Skip to content

Warehouse Inventory Accuracy: How to Calculate It, With a Worked Example

by Ion Rotaru 20 Nov 2026 0 Comments

Picture a regional distributor, call it Northgate Supply, that carries $2,400,000 of inventory on its books across a single 40,000-square-foot warehouse. The controller says the counts are "about 99% right." Then the company's lender schedules a field exam, and nobody can say how that figure was calculated.

That gap is common. Inventory accuracy is simple arithmetic, but only if you decide what counts as a miss before you start. The sections below walk Northgate through the math, from a test count to the number a lender actually cares about.

The Inventory Accuracy Formula

Inventory record accuracy is the number of records without a material discrepancy, divided by the number of records tested, multiplied by 100. The word "material" does the work. Before any counting begins, Northgate defines it as a variance of more than 2 units or more than $100 on a location, whichever is hit first.

Northgate's team then tests 200 locations. Of those, 188 match the system within tolerance and 12 do not. That gives 188 divided by 200, or 94% accuracy. It is a long way from the 99% the controller assumed.

The tolerance you choose should be written down and applied the same way every time. A rate that quietly changes its definition from one quarter to the next is not a trend, just a number that moves.

Test in Both Directions: Sheet-to-Floor and Floor-to-Sheet

The 200 locations should not all come from the same direction. Northgate pulls 100 records from the system and walks to find them on the shelf, which is a sheet-to-floor test. It then picks 100 physical pallets and checks that each appears in the system at the right location and quantity, which is a floor-to-sheet test.

The second direction catches what the first cannot: unrecorded stock, duplicate labels, and goods sitting in the wrong bay. Auditors who observe physical counts routinely use both directions for this reason, along with receiving and shipping cutoff checks so that goods in transit land on the correct side of the count date.

The same two lists are what an outside valuation firm needs. When a lender requires independent support for collateral, firms such as Accountable Inventory Appraisers start from the perpetual inventory report and the physical evidence, so a warehouse that has already reconciled both gives the appraiser far less to untangle.

Why 94% by Count Can Be Worse Than It Looks in Dollars

Unit accuracy treats a $4 fitting and a $4,000 motor as equal records. Money does not behave that way. Of Northgate's 12 failed locations, 9 involve low-value hardware with a combined variance of $6,000. The other 3 involve high-value drive units and account for $48,000.

A record-count score of 94% therefore hides a concentrated dollar problem. The practical fix is to report two rates side by side: accuracy by record and accuracy by value. If the high-value tier fails at a much higher rate than the rest, that tier needs tighter controls, not a bigger sample of bolts.

Calculating Shrinkage in Units and Dollars

Shrinkage is book inventory minus physical inventory. Northgate's ledger shows $2,400,000. After the full count is valued, physical stock comes to $2,328,000. The difference is $72,000, or 3.0% of book value.

The number alone does not say why. Northgate sorts the variance into causes: timing differences from unposted receipts, unit-of-measure errors, mislocated goods, damage, picking mistakes, and theft. A $72,000 gap that is mostly unposted receipts is a paperwork problem. The same gap driven by theft is a security problem, and the two need different fixes.

From Inventory on Hand to Eligible Inventory

Here is where warehouse data meets finance. "Inventory on hand" and "eligible inventory" are different numbers, and a lender lends against the second one. The Office of the Comptroller of the Currency's asset-based lending guidance notes that borrowers are normally expected to certify the amount, type, and condition of their inventory, and that lenders commonly reserve the right to audit it.

Northgate's physical stock includes $90,000 of consigned goods it does not own, $60,000 of damaged or expired product, and $150,000 of items with no sales in over a year. Here is how the figures flow from the ledger to the borrowing base:

Step

Amount

 

Book inventory per ledger

$2,400,000

Physical inventory after count

$2,328,000

Less consigned goods

$90,000

Less damaged or expired

$60,000

Less obsolete stock

$150,000

Eligible inventory

$2,028,000


From there, the lender applies an advance rate. As a reference point, federal bank regulation (Appendix C to 12 CFR Part 327) describes advance rates generally no greater than 65% of eligible inventory valued at the lower of FIFO cost or market. On that basis, Northgate's $2,028,000 supports roughly $1,318,200.

