How Northbay Logistics cut purchase order cycle time from 9 days to 1

The situation

Northbay ran six sites, each with its own way of buying. Site managers ordered maintenance parts, packaging, and contract labour directly. Approvals happened in email when they happened at all, and finance discovered most commitments when the invoice arrived.

The consequence was not fraud. It was fragmentation: eleven suppliers for the same category of packaging materials, three different rates for the same class of contract driver, and a month-end close that took nine working days.

What changed

1. 1Every site moved to a single requisition form with cost centre and category attached.
2. 2Approval thresholds were set from the actual spend distribution, with purchases under $500 auto-approved in budget.
3. 3The top five categories were consolidated to one preferred supplier each, published inside the request form.
4. 4Receipting moved to the requester's phone, which cleared the invoice exception backlog within two months.

The result

Median cycle time dropped from nine days to one within the first quarter. More importantly, site managers stopped routing around the process, because the compliant path was now the fast one. Maverick spend fell from 31% of addressable spend to 8% over three quarters.

Month-end close now runs in four days, and budget owners can answer how much of their quarter is committed without asking finance.

What did not go to plan

Two things. Supplier consolidation took two quarters rather than the one budgeted, because eleven packaging suppliers turned out to be seven legal entities and merging the records meant phoning each one. And the maverick rate at the two smallest sites got worse in month two before it improved — both had a single approver with no delegate, and the fix was a staffing decision rather than a configuration change.

The $740k figure is realised savings against prior-year pricing on consolidated categories, verified by Northbay's controller. It excludes cost avoidance and excludes the close-time improvement, both of which they consider real but did not want in the number.