How an Indonesian courier network fixed first mile, mid mile and last mile on one system

About the company

An Indonesian parcel and courier network collecting from tens of thousands of shippers, from marketplace sellers and small businesses through to enterprise accounts, moving volume through regional sortation hubs and delivering nationwide. It runs a core owned fleet alongside a large network of third-party trucking partners and contracted riders, which is how almost all Indonesian capacity is actually assembled.

Background

For a courier network competing on price, margin is utilisation. Not rates, not volume, but how full the truck leaves, how many drops a rider completes before dark, and how many times a parcel has to be handled before it reaches the door. Indonesian logistics is expensive by regional standards, and a network that wastes capacity has nowhere to hide it.

This company was running three businesses that happened to share a brand. First mile pickup was planned by branch coordinators using phone calls and group chats. Line haul was booked by operations staff ringing partner operators against whatever capacity they had that day. Last mile was assigned by hub supervisors at the start of each shift and then left alone until the shift ended.

Each of the three kept its own records, and none of them agreed. Nobody could follow a parcel end to end without calling three people. Nobody could tell whether a truck had left half empty, because the load was recorded after departure if at all. And nobody could say which partner operators were reliable, because performance existed as opinion rather than as data.

The company implemented Fynd TMS across all three miles.

Challenges, solutions and impact

Challenge #1: First mile planned by hand, branch by branch

Pickup demand arrives unpredictably and in scattered geography. Thousands of sellers, most of them small, each with a handful of parcels and a preferred window. Coordinators built pickup runs from memory and messages, so vehicles covered overlapping ground, some ran nearly empty, and pickups were missed often enough that sellers began hedging by booking with a second courier.

Solution. Fynd TMS plans pickup runs against actual booked demand, clustering shippers geographically and sequencing stops so a vehicle covers a coherent territory rather than a coordinator's mental map. Runs are re-planned as bookings arrive during the day instead of being fixed the night before. Drivers work from an app that sequences their stops and captures collection at the parcel level, so a parcel enters the network with a record attached from the first scan rather than at the hub.

Impact. Pickup capacity went further, with fewer vehicles covering the same territory and materially fewer missed collections. Because parcels are captured at pickup rather than at the hub, the network gained visibility of volume hours earlier, which is what made planning the next leg possible at all.

Challenge #2: Line haul booked on the phone, trucks running half full

Mid mile is where the cost sits. Moving volume between hubs is the most expensive leg in the network and the one with the most waste, because a truck that leaves at 60% full costs the same as one that leaves full. Bookings were made by phone against partner availability, which meant capacity was chosen by who answered rather than by what the lane needed. Consolidation happened only when someone noticed two part-loads heading the same way. Empty return legs were accepted as normal.

Solution. Fynd TMS plans line haul against forecast and confirmed volume per lane, so vehicles are matched to load rather than booked speculatively. Part-loads for the same corridor are consolidated automatically, and backhaul opportunities on return legs surface as capacity to be sold rather than distance to be absorbed. Departures are held against fill thresholds where the service level allows it, and released early where it does not, which turns a judgement call into a rule. Every movement is tracked between hubs rather than disappearing until arrival.

Impact. Load utilisation on line haul rose 34%, which is the single largest cost movement in the programme. Empty running on return legs fell as backhaul became something the system looked for by default. And because movements are now tracked, arrival times at downstream hubs became predictable enough to plan last mile against.

Challenge #3: Last mile fixed at the start of the shift, on addresses that did not resolve

Riders received a list at the start of the shift and kept it whatever happened next. Volume arriving mid-morning waited for the following day. A rider finishing early could not be given work from a neighbouring beat.

The deeper problem was addresses. Much of Indonesia is navigated by kampung and gang rather than street number, geocodes commonly land at the mouth of an alley, and a rider loses minutes to a phone call at the end of every unfamiliar trip. Failed attempts were the expensive consequence, because a reattempt costs a second delivery for one revenue event, and repeated failures on the same address were nobody's problem to fix.

Solution. Assignment became continuous rather than a morning event. Fynd TMS re-evaluates pending deliveries against available riders through the day, so parcels arriving after the shift starts get onto a route the same day and spare capacity is redirected to neighbouring beats. Routing uses times learned from the network's own rider GPS history rather than car-derived traffic feeds, because motorcycles filter through congestion and use alleys that road graphs do not contain.

Challenge #4: A partner network managed without a performance record

Most of the company's capacity belongs to other people. Partner trucking operators and contracted riders carry a large share of volume, and they were managed on relationships, phone calls and separately negotiated rates. There was no record of which operators ran on time on which lanes, no comparison of rates against performance, and settlement was reconciled from paperwork weeks after the fact. Poor performers kept getting work because nobody could prove otherwise.

Solution. Partners now operate inside the same system as the owned fleet. Loads are allocated across own and partner capacity by lane against agreed rate cards, with tendering and acceptance recorded. Every movement generates performance data, so on-time rates, damage and acceptance behaviour accumulate per operator and per lane. Settlement runs off the same movement records rather than off reconciled paperwork.

Impact. Allocation moved from relationships to evidence, which improved reliability on the lanes that mattered without renegotiating a single rate. Settlement disputes and the reconciliation effort behind them fell sharply, and the company could finally see the true cost of a lane rather than the cost of a truck.

Way forward

Next on the roadmap: chaining inter-island legs so a shipment crossing by ferry or sea is planned as one journey rather than handed between operations, forecasting lane volume further ahead to buy partner capacity on better terms, and giving enterprise shippers direct visibility of their own consignments across all three miles.

If your challenges resonate with those in this story, it's time to schedule a call with our experts. Whether you're planning first mile pickup against real demand, filling trucks on line haul, recovering capacity lost to failed deliveries, or allocating across a partner fleet on evidence rather than relationships, our solutions are designed to streamline your operations and drive growth.

With over a decade of experience in commerce and supply chain and a proven track record with 2,300+ brands, Fynd is here to turn your challenges into success stories. Let's drive your growth together.

Book a demo