Inventory Shrinkage in 2026: How to Measure, Book, and Reduce the 1.5% Leak That Silently Kills Retail and E-Commerce Gross Margin

A 3-store boutique chain buys $420,000 of inventory at cost in 2025, records $612,000 of retail sales at keystone-plus, counts $38,000 of inventory on hand at year-end, and posts $352,000 of cost of goods sold from its POS. Gross margin shows 42.5%. The accountant's physical count ties to $28,400 of inventory — $9,600 less than the book, before vendor returns and in-transit are reconciled. That $9,600 is 2.3% of purchases and 1.6% of sales — the National Retail Federation's long-run shrink rate was 1.4–1.6% of retail sales in the 2023 survey, about $112 billion nationally — and it turns the reported 42.5% margin into a true 40.9% margin. No single theft caused it; four small leaks did, and none were booked until the count.

Inventory shrinkage is the difference between book inventory — what the perpetual record says you should have — and physical inventory — what you actually have. For retailers and e-commerce sellers, that difference is not a footnote; it is an unrecorded cost of goods sold that silently overstates margin, taxable income, and inventory assets while hiding where the leak lives. This guide explains how shrinkage hides, how to measure and book it correctly, where the 1.5% actually goes, and the cycle-counting and control habits that move the rate from industry-average toward best-in-class.

How Shrinkage Hides — Book vs. Physical and Why Margin Lies​

Perpetual inventory is the running book:

Beginning inventory
+ Purchases at cost
+ Freight-in
− Cost of goods sold (per POS / shipments)
− Vendor returns, allowances, and markdowns removed
− Recorded shrink (theft, damage, spoilage already booked)
= Book inventory — what the system says is on hand

Physical inventory is what the count finds — counted at cost, by SKU/lot, with in-transit and consignment positions reconciled.

Shrinkage = Book inventory − Physical inventory (when book exceeds physical; the reverse, book short of physical, is an overage — usually a receiving or counting error)

Shrinkage that hasn't been recorded is unrecorded COGS. Until you book COGS — Shrinkage, inventory on the balance sheet is overstated and gross profit is overstated by the same amount. Small businesses that skip a year-end physical and roll perpetual through the return simply overstate both.

The four causes, with how they enter (or fail to enter) the perpetual:

1. External theft — shoplifting and organized retail crime (ORC). The classic external cause, amplified in 2023–2025 by ORC groups who clear shelves for resale. POS records no sale; the shelf empties; perpetual never decrements. E-commerce equivalent: porch piracy after tender of delivery is a carrier-loss/shrinkage question that depends on who bears risk of loss, but the inventory left your books before the loss — the reorder cost is the leak.

2. Employee theft and sweethearting. Discount abuse, voided sales pocketed, merchandise passed to friends without a scan. POS does record something — a discount, a void, a zero — which is why discount and void-rate review catches it while a pure inventory count does not separately identify it.

3. Administrative / process errors. The largest single bucket in many small retailers' audits — not theft:

• Receiving: 12 units invoiced, 10 received, 12 booked. The 2-unit overstatement lives as shrinkage until the count.
• Miscounts: wrong SKU, wrong cost, lot confusion, in-transit double-counted as on-hand and again at the 3PL.
• Pricing: cost entered as retail or vice versa; freight-in not capitalized; vendor allowance netted to revenue instead of reducing purchase cost.
• Returns: customer return credited to revenue but not added back to inventory quantity; vendor return shipped but not removed from on-hand in the system.

4. Damage, spoilage, and obsolescence. Cracked merchandise, expired food, a fashion line that won't move and must be marked to net realizable value. Some of this should be recorded when discovered (damage write-down), not deferred to the count — the count is the backstop, not the primary booking.

For 2026 e-commerce sellers: damage in a 3PL, units lost in FBA inbound, and units stranded in FBA without a removal order are the modern variants of "shrinkage" that live in provider reports, not your shelves.

Measuring It — The Annual Count Isn't Enough​

The once-a-year wall-to-wall count produces a single shrinkage number — a single unknown that mixes theft, errors, damage, and obsolescence. That number satisfies the financial statement, but it cannot diagnose.

