Inventory Shrinkage Causes and Prevention: Practical Guide (2026)

Inventory shrinkage is the gap between the stock your system says you have and the stock you can physically find, when no documented transaction explains the difference. It shows up in manufacturing as unrecorded scrap, damaged components, receiving errors, material consumed on the line without a scan, and theft — and every one of those leaks margin while quietly corrupting the inventory data your planning runs on. This guide covers inventory shrinkage causes and prevention on a plant floor: how to measure the gap and which controls actually hold.

The version of this problem most sites describe is a retail store problem: shoplifted merchandise on the sales floor. Manufacturing losses look different. Material moves in bulk, gets consumed in stages, sits in bins between machines, and changes form as it becomes work in process, finished goods, and eventually scrap. A missing pallet of resin in a warehouse and a missing carton on a store shelf are both shrinkage, but they need different controls and different investigations.

That is the framing for everything below. Causes are grouped by what actually goes missing, then by where in the facility it goes missing, and each one is paired with the control that stops it. If your accuracy number is off and you cannot say why, the measurement section comes first.

Table of Contents

• What Is Inventory Shrinkage?
  • Shrinkage, inventory loss, and write-off are not the same thing
  • Why book inventory diverges from physical inventory
• How to Measure Inventory Shrinkage and Accuracy
• Inventory Shrinkage Causes and Prevention by Loss Type
• How Inventory Shrinkage Causes and Prevention Differ by Location
• Why Inventory Shrinkage Often Goes Unreported
  • Netting variances against scrap
  • Mixing nonconforming inventory with usable stock
  • Delaying transaction corrections
  • Using an inaccurate bill of material
  • Failing to record customer returns
  • Relying on annual counts without reconciliation
• A Practical Inventory Shrinkage Prevention Program
• How to Build Cycle Counts and Root-Cause Checks That Work
  • How to pick count priorities
  • How to run a count that produces evidence
• Which Metrics Should Managers Track?
• Frequently Asked Questions
  • What is the difference between inventory shrinkage and manufacturing scrap?
  • How often should a warehouse perform cycle counts to find inventory shrinkage?
  • Should inventory shrinkage be included in the cost of goods sold?
  • What is an acceptable inventory accuracy rate for manufacturers?
  • How can a company reduce shrinkage without slowing operations?
  • Can insurance cover inventory lost through theft or unexplained inventory variance?
• Conclusion: Start With One High-Risk Inventory Area

What Is Inventory Shrinkage?

Inventory shrinkage is the difference between recorded inventory and physical inventory, expressed in units or dollars, when no approved transaction accounts for it.

Three ideas sit underneath that definition, and mixing them up is why shrink numbers get argued about instead of fixed.

Shrinkage, inventory loss, and write-off are not the same thing

A shrinkage event is unexplained. An inventory loss is explainable but still undesirable — a pallet damaged by a forklift, a container broken in transit. A write-off is the accounting action taken after you have decided the loss is real and will not come back.

Every write-off starts as shrinkage, but not every write-off is shrinkage. Scrapping a batch that failed contamination testing is a documented loss with a cause code and an approval. It still costs you the material, but it is planned, measurable, and it belongs in your cost of goods sold, not in your shrink rate.

Obsolescence and spoilage sit alongside these. A resin lot that ages past its shelf life is obsolescence. Adhesive that cures in the drum is spoilage. Both are real losses, both reduce usable inventory, and both belong in a separate line of your variance report so you can see them apart from the unexplained gap.

Term	What it means	Has a cause code?	Belongs in shrink rate?
Inventory shrinkage	Recorded stock exceeds physical stock with no approved transaction to explain it	No, or provisional	Yes
Inventory loss	Known loss from a specific event such as damage in transit	Yes	Sometimes, tracked separately
Write-off	Accounting removal of inventory value from the balance sheet	Yes	No, this is the accounting entry
Scrap	Material intentionally discarded because it failed a specification	Yes	No, tracked as yield loss
Obsolescence	Stock that will not be used before it becomes unusable	Yes	No, tracked as an aging provision
Spoilage	Perishable or time-sensitive stock that degraded in storage	Yes	No, tracked as spoilage

Why book inventory diverges from physical inventory

Systems do not go wrong on their own. A book-to-physical gap appears when a movement happened and the transaction did not, or when a transaction was posted for something that did not physically occur.

