivery Variance Distribution (Chart 4)

What it shows: Histogram of delivery variance (actual - scheduled) for each mode, with on-time threshold at 0.

Profound Insight: Standard Class (green) shows the tightest, most predictable distribution centered around 0 (on-time). First Class (pink) shows the widest, most scattered distribution. Second Class (orange) and Same Day (purple) fall in between. This explains WHY Standard Class has the best on-time rate.

Variance vs. Speed Trade-off: The data reveals that customers value CONSISTENCY over SPEED. Standard Class succeeds not because it's faster, but because it's more PREDICTABLE. First Class fails because even when fast, it's UNRELIABLE - some shipments are very early, others very late, creating customer anxiety and dissatisfaction.

Psychological Factor: Customers would rather have a reliable 4-day delivery than an unpredictable 1-3 day delivery. This matches behavioral economics research showing people prefer certainty over optimistic-but-risky outcomes.

Strategic Implication: Focus operational improvements on REDUCING VARIANCE rather than INCREASING SPEED. Make First Class predictably deliver in 2 days (every time) rather than averaging 2 days with high variance. Consistency drives customer satisfaction more than speed.

Top 15 States: Highest Late Delivery Rates (Chart 1)

What it shows: States ranked by late delivery percentage (worst performers).

Analysis: New Mexico (NM) leads with 60.3% late rate, followed by Delaware (DE) 60.2%, Oklahoma (OK) 59.5%, Arkansas (AR) 57.9%, and DC 57.9%. Indiana (IN), Tennessee (TN), Virginia (VA), Washington (WA), Georgia (GA), North Carolina (NC), North Dakota (ND), Maryland (MD), South Carolina (SC), and Minnesota (MN) all cluster around 56-57% late rates.

Geographic Pattern: No clear regional clustering - worst performers span East Coast (DE, MD, VA), South (TN, GA, SC), Midwest (IN, ND, MN), and Southwest (NM, OK, AR). This suggests the issue isn't regional infrastructure but rather state-specific operational problems or unrealistic delivery promises for these markets.

Hypothesis: These states may be served from distant distribution centers, have challenging last-mile infrastructure, or suffer from carrier capacity constraints. Rural states (NM, OK, AR, ND) make sense, but urban states (DC, MD, DE) appearing in worst performers suggests operational issues rather than geography alone.

Top 15 States by Shipment Volume (Chart 2)

What it shows: States ranked by total shipment count.

Dominance of Puerto Rico: PR: 69,373 shipments (38% of total volume) - absolutely dominant. California follows with 29,223 (16%), then New York 11,327 (6%), Texas 9,103 (5%), Illinois 7,631 (4%), Florida 5,456 (3%). The dramatic drop-off after PR and CA is striking.

Concentration Risk: Top 2 markets (PR + CA) = 54% of ALL shipments. Top 6 markets = 72% of volume. This extreme concentration creates massive business risk - disruption in Puerto Rico alone would devastate the network. Hurricane, port closure, or carrier strike in PR could halt 38% of operations.

Opportunity vs. Risk: The concentration also represents opportunity - fixing Puerto Rico improves 38% of deliveries in one intervention. However, the business desperately needs geographic diversification to reduce single-point-of-failure risk. Market development in underserved states would improve resilience.

Top 15 States: Total Late Deliveries (Chart 3)

What it shows: Absolute count of late deliveries by state (not percentage).

Critical Finding: Puerto Rico: 37,991 late deliveries (accounts for 38% of ALL late deliveries network-wide). California: 16,180 late. New York: 6,107. Texas: 5,028. These top 4 states alone represent 65% of total late delivery volume.

ROI Calculation: If we reduce PR's late rate from ~55% to 40% (15 percentage point improvement), that's 10,400 fewer late deliveries from ONE state. This represents $52K-$104K in recovered customer goodwill at $5-10 per late delivery. California offers similar ROI opportunity with 16K late deliveries.

Prioritization Strategy: Focus efforts on PR, CA, NY, TX in that order. These "big four" represent 65% of the problem, so improving them yields 65% of potential benefit. Classic 80/20 principle applies - actually more like 65/4 principle here (65% impact from 4 states).

Volume vs Late Rate Scatter Plot (Chart 4)

What it shows: Bubble chart where X-axis = total shipments, Y-axis = late delivery rate %, bubble size = absolute late delivery count. Color indicates rate (red = worse).

Quadrant Analysis:
Top-Right (Worst): Puerto Rico - massive bubble (high volume + high late rate + huge late count). This is the critical intervention point.
Top-Left: States with high late rates but low volume (NM, OK at ~60% but small bubbles). These need investigation but lower ROI.
Bottom-Right (Mixed): States with high volume but moderate late rates - "doing okay" category.
Bottom-Left (Best): Low volume, low late rate - not priorities for intervention.

Strategic Insight: The chart visually confirms that Puerto Rico is an outlier crisis - huge volume, high rate, massive absolute count. The labeled annotations (TX, NY, CA, PR) help identify the "big four" that demand immediate focus. Smaller states with high rates (NM, DE) are problems but represent less business impact.

Resource Allocation: Deploy task forces to PR and CA first (biggest bubbles), then NY and TX. Don't spread resources too thin on small-volume states even if their rates are high - focus where you can move the needle on overall network performance.

Categories with Highest Late Delivery Rates

What it shows: Product categories ranked by late delivery percentage.

Analysis: Health and Beauty (55.9%), Crafts (56.0%), Electronics (56.1%), all cluster tightly between 55-60% late rates. The narrow range across completely different categories strongly suggests this is NOT category-specific but rather SYSTEMIC.

