Resources / Core Concepts
Cycle Counting: Why It's Still Manual (and What Changes That)
This cycle counting guide shows which counting methods work, how to measure inventory record accuracy, and what RFID and vision change about counting by hand.
7 min read
· Updated
November 2026
At a glance
The definition
Cycle counting checks a small, planned slice of locations or items on a rolling schedule, so records are verified continuously instead of once a year.
Why it matters
Picks, replenishment, and customer promises all depend on the record matching the shelf. Small errors compound into short picks and stockouts.
Why it's manual
Counting needs eyes and hands in the bin. Barcodes confirm what was scanned, not what is sitting there unscanned.
Cycle counting is an inventory auditing method that counts a small, scheduled subset of locations or items every day or week, so record accuracy is checked and corrected continuously instead of in one annual physical count. This cycle counting guide covers the methods, the accuracy math, and what is starting to change the manual work.
The WMS says a slot holds 12 units. The picker finds 9. The pick goes short, the order waits, someone investigates, and an adjustment gets posted. Repeat that across thousands of locations and you get a building where nobody fully trusts the numbers, planners hold extra stock just in case, and the annual physical count shuts the operation down for a weekend.
Cycle counting is the fix most teams already know. It is also the first task cut when a shift runs short, because it is still almost entirely manual.
What is cycle counting vs. a physical inventory?
A physical inventory counts everything at once, usually with operations paused. Cycle counting counts a small, planned slice of locations or items every day or week while the building keeps running, so every location gets counted on a rotating schedule.
A well-run program can also change how year-end audits work. Under PCAOB auditing standard AS 2510, when well-kept perpetual inventory records are checked periodically against physical counts, the auditor's observation procedures usually can be performed during or after the end of the period under audit. Whether that lets you scale back a wall-to-wall count is a decision for your finance team and auditors. The records have to earn it first.
Cycle counting methods
ABC counting
Classify items by value, velocity, or both, and count the A items most often. For example, you might count A items monthly, B items quarterly, and C items twice a year. The weakness: slow-moving C items in rarely visited locations are where errors hide longest.
Control group counting
Choose a small group of items and locations and count them repeatedly over a short period. Because they are counted so often, any variance points to a process problem rather than a counting mistake. Use control groups to test your counting procedure before you scale the program, or to diagnose one stubborn error.
Opportunity-based counting
Count when someone is already at the location:
When a location reaches zero or a low quantity, and the picker confirms what is left.
During replenishment, before new stock goes into the forward location.
After a short pick or any pick exception at that location.
When an empty location is about to be reassigned.
Opportunity counts are cheap because the travel is already done, and they land at the moments when an error is easiest to see.
Location sweeps and random samples
Sweeping an aisle or zone in sequence catches items stored in the wrong place, which item-based counts miss. Random samples give you an unbiased estimate of accuracy across the building.
Most programs combine methods: ABC for coverage, opportunity counts for timing, sweeps for misplaced stock, and control groups for diagnosis.
SKU fit check
Send an email to
Measuring inventory record accuracy
Inventory record accuracy (IRA) is the share of counted records that match the physical count within an agreed tolerance.
Formula. Records within tolerance ÷ records counted × 100.
Define the record. Measure at the location level (a SKU in a specific slot), not just the SKU total. A SKU can be right in total and wrong in every location.
Set tolerances by class. Zero tolerance for high-value or controlled items, a small tolerance for low-value items stored in high counts.
Don't net it out. An overage in one location and a shortage in another are two errors, not zero.
Track adjustments too. Adjustment value and frequency by zone, shift, and process show where errors start.
Records drift further than most teams assume. In a study of nearly 370,000 inventory records across 37 stores of one retailer, DeHoratius and Raman (2008) found that 65% of the records were inaccurate. That was a retail store setting, not a warehouse, but the mechanism is the same: every unrecorded or misrecorded physical move makes the record a little less true.
Why cycle counting is still manual
Barcodes confirm transactions, not presence. A scan proves a unit passed through a step. It says nothing about what is sitting in a bin that nobody scanned.
Each-level counting is physical. Counting small items means opening totes, separating mixed contents, and checking the back of the shelf.
Counts compete with the work. Counting a location that is being picked creates false variances, so counts get pushed around peak activity, often onto the same short-staffed shifts.
Variances need judgment. A recount, a look at recent transactions, and a decision about the adjustment are hard to automate.
It is the first task to go. When volume spikes, counters become pickers.
Fit call
Send an email to
What changes that
Counting inside normal work
The cheapest step is usually already in your warehouse management system: zero-balance confirmations, remaining-quantity checks at pick, and count tasks interleaved with replenishment. Each visit to a location becomes a small check, without a separate counting shift.