In a plant with paper travelers, that means a work order was closed without recording consumption. With a WMS and barcode scanning, it means someone used a manual override, or a receipt was posted before the material was counted in. Both produce the same arithmetic: the system says 400 pounds, the rack holds 360, and nobody can point to the transaction that moved 40 pounds.

How to Measure Inventory Shrinkage and Accuracy

Three numbers answer most shrinkage questions: dollar shrinkage, inventory accuracy, and count variance. Compute them on a fixed period and a fixed baseline, or the trend is meaningless.

Dollar shrinkage is recorded value minus counted value for the scope you counted.

Unit variance is recorded units minus counted units for a specific SKU at a specific location.

Inventory accuracy is counted locations that matched the system divided by all locations counted. The location count is a location-location match, not a SKU-SKU match, because most programs count a bin and check whether everything in it was right.

Pick your baseline carefully. Most published retail shrink figures are calculated against sales value, while a plant or warehouse usually wants shrinkage against recorded inventory cost or against the value of the SKUs counted. Mixing a retail-value numerator with a cost denominator produces a number that looks small and means nothing. State the basis on the report, every month, so nobody has to guess which one they are reading.

The period matters too. Monthly catches transactional errors within about a month of them happening. Quarterly smooths out seasonal noise. Annual counts alone leave you with a number once a year and no idea when the loss started.

Method	What it measures	Frequency	Best for	Weakness
Cycle counting	Accuracy of high-risk SKUs on a rolling schedule	Weekly to monthly per ABC class	Continuous visibility and root-cause evidence	Needs frozen movements and trained counters
Annual physical count	Total book value accuracy at a point in time	Once or twice a year	Financial audit and year-end valuation	Long blind spots, high disruption cost
Continuous reconciliation	Transaction-level agreement between ERP, WMS and floor	Daily or weekly, exception-driven	Finding the transaction that broke the chain	Requires clean interfaces and cause codes

Scope matters more than size. Shrinkage is far easier to investigate when a count starts with a limited group of high-value or high-velocity items instead of the whole catalog. A targeted count tells you something in a shift. A full count tells you a number in a week.

Inventory Shrinkage Causes and Prevention by Loss Type

Theft still drives a large share of reported shrinkage, but it is no longer the only cause that moves the number. Process and control errors sit close behind and, unlike theft, they are almost entirely fixable.

Here are the five that account for the bulk of loss in most facilities.

1. External theft — stock removed from the yard, dock, or an unsecured cage by someone who is not an employee. In a plant this usually looks like a bad night on a trailer or a gate that was propped.
2. Internal theft — pilferage by employees and contractors: taking consumables home, feeding personal jobs from production material, or skipping a scan so the stock stays invisible on the books.
3. Process and data entry errors — receipts posted for the wrong quantity, a receipt posted before the count, an issue posted to the wrong work order, or a manual adjustment nobody can explain later.
4. Vendor fraud and short shipments — invoices matching paperwork when fewer units shipped, unauthorized substitutions, or quality issues accepted at receiving and only found months later.
5. Damage, spoilage, and expiration — material crushed in a rack, film and drums degraded by heat, adhesive curing in the drum, shelf life exceeded in a lot nobody flagged.

Beyond those five, manufacturing-specific losses belong in the same conversation: unrecorded scrap from start-up waste and purges, work-in-process variance between issued and consumed material, MRO consumables that walk off with a maintenance job, and tooling or molds that leave the tool room and do not come back.

Cause	How it looks on the floor	Prevention control	Owner
Counting error	Bin counted short because a case was opened and repacked	Blind cycle counts, count sheet shows quantities not locations, second count above threshold	Inventory control
Unrecorded scrap	Purges and start-up waste with no scrap ticket	Scrap ticket with cause code at the machine, scrap bin weighed per shift	Production supervisor
Damage	Forklift impact, crushed drums, split pallet in the rack	Rack guards, aisle speed rules, damage inspection at putaway and pick	Warehouse manager
External theft	Missing pallets at the dock, cut locks, gaps in the yard	Sealed trailers, badge and camera coverage at docks, yard seal checks	Facilities and security
Internal theft	Consumables disappearing from the crib, scans skipped at issue	Role-based access, scan-to-confirm at issue, crib issue logging by name, no unscanned pick paths	Operations manager
Receiving error	Book quantity higher than what came off the truck	Count and scan at the dock before the receipt posts, blind receiving, over/short tolerance alerts	Receiving supervisor
Issuance error	Material issued to a work order that did not consume it	Scan confirmation at issue, backflush verification, WIP reconciliation at work order close	Production control
Supplier shortage	Fewer cases in the carton than the packing list shows	Count at receiving, photograph the seal and carton, file shortage claims within the claim window	Purchasing
Shipping loss	Short shipment, carrier damage, in-transit discrepancy	Load verification scan, photo at pickup, weight capture, claims filed in days not months	Logistics
Record failure	Adjustment made without a cause code or supporting note	Mandatory cause code, blocked overrides above threshold, monthly review of the change log	Finance controller

How Inventory Shrinkage Causes and Prevention Differ by Location

Shrinkage does not spread evenly across a facility. It concentrates where material changes hands without a record, and the control that matters is different at each hand-off.

