Skip to content
PRS India

Hub-and-spoke · ePOD · COD reconciliation

Logistics Platform Development

Logistics platforms fail in the money, not the map. We build hub-and-spoke networks, driver apps that work offline, and a COD ledger. It tells you exactly how much cash is in the field and who is holding it.

Last updated

  • COD as a ledger, not a flag
  • Offline-first driver app
  • E-way bill expiry alerting
  • Scans as immutable events
The real problem

Nobody loses money on the tracking map.

Every logistics platform demo opens with a map and a moving truck icon. We have never once been called in to fix the map. We get called when ₹40 lakh of cash-on-delivery is somewhere between a delivery executive and a bank branch, and nobody can say where. We get called when a vehicle is stopped because an e-way bill ran out on the way and no one was watching the clock. We get called when the driver app needed a network to record a delivery, so three hundred deliveries were typed in from memory at 9pm. And we get called when a parcel missed a line-haul connection, and the system cannot show what just happened. It thinks a shipment only has a start and an end. That is the software we build.

Talk about your network
99.4%
ePOD capture rate

This is because capture works offline. The old system managed 71% — the rest was filled in from memory hours later.

T+1
COD settlement

Cash is matched to a bank line and paid to the merchant the next day, from a ledger and not a spreadsheet.

18%
Fewer failed deliveries

This comes from scoring how good each address is, plus a WhatsApp confirmation before the executive leaves the hub.

4.2hrs
Exception resolution

Middle time from the moment a missed connection is scanned to the moment a new ETA is set and the customer is told.

Where logistics platforms actually break

Every logistics platform we have been asked to rescue failed in one of four places. It was not the tracking map — the tracking map always works. It failed in the money, the network, the driver's phone, or the paperwork.

COD reconciliation is where logistics companies actually lose money

Cash on delivery is still a large share of Indian ecommerce parcels. Every one of those rupees passes through a delivery executive, a hub cashier, a bank deposit slip and a payout file before it reaches the merchant. At each handoff, the number can change.

The systems that fail here treat COD as a status flag on a shipment. The systems that work treat it as a ledger. Cash owed is created the moment the executive marks the parcel delivered. It stays against that executive until it is deposited. The deposit is matched to a bank statement line. The merchant is paid only from cash that has settled. Every entry is double-sided and cannot be changed — a correction is a new reversing entry, never an edit. That is what lets you answer, at 4pm on a Tuesday, exactly how much cash is in the field, who is holding it and for how long. Without that, losses stay hidden until the quarterly audit, and by then the trail is cold.

Hub-and-spoke is a graph problem, and most software pretends it is a list

A parcel does not go from A to B. It goes from a pickup to a first-mile hub, onto a line-haul trip to a regional sorting centre, to a destination hub, into a delivery route with sixty other parcels, and possibly back again on an RTO. Software that sees a shipment as having only "an origin and a destination" breaks the moment you ask a real question. Which bags are on this trip? What is the bag seal number? Which parcels missed the connection, and what is the new ETA for each?

We treat the bag, the trip, the leg and the scan as main building blocks. A scan is a record that cannot be changed. It holds a place, a time, a device and a person. The shipment's state is worked out from its scans, never stored and edited in place. This costs slightly more to build. It is what makes exception handling possible, and that is the whole job of a logistics control tower.

The driver app is offline-first or it is a liability

The delivery executive's phone will lose signal in a basement warehouse, a lift, a gated society and most of rural India. If the app needs a network to record a delivery, the executive will record it later from memory, or not at all. Then your data is fiction.

So we keep a local database on the phone as the main record. Every action is queued as an entry that cannot be changed. When the phone is back online, it syncs, using clear rules for conflicts. The electronic proof of delivery — signature, photo, OTP, geotag — is captured and stored on the phone. It is uploaded when there is signal, with the time of capture and not the time of upload. That difference has settled more than one payment dispute in our clients' favour.

The paperwork is not optional and it is not simple

An e-way bill is required for most consignments above ₹50,000. Its validity depends on distance. It must be extended before it runs out, and a vehicle stopped with an expired one is held and fined. Part-B must be updated when the vehicle changes. FASTag transactions have to be matched to trip cost, or the fuel and toll budget is a guess. We build these as connections that retry on failure, never repeat a step by mistake, and send an alert when a bill is close to expiry. They are not a screen where somebody in operations has to remember.

Modules

What we build into a logistics platform.

We build in this order, because each part is the base of the next. You are running real parcels through the spine long before the last module is done.

Shipment and scan spine

Bags, trips, legs and scans as main records that cannot be changed. Shipment state is worked out from scans, never written over.

Offline-first driver app

A local database on the phone, queued actions, and ePOD with signature, photo, OTP and geotag — all captured with no network.

COD ledger

Double-entry record of cash owed, from delivery to executive to hub deposit to bank line to merchant payout. It cannot be changed, and it can be matched.

Line-haul and trip planning

Vehicle and trip assignment, bag lists with seal numbers, how full each vehicle is, and connection planning across the network.

Route optimisation

Delivery order planned under real limits — time windows, vehicle size, executive shift length, and the map pin that is 200m off.

E-way bill and FASTag

Bills made through the NIC API, Part-B updates, validity tracked against distance, an alert before it runs out, and FASTag matched to trip cost.

Control tower

Lists of problems to act on, not just dashboards. Missed connections, old parcels, undelivered RTO and executives holding cash too long.

Customer communication

WhatsApp Business API for delivery times and confirmation before dispatch — the cheapest way to cut failed deliveries.

