The Road In: Every Shipment from Supplier to Mine

Everyone watches the concentrate going out. Almost nobody watches what comes in — and a mine runs on what comes in. Spare parts, reagents, equipment, consumables: an unbroken line of shipments from suppliers in South Africa and overseas, moving by sea, by air, and finally by road across a multi-week overland corridor to the mine gate. When that line goes quiet, the questions start: has it shipped? has it cleared? where is the truck? And when the answers are wrong, the cost is not demurrage — it is a plant waiting on a part.

Having built an export control tower for our client — a publicly listed tin producer operating in the eastern DRC — we turned the same method around and pointed it the other way.

01Challenge

Inbound is harder to see than outbound, for a simple reason: more hands, more modes. A single consignment might be loaded at a supplier's warehouse, trucked to a port or airport, shipped or flown to East Africa, cleared, transferred at a yard, and then driven for weeks to the mine. Each leg has its own providers, its own paperwork, its own reporting habits — and none of them reports into anything the operator can query.

The symptoms mirror the export side, with one difference in kind: on outbound, a lost day costs money; on inbound, a lost shipment can cost production. The planning questions — when does it land? what is at the yard? which truck is carrying the compressor? — were being answered by email threads and memory.

02Diagnosis

The export platform had already proved the architecture: multiple reporting channels, one reconciled model, automation for the routine, people for the exceptions. The inbound problem added two requirements of its own.

One model for every mode. Sea, air and road legs cannot live in three trackers. A shipment is one thing moving through many hands; the model has to follow the shipment, not the mode — one chain of milestones from the supplier's door to the mine gate, whatever the leg.

The field must be able to write. The last leg — the long one — runs through country where connectivity is intermittent for hours at a time. If the people at the yard and on the road cannot record what they see, the picture dies exactly where it matters most.

03Approach

One chain of milestones per shipment. Every shipment carries a full milestone chain from origin to mine gate — up to 27 tracked checkpoints on the overland leg alone — on one live map. Vessel positions arrive automatically while the cargo is at sea; the road leg is updated from the field. Each shipment shows a plain-English status line — where it is, what happened last, what comes next — readable by anyone from the ops room to the boardroom.

Field and yard apps, offline first. The people at the yard and the border record what actually happens — arrivals, offloadings, loadings, seals, transfers — on their phones, with categorised photos attached to the shipment. No network needed: everything is stored on the device and syncs when the signal comes back. A container photographed at the yard becomes part of that shipment's permanent record, visible to the office minutes later — or as soon as the connection allows.

Reconciliation with guards. Supplier reports, yard records, transporter updates and automatic vessel feeds all land in the same model — reconciled hourly, with rules deciding who wins when sources disagree, and checks that refuse impossible sequences (a shipment cannot arrive before it departs, whatever a spreadsheet says). Gaps get filled; existing records are never silently overwritten.

The paperwork rides along. Waybills, customs papers and photos are filed against the shipment they belong to, and a comment thread per shipment holds the texture that fits no schema — shared between the office and the field, so there is one conversation instead of five.

04Outcome

The system is young next to its seven-year outbound sibling, and already carries the operation's whole resupply picture: more than 200 shipments and 5,000 tracking events across sea, air and road; more than 550 transport legs tracked across the three modes; a corridor GPS-mapped across more than 65 waypoints; up to 27 milestones tracked on every overland shipment.

The practical change is the same one the export side taught us to expect: "when does it land?" now has one answer, the same answer, for everyone who asks — and the ops team spends its time on the shipments that need attention instead of reconstructing the whereabouts of the ones that don't. The morning picture includes what moved yesterday, what is at the yard, and what is late — without anyone having to ask.

05Lessons

1. Follow the shipment, not the mode. Three trackers for three modes means three versions of the truth. One milestone chain per shipment, whatever the leg, is what makes the answer singular.
2. Offline is not an edge case. In this geography it is the main case. A field app that needs a signal is a field app that misses exactly the events you most need recorded.
3. A proven method transfers. The second platform did not require rediscovering the architecture — the channels, the reconciliation, the guards were already proved on the export side. The direction changed; the discipline did not.

Read the export side: A Control Tower for a Tin Corridor →