The pattern is consistent. Every node where stock is received, staged, consumed, or shipped has a distinct failure mode, and a distinct control that closes it.

Location	Typical failure	Control that stops it
Receiving dock	Short or over-shipment accepted without a count	Blind count against the packing list, scan every unit before the receipt posts
Raw material storage	Mislabeled or unscanned putaway, mixed lots, wrong FIFO sequence	System-directed putaway, lot and serial capture, FIFO or FEFO enforced at putaway
Work in process	Issued material never consumed, partial work orders left open	Backflush verification, WIP reconciliation at close, open-order review weekly
Production line	Start-up waste, purges, regrind and scrap not recorded	Scrap ticket with cause code at the station, shift-end reconciliation, downtime tagging
Finished goods warehouse	Short picks, unrecorded samples, sample and giveaway stock leaving the building	Scan-to-confirm picks, sample log with approver, controlled pick faces
Packaging area	Packaging material consumed without issue, overproduction not recorded	Issue packaging by work order, count finished units at pallet close, tie-back to the order
Outbound shipping	Loaded but not scanned, trailer sealed before paperwork, wrong pallet loaded	Load verification scan, seal number recorded, weight capture, photo at pickup

Cross-dock and multi-location transfers deserve their own line. A transfer that moves stock between two internal locations without a system transaction creates phantom inventory at the source and phantom shortage at the destination. Both eventually get adjusted away, and the adjustments get logged as shrink even though nothing was lost. It is common enough to show up in most plant variance reviews, and it disappears the moment you treat every internal movement as a real receipt and issue.

Why Inventory Shrinkage Often Goes Unreported

Most shrinkage is not hidden deliberately. It is absorbed into other lines until nobody owns it.

Six patterns do most of the burying.

Netting variances against scrap

When a physical shortage is offset against recorded scrap, the shrink looks small enough to ignore. The two numbers are then analyzed as one, and the team concludes that the problem is production waste when part of it is a missing pallet in a rack.

Mixing nonconforming inventory with usable stock

Quarantined material that sits in the same bin as released material gets counted, and the count is wrong in both directions. Put it in a physically separate hold area with its own status, and the counting problem disappears along with the argument.

Delaying transaction corrections

A receipt posted three weeks late still gets posted, and the three-week window is exactly where the loss hides. Corrections that arrive after month-end close will not appear in the variance report that would have flagged them.

Using an inaccurate bill of material

If the standard says 3.2 kilograms and the actual issue is 3.6, every work order creates a small variance and the total looks like a normal yield curve. A BOM that was never validated after a process change quietly manufactures shrink all year.

Failing to record customer returns

Returned material that is not logged as a receipt can be sold, scrapped, or simply never exist as far as the system is concerned. The comparison that would have caught it is the one nobody runs.

Relying on annual counts without reconciliation

A year-end count produces one number with no timeline. You learn that the discrepancy was 1,200 units and never learn which transaction caused it. Monthly reconciliation of movement and count is what turns a number into a cause.

A Practical Inventory Shrinkage Prevention Program

Prevention works as a stack. Any single control can be bypassed; a stack where each layer catches what the layer above missed is what holds.

1. Name an accountable owner. Assign one role the accuracy number and one role the write-off. Split ownership means nobody owns the total.
2. Fix SKU and location discipline. One SKU per bin location, one location per SKU, real lot and serial capture. A mislabeled bin is a count that will be wrong twice. If you are still deciding how to capture that data, RFID vs barcode for inventory tracking breaks down the tradeoffs.
3. Make transactions mandatory. No stock moves without a scan or a signed transaction. If a manual path exists for expediting, log it the same way with a reason code.
4. Count by risk, not by calendar. A items counted monthly, B items quarterly, C items annually, with serialized and high-theft-risk goods always on the highest cadence.
5. Restrict adjustments. Role-based access, an approval step above a dollar threshold, a mandatory cause code, and a change log that finance reviews monthly.
6. Secure the physical layer. Cameras, badge access, cages, and controlled crib issue. Treat this as the layer that raises the cost of opportunistic loss, not the layer that does the work.
7. Reconcile every hand-off. ERP to WMS to floor, at least monthly, with exceptions listed and worked rather than netted.
8. Investigate repeat variances. The same SKU going short three quarters in a row is a process defect, not a counting error. Assign a corrective action with a named owner and a date, then verify it held.

