Analytics · data room
Where the exposure sits, and where the delay hides.
Two real datasets, fully anonymised: a transformer maker’s book of 207 trips and the 242 ODC and OWC inspections of 2025. Every chart below is computed from them. Hover a mark for its value.
Portfolio view
A transformer maker’s book: 207 trips, 54 routes, 10 carriers.
All trips start in Pune. Exposure is the declared value at risk. Routes are scored on exposure per trip and distance tail.
Route concentration curve
How quickly exposure accumulates as routes are ranked from largest to smallest.
So what. Ten of 54 routes carry 60.87% of exposure, so route-level terms such as deductibles and escort rules reach most of the money with few rules.
Top 12 routes by exposure
Orange marks routes with a risk score of 60 or more.
So what. A single trip to Odisha sits at the top: high value over about 1,560 km gives it the highest risk score on the book.
Route risk explorer
Distance tail against exposure; bubble size is trip count. Hover for detail.
So what. Risk sits at the long-distance, high-value corner, where there is only one or two trips of history: exactly where an inspection record is the only evidence.
Vehicle concentration curve
Cumulative exposure share as vehicles are ranked from largest to smallest.
So what. The book is spread across 190 vehicles: the top vehicle holds 3.4% and the top 10 hold 16.53%. No single-vehicle dependency, which is a good sign.
Carrier exposure
Ten carriers; names replaced by letters in exposure order.
So what. Carrier A moved 61% of trips and holds 52% of exposure. That is a single-carrier dependency worth a governance control.
Carrier exposure: the table
| Carrier | Trips | Exposure (₹ Cr) | Average per trip (₹ Cr) | Routes |
|---|---|---|---|---|
| Carrier A | 127 | 117.17 | 0.92 | 38 |
| Carrier B | 26 | 42.01 | 1.62 | 9 |
| Carrier C | 15 | 24.06 | 1.60 | 8 |
| Carrier D | 14 | 16.89 | 1.21 | 6 |
| Carrier E | 13 | 10.11 | 0.78 | 7 |
| Carrier F | 3 | 5.59 | 1.86 | 3 |
| Carrier G | 5 | 4.88 | 0.98 | 1 |
| Carrier H | 1 | 3.57 | 3.57 | 1 |
| Carrier I | 1 | 1.46 | 1.46 | 1 |
| Carrier J | 2 | 0.92 | 0.46 | 1 |
Carrier by route hotspots
The ten pairs holding the most exposure.
So what. Five of the top ten pairs belong to Carrier A. Carrier-specific route terms matter more than a blanket rule.
Consignee exposure
Top ten of 56 consignees; names replaced by letters.
So what. The top three consignees account for 28% of exposure. Consignees tend to be the stable node while carriers change, which makes them natural partners for loss-prevention spend.
Consignee exposure: the table
| Consignee | Trips | Exposure (₹ Cr) | Carriers |
|---|---|---|---|
| Consignee A | 14 | 26.38 | 4 |
| Consignee B | 20 | 21.48 | 3 |
| Consignee C | 9 | 15.84 | 1 |
| Consignee D | 16 | 14.73 | 1 |
| Consignee E | 40 | 14.17 | 1 |
| Consignee F | 9 | 9.03 | 1 |
| Consignee G | 3 | 7.85 | 1 |
| Consignee H | 7 | 7.76 | 2 |
| Consignee I | 5 | 7.29 | 2 |
| Consignee J | 3 | 6.45 | 2 |
Route table
The 25 largest routes with their risk score.
