Benchmarks
Warehouse Fulfillment KPIs: Benchmark Ranges and the Levers That Move Them
Pick accuracy, pick rate, labor cost per order, dock to stock, inventory accuracy, space use and perfect orders: how to measure each, what moves it, and which figures have a cited source.
Warehouse benchmarks are a direction, not a destination. Use the ranges below as reference points for spotting a gap worth investigating, and compare against your own trend before any industry figure.
Every figure shows its basis. A superscript number links to a public source in the Sources list at the end, and the sentence states that source's scope. A figure marked no cited source is kept exactly as originally published because none of the public sources checked for this guide states it, so treat it as illustrative.
At a glance
| KPI | Typical | Strong | Basis |
|---|---|---|---|
| Pick accuracy | 97 to 99% | 99.5% (competent); 99.9% or better (world class) | no cited source |
| Pick rate, discrete cart picking | 50 to 80 lines per hour | Not stated | no cited source |
| Pick rate from static bin shelving | 50 to 100 picks per person-hour | Not stated | Bartholdi and Hackman1 |
| Labor cost per order, single-line ecommerce | $1.50 to $3.00 (benchmark) | Can dip under $1.00 (automated) | no cited source |
| Labor cost per order, multi-line B2B | $3.00 to $7.00 (benchmark) | Not stated | no cited source |
| Dock to stock time | 24 to 48 hours | 4 to 8 hours (good); under 2 hours (cross-dock, prioritized receipts) | no cited source |
| Inventory record accuracy | 95 to 98% | 99% (good); 99.5% or higher by location (world class) | no cited source |
| Space utilization by pallet position | Not stated | 80 to 90% (world class) | no cited source |
| Perfect order rate | 90 to 96% | Exceeds 98% (world class) | no cited source |
| Units per labor hour, whole DC | 40 to 80 (common) | 100 plus | no cited source |
Method: How the figures in this guide are sourced
A figure with a superscript number is stated by the cited source, listed under Sources with its publisher and access date. The cited figures come from a warehouse science text by Georgia Tech researchers, and each sentence states its scope. A figure marked no cited source is kept exactly as originally published because none of the public sources checked for this guide states it. It is illustrative, not measured industry data.
Pick accuracy
Pick accuracy is the KPI customers feel first. Typical operations run 97 to 99 percent no cited source, competent ones hit 99.5 percent no cited source, and world-class scan-verified picking reaches 99.9 percent or better no cited source.
For scale, 99.9 percent accuracy is 1,000 defects per million lines, so anything better than 99.9 percent is under 1,000 defects per million lines.
Mispicks cost more than the error itself. Inaccurate orders generate returns, and Bartholdi and Hackman note that handling a return can cost up to ten times the cost of shipping the product out1.
The lever that moves accuracy most is the verification method. Each step up in verification raises the ceiling.
| Verification method | Accuracy | Basis |
|---|---|---|
| Paper picking | Caps around 99.5% | no cited source |
| Barcode scan-to-confirm | Reaches 99.9% | no cited source |
| Pick-to-light or voice with check digits | Past 99.95% | no cited source |
How to measure
- Divide correctly picked lines by total lines picked.
- Measure it from cycle-count audits, not returns, so internally caught errors count.
- Track it weekly by zone to find the bins that drag the average down.
What moves it
- The verification method: paper, barcode scan-to-confirm, or pick-to-light and voice with check digits.
- Fixing the specific bins and zones that pull the average down.
Track it with the Pick Accuracy calculator.
Pick rate
Order picking is the warehouse's most expensive activity. Bartholdi and Hackman, after Frazelle (1996), put it at about 55 percent of warehouse operating costs1.
Pick rate depends on the picking method and the storage medium, so benchmark against your own configuration. The table shows the rates for the main setups.
| Method or storage medium | Pick rate | Basis |
|---|---|---|
| Discrete cart picking | 50 to 80 lines per hour | no cited source |
| Static bin shelving | 50 to 100 picks per person-hour | Bartholdi and Hackman1 |
| Carton flow rack | 150 to 500 picks per person-hour | Bartholdi and Hackman1 |
| Batch and cluster picking | 100 to 150 lines per hour | no cited source |
| Goods-to-person systems | 200 to 400 picks per hour | no cited source |
Travel is the biggest lever, at 50 to 60 percent of a manual picker's shift no cited source. In the breakdown Bartholdi and Hackman reproduce from Frazelle (1996), traveling is 55 percent of order-picking time, ahead of searching at 15, extracting at 10, and paperwork and other activities at 20 percent1. Slotting fast movers into golden zones and batching orders attack that waste directly.
For example, a 22 person crew that picks 20 percent faster needs only about 18 people for the same work, since 22 / 1.2 is about 18. That frees several headcount.
How to measure
- Count lines or picks per paid picking hour, and use one unit consistently.
- Trend it by picker and by method, not as one blended number.
What moves it
- Travel time, cut by slotting fast movers into golden zones.
- Batching orders so one trip serves several orders.
- Matching the picking method and storage medium to each SKU's velocity.
Trend pick rate by picker and by method with your own counts and hours. Pick Rate
Labor cost per order
Labor cost per order is the summary financial KPI, and its range swings with order profile. Single-line, single-unit ecommerce orders benchmark at 1.50 to 3.00 dollars of labor no cited source, and multi-line B2B orders at 3.00 to 7.00 dollars no cited source.
Automated single-line fulfillment can dip under 1.00 dollar no cited source. Rather than chase a universal target, index against your own order mix and drive the trend down.
