Warehouse go-live before peak: which warehouse management features can land in ten weeks
Halden Industrial Supply ships fasteners and safety equipment from three warehouses: Portland (owned), Memphis (leased) and Monterrey (run by a third-party logistics provider). The company carries 11,400 active SKUs, employs 38 pickers, and ships about 2,100 order lines on a peak day. Inventory is tracked only by item, site and warehouse; there are no locations and no license plates. Pickers work from paper lists and key confirmations into a shared spreadsheet at the end of each shift. Receiving clerks type PO quantities into an invoice journal rather than posting a product receipt. Last year: 6% of order lines short-picked, shrink of $340,000, and a two-day manual push to close inventory at month end. The supply chain director wants to switch on every warehouse management feature at once and go live in ten weeks, just before peak. Warehouse supervisors say they cannot free staff for training until after peak, and the Monterrey 3PL has no Dynamics 365 users and will not accept logins. IT has budget for rugged scanners but not for custom code.
1. Which capabilities should realistically go live at Portland and Memphis before peak, and what must exist before paper pick lists can be retired? Explain the sequencing.
Read the worked answer
Separate the two layers first. Inventory management already gives on-hand by item, site and warehouse, reservations, inventory journals and counting journals; it is not what is failing. Warehouse management adds locations, work, work templates and mobile-device execution, and that is the layer that fixes short picks.
Before mobile picking replaces paper, these must be true for a warehouse:
- The warehouse record is enabled for warehouse management processes. This is a one-way decision in practice once transactions exist, so treat it as a design choice, not a switch.
- Location is active in the storage dimension group, and, if you choose plate tracking, license plate is active in the tracking dimension group. Plate tracking is the biggest behavioural change for clerks who currently key quantities; start location-only and add plates after peak.
- Location profiles and formats, default receipt and pick locations, location directives, work templates and work classes per work order type, mobile device menus and menu items, and work users mapped to warehouse workers.
- Barcode labels on locations (and items), scanners enrolled in the mobile app, and trained supervisors who own cycle counting.
Sequence: Portland receiving and putaway plus cycle counting first, outbound picking second, Memphis after peak. Measure short-pick rate, lines per hour and count accuracy before scaling.
2. How should Monterrey be modelled given that the 3PL has no Dynamics 365 users and refuses logins?
Read the worked answer
Do not enable warehouse management processes for Monterrey. Mobile-device execution assumes authenticated work users, and a 3PL that will not take logins cannot confirm work, so the site would sit with open work and unreliable on-hand.
Model Monterrey as a non-WMS warehouse that holds your inventory, fed by transfer orders from Portland or Memphis, with an in-transit warehouse so shipped-but-not-received stock is visible and does not inflate availability at either end. The 3PL sends periodic count and shipment files; you post receipts, adjustments and counts through inventory journals against Monterrey locations defined at warehouse level only. If ownership sits with the 3PL until shipment, model it instead as a vendor with a purchase order and product receipt at the point the goods leave the 3PL.
The judgement point: WMS is a per-warehouse setting, not a legal-entity-wide switch, so Portland and Memphis can run full work execution while Monterrey stays journal-based. Document that Monterrey's on-hand accuracy depends on file timing, and reconcile its in-transit balance weekly during peak.
3. The director insists on a single big-bang go-live for all three sites. What is your counter-argument, and what would you measure to prove or disprove success?
Read the worked answer
The counter-argument is financial, not technical. Warehouse management changes the physical routine of every picker and receiver: new labels, directed putaway, scanner confirmation of every line. A big-bang launch in the peak window converts any design error into unshipped orders, and unshipped orders in peak are lost revenue plus expedite cost, which dwarfs the licence and hardware spend. The stated constraints also make big-bang infeasible: no training capacity until after peak and no custom code budget means every gap must be closed with standard configuration, which needs pilot feedback. Microsoft-side change freezes around peak exist for the same reason.
Propose: Portland pilot four to six weeks before peak on outbound picking and cycle counting, Memphis after peak, Monterrey untouched. Hold a written go/no-go on pilot data.
Measure and report:
- Short-pick rate against the 6% baseline, by picker and by zone.
- Inventory record accuracy from cycle counts versus the two-day month-end close.
- Lines picked per labour hour, to catch a productivity dip that persists.
- Shrink value trend against the $340,000 baseline.
- Open work orders older than one shift, which is the earliest sign that work templates or directives are wrong.