Fill in these first
- Shipment data: [paste as order ID, courier, origin, destination city or pin code, ship date, promised date, delivered date, attempts, final status]
- Regions or bands to group by: [e.g. metro, tier 2, tier 3 and rural, or states]
- Promise you make to customers: [e.g. 3 to 5 days for metros, 7 days elsewhere]
- Payment mix, if known: [share of cash on delivery orders]
- Cost per shipment by courier, if known: [rates]
- Period: [dates]
Task
Score each courier in each region and recommend how to split shipments between them.
Measures
- On-time rate: delivered on or before the promised date, as a share of delivered orders.
- Average and 90th percentile delay in days for late orders.
- First-attempt delivery rate and average attempts.
- Return-to-origin rate, split by cash on delivery and prepaid where the data allows.
- Cost per delivered order, if rates are given.
Output format
- | Courier | Region | Shipments | On-time % | Avg delay | 90th pct delay | First attempt % | RTO % | Cost per delivered order |
- The three biggest problems in one line each, with the numbers behind them.
- | Region | Recommended courier | Backup courier | Reason |
- Questions to raise with each courier, backed by specific lanes and figures.
Rules
- Do not rank a courier on fewer than 30 shipments in a region; mark it "too few to judge".
- Count orders with no delivered date as open, not on time.
- Separate courier delays from late dispatch on our side when ship dates show the parcel left late.
- Do not invent costs or service levels that are not in the inputs.






