Inventory shrinkage costs US retailers and ecommerce operations over $100 billion annually. Your operation’s share of that number is a direct profitability drain — and most of it is attributed to “unknown causes” because the tracking data doesn’t exist to identify the source.
Process design creates the tracking data. Without it, shrinkage is a mystery. With it, it’s a solvable problem.
What Most Operations Get Wrong When Investigating Shrinkage
The instinct when shrinkage increases is to focus on theft prevention: cameras, locked cages for high-value items, bag checks at exit. Theft is a real shrinkage source, but it typically accounts for 35-45% of total shrinkage in most fulfillment operations.
The majority of shrinkage is process-based: receiving errors that overstate inventory counts, picking errors that deplete inventory without record, damaged items that are disposed of without system update, and cycle count discrepancies that accumulate from small errors over time.
Theft-prevention investment that ignores process-based shrinkage addresses less than half the problem.
The diagnostic challenge is that process errors and theft look identical in aggregate inventory data. Both show up as missing units. Without transaction-level records of where each unit went at each workflow step, distinguishing process error from theft requires guesswork.
Process-based shrinkage is solvable without cameras or security upgrades. It requires transaction records — a digital audit trail that shows exactly what happened to every unit at every workflow step.
A Criteria Checklist for Shrinkage Reduction
Receive-to-Light Inbound Verification
Receiving errors are the first shrinkage source. Units that arrive damaged and are disposed of without system update, units miscounted at receiving, and units that arrive but are never formally received into the system are all invisible in your inventory counts. Scan-and-confirm receiving that requires each inbound unit to be individually recorded creates an inbound transaction record that can be reconciled against supplier invoices.
Pick-Level Transaction Audit Trail
Every pick event should generate a record: which SKU, which bin, which quantity, which worker, which time. Warehouse sorting solution hardware with guided confirmation creates this record automatically at each pick. When inventory shows fewer units than the system expects, the pick transaction record shows exactly when the depletion happened and which worker processed it. This data separates process errors from theft by time and worker attribution.
Dimensional scale Weight Verification at Pack
A weight check at the pack station that compares each closed package against the expected weight for the order detects over-picks (extra units placed in the package) and under-picks (missing units). Both represent inventory discrepancies — over-picks deplete inventory faster than orders justify, under-picks create customer issues while misrepresenting pick completion.
Disposal and Damage Recording
Every unit that leaves active inventory outside of a sale must be recorded: damage disposal, QC rejection, returns processing. Unrecorded disposals are invisible shrinkage. Implement a mandatory disposition form — digital or scan-based — for every unit removed from inventory for non-sale reasons.
Practical Tips for Shrinkage Diagnosis
Separate shrinkage by zone first. Different zones have different shrinkage patterns. High-value items in the primary pick zone may show theft-pattern shrinkage (frequent, small quantities of high-value SKUs). Receiving zones may show process-pattern shrinkage (count discrepancies spread across SKUs). Isolate the zone with the highest shrinkage rate before implementing solutions.
Run a reconciliation sprint for your top 50 shrinkage SKUs. For each high-shrinkage SKU, trace its entire 90-day transaction history: receiving records, pick records, disposition records, and cycle count adjustments. The transaction that shows a count reduction without a corresponding record is the shrinkage event. This trace is only possible if transaction records exist.
Implement worker-attributed picking. Knowing that Worker A processed a specific pick doesn’t mean Worker A stole the item — but it narrows the investigation significantly and makes process error more distinguishable from theft. Operations where picking is anonymous cannot attribute discrepancies below the zone level.
Use cycle count results as shrinkage signals, not just inventory corrections. When a cycle count shows a discrepancy, record the discrepancy and trigger an investigation before adjusting the count. The investigation determines whether the discrepancy is a process error or a theft signal. Operations that correct discrepancies without investigating are destroying the evidence.
The Shrinkage Cost Model
Industry average shrinkage rate: 1-2% of inventory value annually. For an operation holding $500,000 in inventory, that is $5,000-10,000 in annual shrinkage.
Operations with full transaction audit trails typically achieve 0.3-0.5% shrinkage rates. The difference at $500,000 inventory value: $3,500-8,500 in annual shrinkage cost reduction.
The process design investment that creates audit trails — guided pick confirmation, scan-based receiving, disposal recording — also delivers pick accuracy and throughput improvements. Shrinkage reduction is the third benefit of the same investment.