The situation
A national distributor moves roughly 40,000 deliveries a month. Orders live in the ERP, despatches in the warehouse system, proof-of-delivery with two haulier portals, and invoices in finance. None of them quite agree — short deliveries, substitutions, returns, disputed PODs — so an operations team spends, say, two days a week per person exporting all four into a spreadsheet that has quietly become business-critical infrastructure. Month-end is worse. Disputes are settled by whoever gives up first, and credit notes leak margin nobody can quantify.
The old options
No off-the-shelf product matches four specific systems and one specific process — this is exactly the work SaaS can’t do. A traditional systems-integration project would have done it, at a cost and timescale that kept it perpetually on next year’s list. So the spreadsheet survived another decade.
What we’d build
A reconciliation system that does the hunting so the team does the deciding:
- Automated feeds from the ERP, warehouse system and both haulier portals — matched continuously, not at month-end
- A live exceptions queue: every delivery that doesn’t reconcile, classified (short, substituted, missing POD, price mismatch) and routed to the right person
- Dispute workflow with the evidence attached — order, despatch, POD, invoice — so a claim takes minutes to assemble instead of an afternoon
- Credit-note control: nothing issued without a reconciled reason code, and a monthly margin-leakage report by cause, customer and route
- Finance-grade audit trail, because this data feeds the accounts
The build
Discovery call, scope in days, and a prototype in days running against a month of real export data — the exceptions queue lit up with the firm’s actual discrepancies, which is usually the moment the room goes quiet. Production integration and parallel running: weeks. This is the unglamorous systems work EBI has done since 2001 — the AI-assisted part is why it now arrives in weeks at a fraction of the old cost.
The economics
Typically 60–80% less than a traditional integration build. The return shows up in three places: the ops days handed back every week, the disputed deliveries recovered because the evidence is one click deep, and the credit-note leakage that stops once every credit has a reconciled cause. The firm owns the system; when a haulier changes portal, it’s a connector change, not a vendor negotiation.
Month three
Month-end reconciliation is a review, not a hunt. The spreadsheet is retired with a small ceremony. And for the first time, the margin-leakage number is on a dashboard — which is usually where the next project comes from.