A cycle-count program diagnoses while it measures:

• A/B/C stratification. Count A items (top ~20% of SKUs by value or velocity) weekly or biweekly; B items monthly; C items quarterly. A 2% shrink on an $80 A-item is $1.60 per unit; on a $4.50 C-item it is $0.09 — accuracy where dollars live reduces variance per hour counted.
• Blind counts and recounts. Counter sees quantity expected = zero; second counter resolves variances above threshold before the record is adjusted.
• Cutoff discipline. All receipts, shipments, transfers, and returns dated before the count are included; nothing dated after is. In-transit inventory — to a store, to a customer, to a 3PL — is owned somewhere; decide where before the count and book the transfer, not after.
• Reconciliation before journal entry. For each variance over threshold, investigate before booking: was the receiving error in the last week? Was the 3PL report stale? Was a bundle broken into components?

Measurement rhythm that works for small retail/e-commerce:

• Daily: exception reports — negative on-hand, voids/discounts > threshold, 3PL discrepancy notes
• Weekly: cycle count on A items plus blind spot-checks on flagged SKUs; review receiving variances (invoiced vs. received)
• Monthly: book provisional shrinkage at a reserve rate (e.g., trailing-12-month shrink % × sales for the month), then true up quarterly after cycle counts
• Quarterly: full reconciliation of perpetual to provider statements (3PL on-hand report, FBA inventory ledger, marketplace returns), with vendor-returns-in-transit cleared
• Annually: wall-to-wall physical that ties to the general ledger — the only number that supports the year-end inventory asset on the tax return

Booking It — The Entries That Keep COGS Honest​

Shrinkage is an additional cost of goods sold, not an operating expense, unless your policy consistently classifies it below gross profit and discloses it (book-tax consistency matters — keep it in COGS for comparability and margin analysis).

Provisional shrinkage (month-end estimate between counts):

Dr COGS — Shrinkage (estimate)      $800
  Cr Accrued Shrinkage Reserve / Inventory Reserve    $800

That reserve is a contra-inventory or accrued liability that anticipates the unknown leak. It is not tax-deductible as a reserve — it is a book estimate that becomes deductible when the physical is taken and the specific shrinkage is identified.

At physical count — truing up:

Where book says $38,000 and physical is $28,400, after clearing known receiving and in-transit reconciling items that explain $1,200 of the gap, the remaining $8,400 is shrinkage:

Dr COGS — Shrinkage                  $8,400
  Cr Inventory                                    $8,400

If a reserve existed, reverse and replace:

Dr Accrued Shrinkage Reserve           $2,400   (cumulative estimates)
  Cr COGS — Shrinkage                             $2,400
Dr COGS — Shrinkage                  $8,400
  Cr Inventory                                    $8,400
  → Net incremental COGS $6,000 vs. prior estimates

Tax note: Inventory shrinkage discovered by physical count reduces ending inventory and increases COGS, reducing taxable income in that year — but a pure reserve without a count is not deductible. Obsolescence write-downs to lower of cost or market (or net realizable value) are deductible only when the goods are actually offered for sale at the reduced price or disposed of in specific fact patterns — a management memo that "this line is obsolete" without a markdown or disposition does not create a deduction.

Damage discovered outside the count:

Dr Loss — Inventory Damage              $450
  Cr Inventory                                    $450

Classify damaged-goods losses inside COGS if they are routine (spoilage, handling), below gross profit if truly abnormal and disclosed — but classify consistently.

E-commerce variants:

• 3PL shortage: COGS — Shrinkage / 3PL Claims Receivable until the carrier/3PL credit or claim resolves; then Cash / Claims Receivable
• FBA lost/damaged reimbursed by Amazon: reimbursements offset the shrinkage — Cash / Inventory or Cash / COGS — Shrinkage Recovery; don't net the recovery against purchases or revenue (it is a cost recovery, not a sale)
• Customer returns: on return, Inventory at cost / COGS reversal and refund liability — not revenue

Finding the 1.5% — Where to Look When the Number Is Big​

When shrinkage exceeds ~1.2% of sales, diagnose by source-specific metric, not total shrink:

External theft signal: shrink concentrates in high-theft categories — fragrances, denim, small electronics, cosmetics, sneakers — with no corresponding receiving variance. Tools: locked cases on high-risk SKUs, kept-stock in cage, exception-based video review on shelf-clearing patterns, ORC intelligence sharing via local retail associations.

Employee-theft signal: shrink tracks to discount rate, void rate, and no-sale rate by employee, not by product. A register with a 4% discount rate against a 1.2% store average is a flag. Require manager approval for post-voids, control the employee-purchase process, and rotate register assignments.