Merchant portal

Bookings, label printing, pickup requests, RTO handling, and a payout statement that matches to the rupee.

India, specifically

The compliance and integration reality.

None of this is optional. This is where a platform built abroad — or by someone who has never watched a vehicle get stopped — quietly falls apart.

  • E-way bill (NIC)

    Required above ₹50,000 for most consignments. Validity depends on distance, Part-B changes with the vehicle, and an expired bill on the road means the vehicle is held and fined.

  • FASTag / NETC

    Toll payments matched to the trip they belong to. Without it, your toll and fuel cost per trip is a guess that somebody defends in a meeting.

  • GST on freight

    Forward charge and reverse charge, place-of-supply rules across state lines, and invoices that match your GSTR filings instead of clashing with them.

  • Addresses that are not addresses

    Indian addresses are landmarks, not coordinates. Scoring each address, checking pincode coverage and rating how sure the map pin is work better than pretending the pin is right.

  • Aggregator and 3PL APIs

    Delhivery, Blue Dart, Shiprocket, Ecom Express and Xpressbees — each with its own contract, its own downtime and its own meaning for a status code.

  • Cash handling controls

    Cash limits for each executive, alarms for old deposits, and a maker-checker (one person enters, another approves) on every manual change. Cash loss is a controls problem before it is a people problem.

How we ship it

The spine first. Always the spine first.

  1. 01

    Walk the network

    Weeks 1–2

    We spend days in your hubs, on a delivery route, and beside the cashier counting cash at the end of a shift. Every logistics platform that fails was designed in a meeting room by people who had not done this.

  2. 02

    Build the scan spine

    Weeks 3–7

    Bags, trips, legs and scan records that cannot be changed. Nothing else can be built right until this is right. Adding it later means rewriting everything built on top.

  3. 03

    Put it in the executive's hand

    Weeks 8–13

    The offline-first driver app with ePOD, tested on the ₹10,000 Android phone your executives really carry, in a basement, with the network off.

  4. 04

    Close the money loop

    Weeks 14–19

    The COD ledger, hub deposits, bank statement matching and merchant payout. This is the part that changes the P&L, and it needs the spine to be trusted first.

  5. 05

    Then the control tower

    Weeks 20–26

    Lists of problems to act on, e-way bill alerts, reports on old parcels, and the numbers that tell an operations head where tomorrow's failure is building — not what yesterday's was.

Tell us where the cash goes missing.

If you can say, to the rupee, how much COD is in the field right now and who is holding it, you may not need us. If you cannot, that is the talk to have — and it takes twenty minutes.

The stack

Built on tools that will still be here in five years.

  • Laravel
  • React Native
  • PostgreSQL
  • Redis
  • Node.js
  • NIC e-Way Bill API
  • FASTag / NETC
  • Google Maps Routes API
  • Delhivery API
  • Shiprocket
  • Razorpay
  • WhatsApp Business API

FAQ

The questions you were going to ask on the call.

As a double-entry ledger, not a status flag on a shipment. Cash owed is created the moment a delivery executive marks a parcel delivered. It stays against that executive until it is deposited. The deposit is matched to a bank statement line. The merchant is paid only from cash that has settled. Entries cannot be changed — a correction is a reversing entry, never an edit. That is what lets you say, at any moment, how much cash is in the field, who is holding it and for how long. Treat COD as a flag and cash loss stays hidden until the yearly audit.

It has to, or your data becomes fiction. Executives lose signal in basements, lifts, gated societies and most of rural India. An app that needs a network to record a delivery just gets filled in later from memory. We build a local database on the phone as the main record, queue every action as an entry that cannot be changed, and sync when the phone is back online, using clear rules for conflicts. ePOD — signature, photo, OTP, geotag — is captured offline and stamped with the capture time, not the upload time. That difference has settled real payment disputes.

Yes, and we build them as proper connections, not a screen someone has to remember. That means e-way bills made through the NIC API, Part-B updates when the vehicle changes, validity tracked against distance, and an alert before a bill runs out. A vehicle stopped with an expired e-way bill is held and fined, and the delivery is late. FASTag payments are matched to trip cost, so toll and fuel budgets are measured and not guessed.

That is the core of it. We treat the bag, the trip, the leg and the scan as main building blocks. A scan is a record that cannot be changed, and it holds a place, time, device and person. Shipment state is worked out from scans and not stored and written over. This is what makes the real questions answerable. Which bags are on this trip? What is the seal number? Which parcels missed the connection? What is each one's new ETA? Software that sees a shipment as just a start and an end cannot answer any of them.

If you are a merchant shipping parcels, you probably should not — use the aggregators. You need a platform when you are the logistics company. You own hubs, you own vehicles, you employ the executives, and you hold the cash owed. The aggregator's dashboard cannot see how full your line-haul vehicles are, how much cash each executive loses, or how many problems each hub has. That is the line. We will tell you honestly which side of it you are on before we quote anything.

A first production version — shipment lifecycle, hub scanning, driver app with ePOD, COD ledger and a control tower — is typically 20 to 28 weeks and starts around ₹28,00,000. We build it in stages. First comes the scan-and-track spine, because everything else is built on it. Then the driver app, then COD, then the problem-list and control tower layer. You are running real parcels through it long before the last module is done.

Let's talk logistics platform development

Bring us the hard version of the problem.

A senior engineer reads every enquiry. You'll get a real answer — scope, risk and a number — within one business day.