| Route | Trips | Exposure (₹ Cr) | P90 distance (km) | Risk score |
|---|---|---|---|---|
| Pune → Surat | 18 | 32.38 | 416 | 40.5 |
| Pune → Ahmedabad | 45 | 19.81 | 648 | 9.5 |
| Pune → West Nimar | 9 | 15.84 | 601 | 42.1 |
| Pune → Kurnool | 16 | 14.73 | 689 | 21.7 |
| Pune → Kachchh | 9 | 12.09 | 981 | 34.3 |
| Pune → Shajapur | 9 | 9.87 | 739 | 26.5 |
| Pune → Gulbarga | 7 | 9.54 | 378 | 29.8 |
| Pune → Bangalore | 9 | 9.42 | 842 | 25.7 |
| Pune → Jamnagar | 7 | 7.55 | 953 | 27.1 |
| Pune → Panipat | 4 | 6.72 | 1,611 | 46.5 |
| Pune → Hoshangabad | 7 | 6.25 | 872 | 21.9 |
| Pune → Tumkur | 5 | 5.46 | 771 | 26.6 |
| Pune → Sundergarh | 2 | 5.44 | 1,405 | 74.8 |
| Pune → Angul | 3 | 4.23 | 1,480 | 38.4 |
| Pune → Golaghat | 3 | 4.09 | 2,877 | 40.5 |
| Pune → Ludhiana | 3 | 3.79 | 1,812 | 35.1 |
| Pune → Dhenkanal | 1 | 3.57 | 1,559 | 100.0 |
| Pune → Vadodara | 2 | 3.39 | 548 | 39.9 |
| Pune → Bellary | 4 | 3.30 | 610 | 18.9 |
| Pune → Raigarh(MH) | 1 | 3.26 | 135 | 60.5 |
| Pune → Durhapur | 1 | 2.62 | 2,000 | 75.5 |
| Pune → Durgapur | 1 | 2.62 | 2,000 | 75.5 |
| Pune → Bardhaman | 1 | 2.62 | 1,956 | 75.3 |
| Pune → West Midnapore | 2 | 2.54 | 1,840 | 35.5 |
| Pune → Kolkata | 1 | 2.52 | 1,953 | 72.4 |
2025 inspection dataset
242 inspections, 28 Jan 2025 to 20 Dec 2025.
17 carriers, 89 consignees and 71 routes. Delay is the absolute gap in days between factory stuffing and the lorry receipt.
Inspections by month, 2025
Volume built through the year and peaked in September.
So what. Inspections grew from one in January to 43 in September: the dataset is dominated by the second half of the year.
Delay distribution
Mean and P90 against the tail. Orange bars sit beyond P90.
So what. The mean is 1.41 days and P90 is 2, but the tail runs to 11 days. Averages hide the movements that cost money.
Concentration: shipments against exposure
Herfindahl-Hirschman index; 1,500 and 2,500 mark moderate and high.
So what. Carrier concentration is high: 2,518 by shipments against 2,233 by exposure. Operational dependence is greater than financial dependence. Routes (408) and consignees (347) are well spread.
Frequency against exposure, by carrier
Each bubble is a carrier. Size shows P90 delay; orange is 4 days or more.
So what. Frequency and exposure rise together for the big carriers, but two smaller carriers (03 and 09) carry the worst tails, with P90 delays of 4.5 and 5.5 days. Volume alone does not rank risk.
Busiest routes
Orange routes have a P90 delay of 3 days or more.
So what. Most busy routes hold a P90 of 1 to 2 days. Pune → Surat is the exception at 2.6 days over 15 inspections.
Busiest routes: the table
| Route | Inspections | Exposure (₹ Cr) | Mean delay (d) | P90 delay (d) |
|---|---|---|---|---|
| Pune → Ahmedabad | 33 | 12.64 | 1.3 | 2.0 |
| Pune → Surat | 15 | 31.33 | 1.6 | 2.6 |
| Pune → Kurnool | 14 | 14.17 | 1.57 | 2.0 |
| Vadodara → Nalgonda | 12 | 14.35 | 1.0 | 1.0 |
| Pune → Shajapur | 9 | 10.86 | 1.33 | 2.0 |
| Vadodara → Bangalore | 9 | 8.44 | 1.33 | 2.0 |
| Vadodara → Surat | 8 | 5.80 | 1.0 | 1.0 |
| Vadodara → Thane | 8 | 4.37 | 1.0 | 1.0 |
| Pune → West Nimar | 8 | 15.62 | 2.25 | 3.0 |
| Pune → Tumkur | 7 | 7.72 | 1.43 | 2.0 |
| Pune → Bangalore | 7 | 8.84 | 1.57 | 2.0 |
| Vadodara → Jaipur | 5 | 4.08 | 0.8 | 1.0 |
Routes with the worst tail
Ranked by P90 delay. Counts are in the tooltip.