How to measure
- Divide warehouse labor cost by orders shipped in the same period.
- Report it monthly, split by order profile.
- Pair each movement with the pick rate change that caused it.
What moves it
- Pick rate.
- Batch density.
- Reducing indirect labor to below 20 percent of paid hours no cited source.
Work out labor cost per order from your own hours, wages and order volume. Warehouse Labor Cost per Order
Dock to stock time
Dock to stock time gates how fast inventory becomes sellable. Typical warehouses run 24 to 48 hours no cited source, good ones 4 to 8 hours no cited source, and cross-dock or flow-through operations achieve under 2 hours for prioritized receipts no cited source.
Every hour of dock to stock is inventory you own but cannot pick, so on fast movers it shows up as phantom stockouts. Receiving and put-away also carry real cost: Bartholdi and Hackman put put-away at about 15 percent and receiving at about 10 percent of warehouse operating costs1.
How to measure
- Time each receipt from arrival at the dock to available to pick in the warehouse management system (WMS).
- Set a tighter target for A items than for slow C stock.
What moves it
- Advance ship notice (ASN) quality.
- Receiving labor scheduled to inbound volume.
- Directed putaway that removes the search for a location.
To test whether directed putaway or another WMS capability pays back against your dock to stock gap, use the WMS ROI calculator.
Inventory record accuracy
Inventory record accuracy underpins everything downstream. Typical is 95 to 98 percent no cited source, good operations hold 99 percent no cited source, and world-class sustains 99.5 percent or higher at the location level, not just at the aggregate SKU level no cited source.
The distinction matters. Aggregate accuracy can look high while individual bins are wrong in ways that break picks.
How to measure
- Count by location, not only by SKU total.
- Score a location as accurate only when both item and quantity match the record.
- Size the count program so its coverage matches your accuracy target.
What moves it
- Cycle counting by ABC class.
- Root-causing every variance rather than just adjusting it.
- Transaction discipline at receiving and picking.
Size the count program and check results with the tools in the warehouse and fulfillment calculators.
Space utilization
Space utilization has a counterintuitive target. World-class is not 100 percent; it is 80 to 90 percent by pallet position no cited source.
Above 90 percent you lose the slack to receive, slot and reslot efficiently, and honeycombing, the waste of partly empty lanes and locations, forces double handling no cited source. Below 70 percent you pay for cube you do not use no cited source.
How you assign locations sets the ceiling. Bartholdi and Hackman explain that under dedicated, fixed-slot storage each location empties between replenishments, so on average storage capacity is only about 50 percent utilized1. Shared storage, which lets an emptied location take another product, is their remedy.
How to measure
- Measure by cube, not floor area.
- Net out aisles and clearance before you divide.
- Report pallet storage by pallet position.
What moves it
- Storage-medium fit: pallet, case-flow and shelf media matched to velocity.
- Dynamic or shared slotting instead of fixed slots.
Before you lease more building, check whether reslotting closes the gap with the space tools in the warehouse and fulfillment calculators.
Perfect order rate
Perfect order rate multiplies the on-time, complete, damage-free and accurate-documentation rates, so every component drags the total. Typical is 90 to 96 percent no cited source, and world-class exceeds 98 percent no cited source.
For example, four components at 99 percent each give 0.99 x 0.99 x 0.99 x 0.99 = 0.9606, or 96.1 percent perfect orders.
How to measure
- Track the on-time, complete, damage-free and documentation rates separately.
- Multiply the four rates measured over the same set of orders.
What moves it
- Attacking whichever component is weakest first.
- Pick accuracy, which feeds both the complete and the accurate components.
Let the WMS ROI calculator quantify whether an automation lever pays back against the specific KPI gap you are trying to close.
Units per labor hour
Units per labor hour across the whole distribution center (DC) captures total productivity. Figures of 40 to 80 are common no cited source, and 100 plus signals strong process or automation no cited source.
How to measure
- Divide units shipped by all labor hours in the building, direct and indirect.
- Keep the period and the hour definitions the same every time.
What moves it
- Pick rate and batch density.
- The share of indirect labor in paid hours.
- Automation, where the payback case holds up.
Test whether an automation or WMS project pays back against your productivity gap. WMS ROI
Where to start and how often to review
Sequence improvement in the order that protects revenue and compounds. Fix inventory and pick accuracy first, because bad data and mispicks corrupt every other metric and every downstream cost.
Then attack pick rate and dock to stock, which convert directly into throughput and availability. Space utilization comes last unless you are physically constrained.
Review cadence
- Pick accuracy and inventory accuracy: weekly.
- Labor cost per order and pick rate: monthly.
- Space utilization and dock to stock: quarterly.
How to build your own baseline
- Write down each KPI's definition and keep it fixed from period to period.
- Collect several months of your own data before you set a target.
- Compare against your own trend first and published ranges second.
- Target a specific percentage-point gain each quarter and verify it in the relevant calculator.
- Do not declare victory on a single good week.
Sources
- Warehouse & Distribution Science, Release 0.98.1. Georgia Institute of Technology, H. Milton Stewart School of Industrial and Systems Engineering (J. J. Bartholdi III and S. T. Hackman). published August 16, 2019. Section 2.4 (p. 15, dedicated storage utilization); Sections 3.1 to 3.3 (pp. 24-25, cost shares and the order-picking time table after Frazelle 1996); Section 3.4 (p. 29, returns); Sections 5.1.2 and 5.1.3 (pp. 42-43, pick rates by storage medium). Accessed October 1, 2026.
Published 2026-07-01.