Learning objectives
- Compute a simple warehouse cost per order and cost per line
- Apply an ABC slotting approach to reduce travel time
- Use order profile data to size receiving and shipping capacity
Anatomy of a warehouse
A warehouse performs four functions: receiving product from inbound carriers, putaway into storage, picking to fulfill orders, and shipping to outbound carriers. Each function uses labor, equipment, and space, and each has a different cost driver. Receiving is paced by truck arrivals and unload time per truck. Putaway is paced by inbound volume and the distance from receiving dock to slot. Picking is paced by order volume, lines per order, and pick path. Shipping is paced by carrier cut-off times and staging area. A useful diagnostic is to compute labor hours per function as a share of total warehouse labor hours and watch how the mix moves over time. In a conventional case-and-piece-pick operation, picking is normally the largest single block by a wide margin, with receiving, putaway, and shipping each taking a much smaller share. Published benchmark ranges circulate widely, but they vary enough by automation level, order profile, and how each firm books indirect labor that comparing your split to someone else's is a weak test. The stronger test is internal: measure your own split, then ask why it is what it is and whether it is drifting. If picking is a smaller share than you would expect, the likely explanations are over-staffing at the docks, a layout that forces excessive travel booked to another function, or an order profile with far more lines per order than the design assumed.
ABC slotting
Measure picking time and travel before deciding how much a slotting change could save. ABC slotting ranks SKUs by velocity (typically pick frequency or pick lines per period) and places A items in the most accessible, closest-to-pack locations, B items in the next tier, and C items in the most remote or least ergonomic locations. A conventional starting split is the top 20 percent of SKUs generating around 80 percent of pick lines (A), the next 30 percent generating most of the remainder (B), and the bottom half generating very little (C), but the actual curve should be measured from your own pick history rather than assumed, because assortment breadth changes it a lot. The size of the prize is likewise site-specific: it depends on how far travel currently is, how concentrated the velocity curve is, and how much of the pick face is already well placed. The way to size it before committing is to compute current average travel per line from the WMS, re-rank slots against the last quarter's pick frequency, and recompute travel under the proposed assignment; the difference between those two numbers is the estimate, and it is worth far more than any published range. Slotting is not a one-time project. Velocity rankings shift with promotions, seasonality, and assortment changes, so the ranking should be re-run on a regular cadence, at an interval chosen from the rate of assortment and demand changes, and the drift itself is the signal for how urgently a re-slot is needed.
Capacity sizing
Sizing a warehouse starts with peak day order volume, which is typically 1.4-1.8 times the average day, depending on the business. From peak orders, derive peak lines per order, then peak picks (lines x picks per line for multi-line orders). Divide peak picks by the picking rate (units per hour per picker, or lines per hour depending on the metric), and you get the number of picker-hours required at peak. Add indirect labor (receiving, restocking, shipping, supervision) at 25-40 percent of direct labor depending on automation. Divide total labor hours by paid hours per FTE to get headcount. Cross-check with floor space using a storage cube model and with receiving and shipping dock counts using truck arrival patterns. Under-sizing any one of these creates a bottleneck that shows up as missed ship dates.
Worked example
Problem
A warehouse handles 4,500 order lines per day on average. Each line averages 1.6 picks (multi-line orders with cases), so total picks per day = 4,500 x 1.6 = 7,200. Pick rate is 90 picks per hour per picker. Indirect labor is 30 percent of direct labor. Paid hours per FTE per day = 7.5 (one shift, with breaks). Compute (a) direct picker headcount at average day, (b) total headcount including indirect, and (c) headcount at peak day using a peak factor of 1.6.
Step by step
- Direct picker hours per day = 7,200 / 90 = 80 hours.
- Direct picker headcount at average = 80 / 7.5 = 10.67, round up to 11 FTEs.
- Indirect labor = 0.30 x 11 = 3.3 FTEs, round up to 4 FTEs.
- Total headcount average day = 11 + 4 = 15 FTEs.
- Peak day picks = 7,200 x 1.6 = 11,520. Direct hours = 11,520 / 90 = 128 hours. Direct headcount = 128 / 7.5 = 17.07, round up to 18.
- Indirect at peak (assume scales with direct) = 0.30 x 18 = 5.4, round to 6. Total at peak = 18 + 6 = 24 FTEs.
- If the firm cannot flex labor up, it must either run overtime, reduce pick rate per FTE, or push some volume to the next day, each of which has a known cost.
Answer. Average-day total headcount = 15 FTEs (11 direct, 4 indirect). Peak-day total headcount = 24 FTEs (18 direct, 6 indirect). The 9-FTE swing should be planned for via seasonal labor, cross-training, or demand smoothing.
Practice
Work each question before opening the solution.
-
A warehouse has not been re-slotted in 14 months. The operations manager claims velocity rankings have shifted due to two new product launches. What is the cheapest diagnostic to justify a re-slot?
Show solution for question 1
Pull the top 100 SKUs by pick frequency today and compare with the top 100 from when the slotting was last set. The churn rate in that list is the diagnostic: if a quarter of the list has turned over, a large share of today's fastest movers are sitting in slots that were assigned to yesterday's. To convert that into a business case rather than an assertion, take the current average travel distance per pick line from the WMS, re-rank the slots against the last quarter of pick data, and recompute travel under the new assignment. The difference in travel distance, multiplied by pick volume and the picker labour rate, is the annual saving, and it can be compared directly with the labour hours the re-slot itself will consume. Both numbers come out of data the site already has, which is what makes this the cheap diagnostic.
-
Why is the receiving function sometimes the bottleneck even though it is only 10-15 percent of labor?
Show solution for question 2
Receiving is paced by truck arrivals, which can be lumpy and concentrated around specific times of day. If the dock door count or unload crew is sized for average arrivals, a few peak-hour arrivals back up the dock and push product into the next shift's pick queue. The fix is usually a small one: add a dock door, stagger receiving windows, or pre-stage unload crews.
-
A warehouse runs two shifts and is considering a third. What cost categories change and what stays the same?
Show solution for question 3
Total direct labour hours rise roughly in proportion to the added volume, though hours per FTE do not change, since a third shift means more people rather than longer days, and the wage rate itself usually rises because night shifts carry a premium. Supervision scales less than proportionally: a third shift needs a supervisor but not a full duplicate of the day-shift management structure. Equipment is where the third shift earns its money, because forklifts, conveyor, and the WMS are already paid for and can be handed over at shift change rather than duplicated. Facility rent and base-load utilities are fixed and get spread over more throughput; activity-driven utilities grow. The decision should compare the marginal cost per pick on the third shift, including the night premium, against the service or capacity benefit of the extra throughput, and should also weigh the fact that a third shift removes the overnight window in which maintenance and cycle counting normally happen.