Case study · ITC wholesale distributor network

Built inside a real ITC distributor, one working day at a time

RouteSync did not start as a product. It started as a specification written against the way one ITC wholesale distributor in Hyderabad actually runs its last mile — and every feature in it survived a real operator, real drivers and real order files before it was called finished.

Pilot at a glance
Principal
ITC
Operation
Wholesale distributor, Hyderabad
Retailers
1,194
Beats
76
Fleet
Vans + drivers, ~30 stops each
Order source
Trishul CSV export
Rollout plan
3 distributors → 50
The situation

A paper-and-phone last mile

The distributor receives orders through ITC’s Trishul system and delivers them to roughly twelve hundred kirana shops across the city. Vans go out each morning; each driver covers around thirty retailers on a run, sometimes making a second trip after lunch.

The dispatch-and-monitor part of that operation ran entirely on paper and phone calls. A salesman printed the day’s order list, decided which orders were physically ready, grouped them by beat, and wrote out the stop order from memory. Once a van left the warehouse there was no way to know where it was or which shops had been served, other than calling the driver. Deliveries were confirmed verbally. Nothing about the day was measurable afterwards.

The beat system — retailers grouped into fixed weekly routes — was doing a job it was never designed for. It existed so a human could remember roughly where shops were. It was being used as the planning constraint, which meant a van’s load had nothing to do with the day’s actual orders. Quiet days went out at a third full. Peak days were overloaded and nobody could tell until the van was on the road.

What the operation was losing
  • Time — a skilled person’s morning, every morning, on work a computer does better.
  • Capacity — vans running well under their rated load on ordinary days.
  • Responsiveness — a new retailer waited for the next beat cycle before it could be served.
  • Evidence — no proof of delivery, so disputes came down to whose word it was.
  • Insight — no fill rate, no utilization, no reason codes. Nothing to improve against.
  • Continuity — the plan lived in one person’s head.
The build

Four applications on one dataset

The scope was deliberately narrow: replace the dispatch-and-monitor portion of the flow, leave the ERP alone as the source of commercial truth, and give each person exactly the screen they need.

Brand-side portal

Onboard distributors, manage the shared product catalogue, and see every distributor’s day in the same shape.

Distributor portal

Import, plan, dispatch, track, report. The operator’s whole day lives here.

Driver app

Installable on any Android phone, in English, Hindi or Telugu. Stops, navigation, OTP confirmation, GPS capture.

Field salesman app

Fix a wrong retailer pin, mobile number or address from the field, with a photo and a GPS accuracy reading.

The turning point

Beats stopped being the plan

The first version gave the operator better information — road distances, retailer coordinates, order weights — and let him sequence the trip himself. It worked, and it was faster than paper. But it still asked a person to solve a clustering problem every morning.

Before · fixed weekly beats

The route decided the orders

  • Same driver, same shops, same day each week.
  • Load followed tradition, not today’s order book.
  • New retailers waited weeks for a slot.
  • One delay cascaded through the whole beat.
After · daily auto-routing

The orders decide the route

  • Today’s ready pool clustered by geography, across the whole fleet at once.
  • Each van filled to a target the distributor sets, and never past it.
  • A new retailer ships the same day it orders.
  • Beat survives as a reporting tag — useful, no longer binding.
The order file already knew what each shop had bought and what it weighed. The retailer master already knew where every shop was. The vehicle master already knew what each van could carry. Nobody had ever put those three facts in the same room and asked for the answer.
What the pilot taught us

The decisions that only a live operation produces

These are the things that were not in any specification. They came from watching the system meet real drivers, real shopkeepers and real Tuesday afternoons — and they are the reason the product holds up.

Never change data behind the operator’s back

No nightly rollover, no auto-cancelling of stale trips. The dashboard detects a stuck state and offers a button; the record only changes when a person clicks it. Silent corrections destroy an operator’s trust faster than bugs do.

Translate the screen, never the data

The driver app speaks English, Hindi and Telugu — but every stored status, role and code stays in English. Reports, queries and audits are identical no matter what language the driver picked.

Bad GPS was mostly bad master data

We investigated complaints that driver GPS was “kilometres off.” Around nine in ten fixes were accurate to within fifty metres. The largest apparent errors were good GPS readings sitting far from a wrong retailer pin — a master-data problem wearing a GPS costume. That finding produced the field pin-correction queue.

No driver surveillance features

It would have been easy to build monitoring dashboards. We deliberately didn’t. The effort went into making the app accurate and fast for the drivers who use it properly — which is a better lever than watching the ones who don’t.

Humans make mistakes; give them an undo

A driver marks a stop delivered by accident. An operator closes a trip with orders still attached. Both got an explicit correction path in the UI, because the alternative is a support request and a hand-written SQL statement.

Shops don’t move — so cache forever

Road distance between two fixed shops never changes. Caching it permanently, and invalidating only when a pin is edited, turns mapping from a per-trip cost into a one-time one. It is the difference between a product that scales to fifty distributors and one that doesn’t.

Where it stands

What the distributor has now

The pilot distributor runs its full delivery day on RouteSync — import to dispatch to proof of delivery to report — with the same fleet and the same people.

Planning is a click, not a morningA full day of trips is proposed in seconds and reviewed in minutes.
Every delivery has evidenceOTP confirmation, GPS coordinates and a timestamp on every stop, plus a structured reason on every failure.
Retailers stopped calling to askTwo WhatsApp messages per delivery replaced the “where is my order” phone traffic.
The day is measurableFill rate, delivery latency, exception reasons and three utilization measures — daily, per trip, per order.
The knowledge left one person’s headAny trained operator can run the day. The system holds the map.
routesync.app/wd/reports
Reports · last 30 days Download .xlsx
Delivery latency
74.7%
closed on D+1
Order fill rate
92.4%
stops closed / total
Exception rate
13.8%
undelivered / total
The three utilizations
Weight utilization72%
Time utilization88%
Delivery utilization57%
Workbook · 6 sheets
Reports · daily rollup Summary · per trip Details · per hop Invoice level · per order Delivery aging Vehicle raw
The end-of-day view the distributor now has

Screen values shown throughout this site are representative of the pilot instance. In a walkthrough we’ll show you live figures and explain exactly how each one is computed.

Rollout

From three distributors to fifty — same software

There is no pilot version and production version. Scaling is an onboarding sequence, not a rebuild, and every distributor added runs the identical codebase on the identical schema.

Create the distributor

Code, name, city and warehouse coordinates — the origin point for every distance the system computes.

Assign the operator

One primary user with the distributor role. Login is a mobile number; password resets are admin-side.

Add the vehicles

Registration, rated weight and volume, and whether each one participates in auto-routing.

Add the drivers

Mobile number is the login. Import the retailer master, and the distributor is ready to plan its first day.

Run your distributor’s day through it

Send one day’s order file and your vehicle list. We’ll plan it on the live system and walk you through what came out — including whatever didn’t fit, and why.

Quick support