The same regulation allows up to 85% of net orderly liquidation value, or NOLV, where an appraisal exists. Suppose an independent appraisal puts NOLV on the eligible stock at $1,400,000. Eighty-five percent of that is $1,190,000, which is lower than the cost-based figure. That is typical: liquidation value reflects what the goods would net after discounts, commissions, freight, and handling, not what Northgate paid for them.

These percentages are examples, not universal rules. Actual advance rates depend on the loan agreement, the inventory type, marketability, concentration, and the lender's own underwriting. The National Credit Union Administration's examiner guidance, for instance, describes collateral value as generally the lesser of cost or liquidation value, and points examiners to quantity, condition, location, and the reliability of inventory controls.

Cycle Counting to Keep the Rate From Slipping

A full physical count is a snapshot. Cycle counting is the habit that keeps the next snapshot from surprising anyone. Northgate moves to a risk-based schedule: its 3 high-value drive unit lines are counted monthly, mid-value items quarterly, and low-value stable items twice a year.

The point is to match effort to dollars, not to hit a fixed calendar. No law dictates "weekly" or "annual" counts for every business, and the right cadence depends on value, movement, shrinkage history, and any lender requirements.

When a count turns up a variance, classify it before adjusting the system. Unposted receipts, duplicate records, and mislocated goods each point to a specific process breakdown. If every variance is simply netted out, the root cause survives and the next count finds it again.

What Good Warehouse Data Looks Like Before a Lender Visits

Northgate's cleanup does not require new software, only discipline. Location labels are corrected so every pallet can be found. Consigned and customer-owned stock gets an ownership flag in the system, and damaged or quarantined goods get their own status code so they never inflate available stock.

Aging reports are run by product family, because a company-wide turnover average can conceal a dead category. Days inventory outstanding, calculated as average inventory divided by cost of goods sold times 365, helps management spot slow movers, but it does not prove the stock is saleable or eligible as collateral.

The result is a documented trail from ledger to shelf and back. When Northgate's lender arrives, the question shifts from "can you prove these numbers" to "how do these numbers hold up," which is a much better conversation to be having.

Picture a regional distributor, call it Northgate Supply, that carries $2,400,000 of inventory on its books across a single 40,000-square-foot warehouse. The controller says the counts are "about 99% right." Then the company's lender schedules a field exam, and nobody can say how that figure was calculated.

That gap is common. Inventory accuracy is simple arithmetic, but only if you decide what counts as a miss before you start. The sections below walk Northgate through the math, from a test count to the number a lender actually cares about.

The Inventory Accuracy Formula

Inventory record accuracy is the number of records without a material discrepancy, divided by the number of records tested, multiplied by 100. The word "material" does the work. Before any counting begins, Northgate defines it as a variance of more than 2 units or more than $100 on a location, whichever is hit first.

Northgate's team then tests 200 locations. Of those, 188 match the system within tolerance and 12 do not. That gives 188 divided by 200, or 94% accuracy. It is a long way from the 99% the controller assumed.

The tolerance you choose should be written down and applied the same way every time. A rate that quietly changes its definition from one quarter to the next is not a trend, just a number that moves.

Test in Both Directions: Sheet-to-Floor and Floor-to-Sheet

The 200 locations should not all come from the same direction. Northgate pulls 100 records from the system and walks to find them on the shelf, which is a sheet-to-floor test. It then picks 100 physical pallets and checks that each appears in the system at the right location and quantity, which is a floor-to-sheet test.

The second direction catches what the first cannot: unrecorded stock, duplicate labels, and goods sitting in the wrong bay. Auditors who observe physical counts routinely use both directions for this reason, along with receiving and shipping cutoff checks so that goods in transit land on the correct side of the count date.

The same two lists are what an outside valuation firm needs. When a lender requires independent support for collateral, firms such as Accountable Inventory Appraisers start from the perpetual inventory report and the physical evidence, so a warehouse that has already reconciled both gives the appraiser far less to untangle.

Why 94% by Count Can Be Worse Than It Looks in Dollars

Unit accuracy treats a $4 fitting and a $4,000 motor as equal records. Money does not behave that way. Of Northgate's 12 failed locations, 9 involve low-value hardware with a combined variance of $6,000. The other 3 involve high-value drive units and account for $48,000.