One caution on the security layer. Heavy-handed surveillance damages the floor and erodes the reporting you need. Procedural discipline, clear store-keeping rules, visible supervision, and greeting routines tend to work better than an accusatory atmosphere. If your controls only work when nobody suspects anyone, they will stop working the day turnover spikes.

How to Build Cycle Counts and Root-Cause Checks That Work

Cycle counting is the control that turns a monthly number into a named cause. The design details matter more than the frequency, and the choice of method is covered in cycle counting vs physical inventory.

How to pick count priorities

Use ABC analysis on value, and layer velocity and risk on top. A class holds roughly 80 percent of the value, B items the next 15, and C items the remainder, so counting effort should follow that curve rather than being spread evenly. ABC inventory analysis explained walks through how to build the classes.

The 80/20 rule people ask about in inventory is the same logic: a small share of SKUs carries most of the value and most of the variance, and those are the ones you count most often. This is also why a full count of everything is usually the wrong first move.

How to run a count that produces evidence

1. Freeze movements for the SKUs and locations in the count. A count with picks happening during it is a count of a moving target.
2. Count blind. The counter does not see the system quantity. Revealing it first anchors the count to whatever the system says.
3. Count locations, not SKUs. A location that matches means everything in that location matched, and it is far faster than a SKU-level sweep.
4. Require a second count above a threshold, or for any high-value or serialized item.
5. Record the evidence at the time: photos of the bin, the lot number, the seal, the label condition. Evidence captured three weeks later is not evidence.
6. Trace the discrepancy to transactions — every receipt, issue, move, and adjustment on that SKU in the count window.
7. Escalate repeat losses on the same location or the same item to a formal root-cause review with a corrective action.

Two design flaws quietly destroy count reliability. Unblinded counts and predictable schedules let experienced operators learn when a bin is checked, which raises the count error and lowers the count’s value as evidence. Fix both: blind counts, and a schedule driven by a rule the counter does not control.

Which Metrics Should Managers Track?

Track a small set, review them on a fixed cadence, and give each one an owner. Nine metrics cover the operation well enough to run a program on.

Metric	What it tells you	Review cadence	Owner
Inventory accuracy	Location-to-location match rate	Monthly	Inventory control
Shrinkage rate	Value lost as a share of recorded inventory, on a stated basis	Monthly	Operations manager
Count variance value	Dollar value of variances found, split by cause code	Weekly	Inventory control
Unrecorded scrap rate	Scrap value as a share of material issued to production	Weekly per line	Production supervisor
Receiving-to-putaway time	How long received stock sits unprocessed, an unrecorded-stock proxy	Weekly	Warehouse manager
Transaction error rate	Postings later reversed or adjusted, per thousand transactions	Monthly	Finance controller
Adjustment value	Total value of inventory adjustments posted in the period	Monthly	Finance controller
Repeat-variance rate	Share of SKUs or locations short more than once in a period	Quarterly	Operations manager
Days to root cause	Average days from variance found to cause assigned	Monthly	Operations manager

Set thresholds rather than reacting to noise. A variance under a defined dollar value on a C item is rarely worth the count time. Repeat-variance rate is the metric that tells you whether your fixes are working, because accuracy can stay flat while you slowly change which problems you have.

Frequently Asked Questions

Conclusion: Start With One High-Risk Inventory Area

The fastest way into inventory shrinkage causes and prevention is to stop trying to fix the whole facility at once. Pick the location where value and variance history overlap, and work that one area end to end.

Choose the area, pull the last 60 days of receipts, issues, moves, and adjustments for its top SKUs, and reconcile them. Run one controlled blind count with movements frozen. Classify the variance against the cause list above, assign a corrective action with a named owner and a due date, then re-count in 30 days to confirm it held. Repeat on the next area.

Inventory shrinkage causes and prevention is a reporting discipline before it is a security problem. Once every loss carries a cause code and an owner, the trend tells you which control is missing.