The Fourth DC That Doesn't Pay for Itself on Paper
Northline Provisions makes refrigerated sauces in two plants (Fresno, Allentown) and serves 4,200 retail and foodservice accounts from three leased DCs (Atlanta, Columbus, Kansas City). Average order size has fallen 34% in two years while order count rose 41%; on-time-in-full has slipped to 91.4% against a 97% contract standard. The VP of sales wants a fourth DC in Dallas. The network study projects $2.1M in annual freight savings, $1.4M in added fixed occupancy, and $0.9M in added inventory carrying cost at the current 6.2 turns. Two of the top five accounts are in Texas and have already claimed service credits. Capital is tight and the CFO wants a two-year payback.
1. Should Northline approve the Dallas DC? Build the economic case and state what must be true for the lease to be justified.
Read the worked answer
On the study's own numbers the project fails: $2.1M − $1.4M − $0.9M = −$0.2M per year, so payback never arrives.
But the study is built on averages, which is the wrong basis for a marginal node. Rebuild it:
- Freight savings are lane-specific. Only lanes that actually change mode or zone count. A new node also destroys some existing multi-stop consolidation and can raise cost on the lanes that remain.
- Incremental inventory is not average inventory. A fourth stocking point adds safety stock concentrated in the slowest movers; at 6.2 turns the $0.9M figure is probably optimistic unless the company repools slow SKUs at a single node.
- Occupancy must be a cash cost, not an allocation.
- Add revenue at risk from the Texas accounts and any other accounts that fall inside a new one-day zone, discounted by the probability of losing them.
The decision rule: approve only if the revised net is positive and it depends on retaining identifiable revenue. Even then, test lower-commitment substitutes first - a 3PL cross-dock in Dallas holding the top ~150 SKUs, zone skipping, or a bought next-day service out of Kansas City. If a full DC is chosen, take a 3-year lease with expansion and exit rights, not five.
2. Sales argues that a fourth DC 'fixes service.' How do you respond?
Read the worked answer
Challenge the premise before funding it. Decompose the 91.4% OTIF into three failure buckets - no stock at the shipping node, missed order cut-off, and carrier delivery failure - and cut it by account and by order size.
A 34% drop in order size with a 41% rise in order count points to a cost-to-serve and cut-off problem, not purely a distance problem. If most misses are availability misses, adding a stocking node makes service worse, because the same inventory is spread over more locations and slow movers go out of stock in more places.
The cheaper response is segmentation:
- A accounts: next-day commitment, dedicated carrier or inventory forward-positioned at a 3PL against a named SKU list.
- B accounts: two-day standard.
- C accounts: minimum order value or three-day window.
Publish the policy so sales can sell against it. Run a root-cause review on every miss for 90 days and report OTIF by segment monthly. Revisit the DC only if a genuine distance-driven gap survives those actions.
3. What assumptions in the network model would you stress-test before signing anything?
Read the worked answer
Test the assumptions that change the sign of the answer, not the ones that shave pennies:
- Inventory scaling. The model likely applies a square-root or average-inventory rule. Slow movers violate it. Model the new node SKU by SKU and confirm the $0.9M.
- Freight rate volatility and carrier capacity in Dallas, especially for refrigerated LTL. A 10% rate swing erases the entire $0.2M gap.
- Volume transfer. Moving volume out of Columbus and Kansas City may leave those sites unable to cover their fixed costs - the savings may be a transfer, not a reduction.
- Demand downside. What happens to payback if volume drops 15%? A lease is fixed; freight savings are variable.
- Service credit exposure, i.e., whether the credits stop or merely shrink.
- Lease exit cost and whether the site can be sublet.
Run a sensitivity that answers one question: what freight savings level makes NPV zero? If that number is within normal forecast error, the project is a coin flip and should be staged through a 3PL trial, not a capital commitment.