A record-count score of 94% therefore hides a concentrated dollar problem. The practical fix is to report two rates side by side: accuracy by record and accuracy by value. If the high-value tier fails at a much higher rate than the rest, that tier needs tighter controls, not a bigger sample of bolts.

Calculating Shrinkage in Units and Dollars

Shrinkage is book inventory minus physical inventory. Northgate's ledger shows $2,400,000. After the full count is valued, physical stock comes to $2,328,000. The difference is $72,000, or 3.0% of book value.

The number alone does not say why. Northgate sorts the variance into causes: timing differences from unposted receipts, unit-of-measure errors, mislocated goods, damage, picking mistakes, and theft. A $72,000 gap that is mostly unposted receipts is a paperwork problem. The same gap driven by theft is a security problem, and the two need different fixes.

From Inventory on Hand to Eligible Inventory

Here is where warehouse data meets finance. "Inventory on hand" and "eligible inventory" are different numbers, and a lender lends against the second one. The Office of the Comptroller of the Currency's asset-based lending guidance notes that borrowers are normally expected to certify the amount, type, and condition of their inventory, and that lenders commonly reserve the right to audit it.

Northgate's physical stock includes $90,000 of consigned goods it does not own, $60,000 of damaged or expired product, and $150,000 of items with no sales in over a year. Here is how the figures flow from the ledger to the borrowing base:

Step

Amount

 

Book inventory per ledger

$2,400,000

Physical inventory after count

$2,328,000

Less consigned goods

$90,000

Less damaged or expired

$60,000

Less obsolete stock

$150,000

Eligible inventory

$2,028,000

From there, the lender applies an advance rate. As a reference point, federal bank regulation (Appendix C to 12 CFR Part 327) describes advance rates generally no greater than 65% of eligible inventory valued at the lower of FIFO cost or market. On that basis, Northgate's $2,028,000 supports roughly $1,318,200.

The same regulation allows up to 85% of net orderly liquidation value, or NOLV, where an appraisal exists. Suppose an independent appraisal puts NOLV on the eligible stock at $1,400,000. Eighty-five percent of that is $1,190,000, which is lower than the cost-based figure. That is typical: liquidation value reflects what the goods would net after discounts, commissions, freight, and handling, not what Northgate paid for them.

These percentages are examples, not universal rules. Actual advance rates depend on the loan agreement, the inventory type, marketability, concentration, and the lender's own underwriting. The National Credit Union Administration's examiner guidance, for instance, describes collateral value as generally the lesser of cost or liquidation value, and points examiners to quantity, condition, location, and the reliability of inventory controls.

Cycle Counting to Keep the Rate From Slipping

A full physical count is a snapshot. Cycle counting is the habit that keeps the next snapshot from surprising anyone. Northgate moves to a risk-based schedule: its 3 high-value drive unit lines are counted monthly, mid-value items quarterly, and low-value stable items twice a year.

The point is to match effort to dollars, not to hit a fixed calendar. No law dictates "weekly" or "annual" counts for every business, and the right cadence depends on value, movement, shrinkage history, and any lender requirements.

When a count turns up a variance, classify it before adjusting the system. Unposted receipts, duplicate records, and mislocated goods each point to a specific process breakdown. If every variance is simply netted out, the root cause survives and the next count finds it again.

What Good Warehouse Data Looks Like Before a Lender Visits

Northgate's cleanup does not require new software, only discipline. Location labels are corrected so every pallet can be found. Consigned and customer-owned stock gets an ownership flag in the system, and damaged or quarantined goods get their own status code so they never inflate available stock.

Aging reports are run by product family, because a company-wide turnover average can conceal a dead category. Days inventory outstanding, calculated as average inventory divided by cost of goods sold times 365, helps management spot slow movers, but it does not prove the stock is saleable or eligible as collateral.

The result is a documented trail from ledger to shelf and back. When Northgate's lender arrives, the question shifts from "can you prove these numbers" to "how do these numbers hold up," which is a much better conversation to be having.

Leave a comment

Please note, comments need to be approved before they are published.

Thanks for subscribing!

This email has been registered!

Shop the look

Choose Options

Edit Option
Back In Stock Notification
is added to your shopping cart.
this is just a warning
Login