Services · Storage Strategy
How the engine works.
From three WMS extracts to a fitted storage model in six stages. Every figure on this page is engine output for a fictional sample warehouse.
Three extracts. No IT project.
The engine reads flat files your WMS already produces. Two are required, one of a pair is required, and one is optional and adds the time-based views.
-
Item master
Required
SKU dimensions, weight, unit of measure, and case pack.Sample run: 6,500 rows
-
Inventory
Required
Every location with its dimensions and type, and the SKU and quantity it holds.Sample run: 11,129 rows
-
Volume or order lines
One of two
Units sold per SKU, either summarized or as dated order lines.Sample run: 5,777 SKUs summarized
-
Volume by day
Optional
Daily units per SKU. Adds seasonality, peak-day, and day-of-week views.Sample run: 859,143 rows
Every row accounted for.
Before anything is modeled, each table gets a quality pass. Rows that cannot be used are removed and listed with the reason. Rows that look wrong but are still usable are kept and flagged, so you can correct the source instead of finding the problem in the results.
Removed, with the reason
- Duplicate location ID
- Invalid location or SKU dimension
- SKU missing from the item master
- Duplicate SKU with mismatching dimensions
- Invalid inventory quantity or sales days
Kept, and flagged
- SKU name likely mangled into scientific notation
- Mixed unit-of-measure values for one SKU
- A dimension over 48 in, or every dimension at 1 in
- Weight over 50 lb
- Zero quantity or zero sales
Each table gets a summary of rows, SKUs, removed rows, and flagged rows that you review before the model runs.
Know the demand before you size the space.
The engine profiles demand and the catalog: how volume moves through the year and the week, how concentrated it is, and how inventory sits against velocity. SKUs are grouped into cube bins, AA through D, with summary metrics for each.
SKUs per cube bin · sample run
The peak day ran 2.3 times the average.
The busiest day, Cyber Monday, moved 60,659 units. Peak-to-average sets how much headroom pickfaces need, so the model sizes to the demand you actually see, not the mean.
35% of SKUs carry 80% of unit volume.
Concentration decides where pickface investment goes. The A band is where location size and position pay back.
Inventory against velocity.
Each point is a SKU. Points high and to the left hold far more stock than they sell.
Monday carries the most volume.
Share of weekly volume by day sets the rhythm for labor and replenishment.
Your rules, not a generic sort.
A handful of parameters drive the fit, and we set them with you. The starting days-on-hand values come from your own data: the average your current pickfaces already hold.
| Parameter | What it controls | Sample run |
|---|---|---|
| Non-pallet days on hand | Minimum days of demand a bin or carton-flow pickface must hold | 13.0 days |
| Pallet days on hand | The same rule for pallet pickfaces | 23.9 days |
| Maximum utilization | How full a pickface or reserve location may be filled | 100% / 100% |
| Minimum units | A floor on units per location, so slow movers do not get tiny slots | 12 units |
| Ceiling height | Clear height for stacking bays and computing footprint | 30 ft |
| Sales and inventory growth | Multipliers to test future demand and stock levels | 1.0× / 1.0× |
| Location libraries | Which pickface, reserve, and bay types the fit may use: your current set, our library of about 6,500 pickface and 150 reserve types, both, or a custom upload | Current set, 24 bay types |
Comparing futures is a rerun, not a new project. Change growth, a days-on-hand policy, or the rack set, and the whole model updates.
Your own racking can replace the default bay types: describe it in Bay Creator in the client portal.
Every SKU gets a home. Every location gets a bay.
Each SKU is assigned the pickface that holds its days-on-hand requirement within the utilization cap. Remaining inventory goes to reserve, and locations are packed into bays. Every result is compared with your current layout.
- −34%Location cube203,192 → 134,987 cu ft
- −27%Location footprint4,956 → 3,637 sq ft
- 73%Average pickface utilizationUp from 38% today
- 886Bays to buildAcross six bay types
Pickfaces run about 30% full. The fit takes most types past 70%.
Each location type is shown twice: what the current slotting achieves, and what an optimal fit would. The gap is recoverable space.
Which location types to add and retire.
Current count against fitted count for every pickface type.
886 bays across six types.
Locations packed into bay types give a bay count and loaded footprint for the building.
A model you can take into the capital meeting.
You leave with the evidence and the working files behind it, ready for operations, engineering, and finance.
Bay layout, exported to CAD.
Fitted bays are laid out and exported as a DXF drawing your engineers can open and edit. Shown here redrawn from the sample run's file.
- Report
Every figure with written commentary generated from your numbers.
- SKU-level workbook
One row per SKU with its current and fitted location, plus pickface, reserve, bay, and overall summaries.
- CAD bay layout
A DXF drawing of the fitted bays.
- Definitions and parameters
Column definitions for every table, and the exact settings behind the run so any result can be reproduced.
Figures are engine output from a baseline run on a fictional warehouse with generated SKUs, sales, and stock, reproduced as the engine draws them. The bay layout is redrawn from the engine's DXF file.