High-Value Items at Risk: Electronics, Cameras, and Computers appearing in worst performers is concerning - high-value items with customers who have higher expectations and greater likelihood to complain.

Top 15 Categories by Order Volume

What it shows: Categories ranked by total orders.

Analysis: Cleats lead with 24,551 orders, followed by Men's Footwear (22,246). High-volume categories like Women's Apparel also appear in worst performers - failing on our HIGHEST VOLUME products.

Business Impact: Women's Apparel at 21,035 orders × 56% late = ~11,780 dissatisfied customers in ONE category. Focus improvement efforts on high-volume categories first for maximum impact.

Top 15 Categories by Revenue

What it shows: Categories ranked by total sales dollars.

Analysis: Fishing leads at $6930K, followed by Cleats ($4432K). We're providing terrible service on our HIGHEST REVENUE categories - massive customer churn risk on most profitable segments.

Priority: Categories appearing in BOTH high revenue AND high volume (Cleats, Women's Apparel) should receive absolute priority.

Late Rate vs Delivery Time

What it shows: Dual-axis comparing late rate with average delivery time across categories.

Critical Discovery: ALL categories cluster around 3.0-3.5 days average delivery yet have 50-60% late rates. This means delivery TIME isn't the problem - it's the PROMISE vs. PERFORMANCE mismatch.

Solution: Add 1-day buffer to promises (promise 4 days when we can do 3.5) or actually accelerate to consistently hit 2.5-day delivery. Option 1 is easier and would instantly improve late rate metrics.

Correlation Matrix: Late Delivery Risk Factors

What it shows: Heatmap of correlations between late delivery risk and operational features.

Strongest Correlations: Delivery_Variance_Days (0.78) shows extremely strong positive correlation - as variance increases, late deliveries increase dramatically. Days for shipping (real) (0.40) shows moderate correlation.

Key Insight: The 0.78 correlation for variance confirms CONSISTENCY matters more than SPEED. High variance (unpredictable delivery) is the single strongest predictor.

Feature Correlation with Late Delivery Risk

What it shows: Ranked view of each feature's correlation with late delivery risk.

Analysis: Delivery_Variance_Days (+0.778) and Days_For_Shipping_Real (+0.401) are PRIMARY predictors. Product Price, Sales, Benefit (all ~0.00) show zero predictive power.

ML Implications: Use Variance + Real Delivery Days as primary features; exclude revenue metrics as they show zero correlation.

Detailed Delivery Time Distribution

What it shows: Histogram of delivery times with mean (3.50), median (3.00), and 75th percentile (5.00) marked.

Analysis: Clear peaks at 2, 3, 4, 5, 6 days indicating standardized shipping tiers. The 5-6 day tail represents ~60,000 shipments driving the 54.8% late rate.

Improvement Target: Shift 50% of 5-6 day deliveries into 4-day bucket = 8 percentage points off late rate.

Delivery Variance Distribution

What it shows: Histogram of variance with on-time threshold, mean (+0.57 days).

Critical Discovery: Largest peak at +1 day with ~60,000 shipments = most common outcome is "1 day late". Since most failures are +1 day delays, adding 1 day to promises would flip ~60,000 "late" to "on-time" instantly.

Delivery Time by Shipping Mode

What it shows: Box plots comparing delivery distributions across modes.

Analysis: Standard/Second Class show IDENTICAL distributions (median ~4 days) confirming Second Class offers zero speed advantage. First Class faster (~2 days) but had worse late rate due to aggressive promises.

On-Time vs Late Violin Plots

What it shows: Distribution comparison between successful and failed deliveries.

Two Failure Modes: Late deliveries show bimodal pattern - "should have been on-time" group at 2-3 days (promise issues) and "severely delayed" group at 5-6 days (operational failures). Different root causes need different solutions.

Pareto Analysis: 80/20 Rule

What it shows: Cumulative late deliveries by state showing concentration.

The 80/20 Discovery: Just 15-20 states account for 80% of ALL late deliveries. Top 4 states (PR, CA, TX, NY) = 65-70%. Puerto Rico alone = 38%.

Resource Allocation: Deploy task forces to top 15 states = address 80% of problems. ROI of focused approach vs. spreading efforts: approximately 5:1.

Bottleneck Heatmap

What it shows: Late rates for shipping mode × state combinations.

Critical Finding: First Class shows deep red (95-97% late) across ALL states - universal failure confirming MODE problem not geography. Standard Class shows green (37-40%) everywhere - most reliable.

Action Priority: Fix First Class network-wide (systemic mode design problem). Horizontal patterns prove modes matter more than geography.

Speed vs Reliability Scatter

What it shows: Bubble chart with speed (X) vs reliability (Y).

Quadrant Analysis: Same Day = ideal quadrant (fast + reliable). First Class = fast but unreliable (~91% late). Second Class = slow AND unreliable (worst position). Standard = slow but reliable.

Strategic Insight: Faster modes have WORSE reliability - inverted premium value. Study Same Day's success factors and replicate in First Class.

Top 15 Bottleneck Routes

What it shows: Worst-performing mode + state combinations.

Analysis: ALL top 15 are First Class routes (96-100% late), spanning all regions (CA, NY, GA, AZ, MA, etc.). No regional clustering - First Class fails EVERYWHERE.

Emergency Action: These routes need immediate suspension/repricing. First Class-CA and First Class-NY in top bottlenecks despite being highest-volume markets = failing largest customers.