Northline Foods: Third DC or Flow-Through?
Northline Foods makes refrigerated sauces in three plants and distributes through two DCs - Memphis (62% of throughput) and Reno (38%). Over three years, 40% of case volume has migrated from West Coast grocery chains to Southeast and Mid-Atlantic retailers, mostly club and mass formats that order full pallets and expect two-day replenishment. Outbound freight is now $18.4M a year, and the average haul from Memphis to the new demand base is 640 miles. OTIF has slipped to 91.2% against a 96% target, and the largest single cause is short shipments when Memphis runs out of the highest-velocity SKUs. A network study proposes an Atlanta DC: $4.2M capital, $1.9M annual fixed cost, a projected $4.5M freight reduction, and $0.4M of added inventory carrying cost. Reno would keep serving the West at current volume. Separately, two national retailers have told Northline they will not expand SKU counts until they receive lot-level traceability data linked to the shipping location.
1. Should Northline approve the Atlanta DC? Build the financial and service case, and state the conditions under which you would reverse the recommendation.
Read the worked answer
Base case: $4.5M freight − $1.9M fixed − $0.4M incremental carrying = $2.2M net per year, a payback of 4.2 / 2.2 = 1.9 years. That clears most hurdle rates, but two assumptions carry the case.
Freight savings are variable; the fixed cost is not. If Southeast volume lands 20% below plan, savings fall to $3.6M and net benefit to $1.3M (3.2-year payback) - still fundable. At a 35% shortfall, net benefit is $0.6M and payback exceeds six years, which is not. The decision is therefore really a question about demand concentration: how firm are the club and mass commitments?
The $0.4M inventory figure is likely understated. Splitting the same demand across three nodes generally raises total safety stock unless the nodes serve independent demand; Atlanta's demand overlaps Memphis's, so the square-root pooling benefit does not fully apply. A third node also adds a second echelon of handling and one more variance source for short-shipment risk.
Reverse the decision if: retailers will not commit volume in writing; a Memphis cross-dock or a 3PL flow-through in Atlanta can deliver the same transit time at lower capital; or carrier lane renegotiation closes the service gap without a facility.
Recommendation: approve a staged entry - shared 3PL space in Atlanta now, owned DC only after four quarters of proven volume.
2. What must the network design include to meet the retailers' lot-level traceability requirement, and where should the data be captured?
Read the worked answer
Traceability is a process-and-data design problem surfaced by a commercial request, not an IT project. The design must define the critical tracking event chain and the key data elements at each handoff:
- Production: lot/batch ID assigned at batching, tied to raw material lots, line, date/time, and quantity. This is the anchor; everything downstream inherits it.
- Receiving/putaway at each DC: lot ID scanned against the advance ship notice; location identified by a standardized location code (GLN) so a recall can be scoped to a node, not just a region.
- Pick and ship: lot ID linked to order, ship-to location, and carrier; case-level scanning is normally sufficient unless the retailer contract specifies unit-level.
- Upstream and downstream: co-packers, ingredient suppliers, and carriers must be contractually obligated to supply and retain the same elements; the Kestrel-style weak link is almost always a third party outside the four walls.
Design implications: master data governance (one item, lot, and location standard across plants and DCs); retention rules; a mock recall test with a stated target, typically evidence assembled within four hours; and a decision on whether Atlanta captures data via the WMS or a bolt-on serialization layer. Cost should be tested against the revenue at stake - SKU expansion from two national retailers usually justifies the investment, but case-level scanning at the DC may be enough. Do not approve the DC without the data model, because retrofitting location-level traceability into a new facility after go-live is materially more expensive.