Process-error signal: shrink is diffuse, spikes after physical inventory or after a system migration, and correlates with receiving-variance reports. The fix is upstream: barcode-scan receiving against the PO (not keying), two-way match (PO vs. receipt) before the vendor invoice is approved, daily negative-on-hand alert clearing, and 3PL report reconciliation on a fixed weekday.

Damage/spoilage signal: shrink clusters in fragile, perishable, or handled categories, with a visible damage cage or spoilage log that doesn't match the booked write-downs. First-in, first-out (FIFO) rotation discipline, temperature logs where relevant, packaging standards, and a disposition workflow — markdown before shrink, donation before dump, destruction with certificate where required — reduce this bucket and create deductible documentation.

Set shrink targets not as a wish but as a control:

• Industry average 2023 (NRF): ~1.4% of retail sales at retail value (~1.5% at cost after conversion — retail shrink % is commonly reported at retail prices and translates to a smaller cost % for keystone retailers)
• Best-in-class small retail: 0.5–0.9% with disciplined cycle counts, receiving controls, and locked high-risk stock
• E-commerce pure-play: often lower at 0.4–1.0% excluding 3PL/carrier losses, but carrier and 3PL shortages must be tracked separately to see the true rate

Convert carefully: NRF's 1.4% at retail for a store at 2× keystone is about 2.8% of cost equivalent? No — retail $100 at 50% margin has $50 cost, shrink $1.40 at retail is shrink $0.70 at cost on $50 cost = 1.4% at retail ≈ 1.4% at cost when margin is keystone? Actually same numerator/denominator scaling: shrink cost = shrink retail × cost/retail ratio. For a given shrink dollars, both cost and retail scale together. So 1.5% at retail ≈ 1.5% at cost for mark-up uniformity — but don't mix them in the same ratio.

A Close That Fits the Count Calendar​

This month — baseline the number:

• Run a full physical or a complete rolling cycle within 30 days and book the shrinkage to COGS in that month — that entry is the "before" that justifies controls going forward. Reconcile inventory GL to the physical by SKU, clear vendor returns in transit, and confirm 3PL/FBA provider on-hand ties out.

Ongoing — make the count routine:

• Schedule the cycle-count calendar by ABC class for the next 12 months, assign blind counters, and set variance-investigation thresholds (e.g., >$50 or >2 units triggers a recount and receiving audit before the adjustment is posted). Weekly A-counts, monthly B-counts, quarterly wall-to-wall for the lane that needs it.

At month-end — reserve honestly:

• Book the provisional shrinkage percentage monthly (trailing-12 shrink % × sales at cost for the month) to the reserve; reverse and true up quarterly after cycle counts so monthly margin is honest, not just year-end margin. Show shrink in monthly gross-margin reporting — hiding it until December makes every merchandising decision for 11 months on a false margin.

At year-end — take, reconcile, and report:

• Count wall-to-wall with cutoff discipline, reconcile the shrinkage roll-forward (beginning unidentified + provisional charges − true-ups = ending physical shrink), keep the count sheets, recount logs, receiving variance package, and 3PL statements as one workpaper. On the return, ending inventory is the physical, not the perpetual — the shrinkage flowed through COGS in the adjustments above.

The Bookkeeping Connection​

Shrinkage rewards the habit that makes plain-text accounting powerful: every purchase, receipt, sale, return, damage cage ticket, and count sheet is a dated, SKU-tagged event — not a year-end plug from physical. When purchases by vendor, receipts by quantity, COGS by POS SKU, reserves by month, and counts by location live in the same version-controlled ledger, the story from "$420,000 purchased, $38,000 perpetual, $28,400 physical, $9,600 shrink (1.6% of sales)" to "$352,000 POS COGS + $8,400 shrink = $360,400 COGS, 40.9% true margin, reserve trued, workpapers tied to 3PL and vendor returns" is traceable and explainable to a manager who must decide what to lock, what to recount, and what to discontinue — and to a preparer who must set ending inventory to the physical, not the wish.

Simplify Your Financial Management​

Margin is not what the register printed; it is what the count proved after shrink, damage, and errors took their cut. Beancount.io gives you plain-text, version-controlled accounting where purchases, receipts, COGS, shrinkage, and physical inventory stay explicitly linked — no hidden reserves, no vendor lock-in, and AI-ready when you want help turning this week's cycle count into next month's margin. Get started for free and make gross margin earn its name after the count.