RFID
RFID reads tagged items without line of sight, so a portal or handheld can capture many units at once. Its effect on accuracy often shows up at receiving. In a study run from June 2017 to July 2018 by the Auburn University RFID Lab and GS1 US, 69 percent of inbound orders contained errors when RFID was not implemented, while brands and retailers that used RFID to reconcile shipments were capable of achieving 99.9 percent order accuracy. The limits are practical: the cost of tagging every unit, read reliability around metal and liquids, and the need for tags to be applied at the source.
Vision and mobile scanning
Cameras on mobile robots, drones, or fixed mounts can read location and pallet labels on high racks, which suits pallet and case storage. Each-level bins are harder. Items are small, stacked, and hidden behind each other, so counting them still tends to need a hand in the bin.
Fix the cause, not just the count
A count corrects one record. Finding the cause fixes the process that broke it. WHO guidance for medical product storage sets a useful standard for any warehouse: periodic stock reconciliation should compare actual and recorded stock, and the root cause of discrepancies should be identified with corrective action to prevent recurrence. The usual causes:
Receipts accepted at the paperwork quantity without a count.
Putaway to a different location than the one confirmed.
Unit of measure mix-ups, such as a case recorded as an each.
Picks from a neighboring bin with a similar-looking item.
Damaged, sampled, or returned units moved without a transaction.
Tag each adjustment with a cause code. After a few weeks, the pattern usually points to one or two processes.
Site walkthrough
Send an email to
Common cycle counting mistakes
Showing counters the expected quantity. Blind counts are more accurate.
Counting a location while it is being picked.
Posting adjustments above tolerance without a recount.
Measuring accuracy only on A items.
Dropping the schedule at peak, when errors multiply fastest.
Where Avatar fits
Avatar robots pick from traditional racks and bins and can use the same scanner devices as your staff. That makes record accuracy in the pick zone as important for robots as it is for people, since a robot sent to an empty slot creates an exception. Learn more on the Avatar picking page, or see how counting fits the bigger picture in what inventory management covers.
How It Works
The process in four steps.
1
Set scope and tolerances
Classify items by value and velocity, decide to measure accuracy at the location level, and set a tolerance for each class, with zero tolerance for high-value or controlled items.
2
Build the cycle counting schedule
Set count frequency by class, add opportunity triggers such as zero-balance confirmations and replenishment counts, and schedule counts away from active picking.
3
Count blind and recount variances
Counters record what they find without seeing the expected quantity. Any variance above tolerance gets a recount and a transaction review before an adjustment is posted.
4
Find causes and track accuracy
Tag every adjustment with a cause code, trend inventory record accuracy by zone and shift, and fix the processes behind repeat errors.
Frequently Asked Questions
What is the difference between cycle counting and a physical inventory?
A physical inventory counts everything at once, usually with operations paused. Cycle counting counts a small portion of locations or items on a rolling schedule while the warehouse keeps running. Over a full cycle, every location is counted, and errors are caught and corrected much sooner.
How often should you cycle count?
It depends on the item class. High-value or fast-moving A items are usually counted most often, and slow C items least often. Add opportunity counts, such as when a location hits zero or is replenished, so counts happen when someone is already there.
What is a good inventory record accuracy rate for cycle counting?
There is no single right number for every operation. Set targets and tolerances by item class, measure at the location level rather than the SKU total, and watch the trend. A stable or improving rate with falling adjustment value matters more than one month's figure.
Can cycle counting replace an annual physical inventory?
In some cases. Auditing standards allow observation procedures to be performed during or after the period when well-kept perpetual records are checked periodically against physical counts. Whether that applies to your business is a decision for your finance team and auditors.
Sources
1. Public Company Accounting Oversight Board. AS 2510: Auditing Inventories, paragraph .10. PCAOB.
2. DeHoratius, N., & Raman, A. (2008). Inventory record inaccuracy: An empirical analysis. Management Science, 54(4), 627–641.
3. Auburn University RFID Lab & GS1 US (2018). EPC/RFID Data Exchange Study (Project Zipper): Executive Summary. GS1 US.
4. World Health Organization (2020). Good storage and distribution practices for medical products, WHO Technical Report Series No. 1025, Annex 7, sections 13.1–13.2. WHO.
Keep Reading
Core Concepts
What Is Inventory Management? A Practical Definition for Warehouse Teams
What is inventory management? A warehouse definition covering ABC, reorder points, FIFO vs. FEFO, record accuracy, and the metrics worth tracking.
5 min read
Read guide →
Core Concepts
WMS Explained
A WMS tracks inventory and directs warehouse tasks but never moves a unit. See what it controls, where control stops, and how it differs from ERP, WES, and WCS.
5 min read
Read guide →
Solution
Picking
See how Avatar works here.
Visit page →
Ready To Scale Your
Workforce?
Tell us your hardest jobs. We’ll show you what an Avatar robot can do on day one.