So what. Pune → Golaghat reaches a P90 of 10.7 days from 2 inspections. Thin routes can have the worst tails, though one or two inspections are a small sample.
Carrier by route heatmap
Top six carriers by eight busiest routes; cell value is P90 delay in days.
So what. On Pune → Surat the P90 runs from 1.0 day for one carrier to 3.8 days for another. Delay sits in the pair, not in the route alone. Inspect the pair.
Consignee concentration curve
How quickly exposure accumulates across consignees.
So what. The top ten consignees hold 48% of exposure and the top 30 hold 76%: demand-side exposure is broad but not flat.
Inspection status
Where each inspection stood in the workflow.
So what. Seven in ten inspections are approved; about one in seven files waits on documents, the clearest process bottleneck in the data.
Exposure per inspection
Distribution in ₹ lakh.
So what. The median inspection carries about ₹87.0 lakh, but a long right tail goes past ₹300 lakh.
Carrier profile
Frequency, exposure and delay for every carrier.
| Carrier | Inspections | Exposure (₹ Cr) | Mean delay (d) | P90 delay (d) | Routes |
|---|---|---|---|---|---|
| Carrier 01 | 106 | 98.76 | 1.48 | 2.0 | 32 |
| Carrier 02 | 52 | 45.87 | 1.08 | 1.0 | 14 |
| Carrier 03 | 18 | 33.93 | 2.33 | 4.5 | 8 |
| Carrier 04 | 11 | 9.53 | 1.09 | 1.0 | 7 |
| Carrier 05 | 10 | 15.51 | 1.3 | 2.0 | 4 |
| Carrier 06 | 9 | 7.89 | 1.11 | 1.2 | 6 |
| Carrier 07 | 8 | 6.54 | 0.88 | 1.3 | 6 |
| Carrier 08 | 6 | 9.66 | 1.17 | 1.5 | 5 |
| Carrier 09 | 6 | 2.85 | 2.83 | 5.5 | 5 |
| Carrier 10 | 5 | 1.85 | 1.0 | 1.0 | 3 |
| Carrier 11 | 3 | 3.22 | 2.0 | 2.0 | 1 |
| Carrier 12 | 2 | 1.22 | 1.0 | 1.0 | 1 |
| Carrier 13 | 2 | 2.96 | 1.0 | 1.0 | 2 |
| Carrier 14 | 1 | 1.60 | 2.0 | 2.0 | 1 |
| Carrier 15 | 1 | 0.87 | 1.0 | 1.0 | 1 |
| Carrier 16 | 1 | 0.22 | 2.0 | 2.0 | 1 |
| Carrier 17 | 1 | 4.23 | 1.0 | 1.0 | 1 |
Data quality
Seven quality gates, all passed.
| Gate | Status | Result |
|---|---|---|
| Exposure value is non-negative | PASS | 0 negative rows |
| Exposure value is present | PASS | 207 of 207 |
| Origin and destination present | PASS | all rows |
| Distance present | PASS | 207 of 207 |
| Near-zero distance check (under 5 km) | PASS | 0 rows |
| No duplicate e-way bill keys | PASS | 0 duplicates |
| Geo columns present | PASS | all four |
Method and limits
What is measured, and what is not claimed.
Anonymised
Consignors, consignees, carriers, insurers, vehicles and drivers are replaced by labels (Carrier A to J, Carrier 01 to 17, Consignee A to J, Consignee 01 to 89) assigned in rank order. Vehicle numbers, driver details and document numbers are never shown. Cities are kept.
Measures
Exposure is declared value at risk. HHI is the sum of squared percentage shares. P90 is the delay that nine in ten movements beat. Route risk scores blend exposure per trip and distance tail.
Limits
This is not a claims predictor or a loss-probability model. Small routes have few inspections, so their tails are indicative. The two datasets are different samples and should not be added together.
Counts and values were computed by iCargoFinsure from the anonymised files supplied. The research paper explains the framework behind them. Read the research.
See your own book
Run these views on your inspections.
Every inspection on the platform feeds the same concentration, tail and hotspot views.