A practical starting schedule is to cycle count A items monthly, B items quarterly, and stable C items every six to twelve months. Count specific high-risk items or locations more frequently, and investigate unexpected shortages as they occur. These intervals are a framework, with the final schedule shaped by your products, warehouse activity, and count results.
A warehouse might perform cycle counts every day while checking each individual SKU only a few times a year. Daily counting describes the team’s routine. Item frequency describes how long a particular inventory record goes without physical verification. A good program makes both decisions explicit. It identifies which stock needs attention, sets achievable due dates, and assigns enough labor to complete the work. That is where a counting schedule starts protecting orders rather than simply filling a calendar.
How often should you cycle count A, B, and C inventory?
You can use the following cycle counting schedule as an initial planning guide. Kardex publishes similar sample intervals and emphasizes adjusting them to the warehouse’s risk and results.[1]
| Inventory group | Starting frequency | Reason to count sooner |
|---|---|---|
| A items with high financial or service exposure | Monthly | Frequent handling, costly shortages, or recurring discrepancies |
| B items with moderate exposure | Quarterly | Rising activity, new customer commitments, or repeated errors |
| C items with low exposure and stable records | Every six to twelve months | An error starts affecting fulfillment or the item’s risk changes |
| Specific items or locations with repeat discrepancies | Weekly during investigation | The underlying process is still producing errors |
| Unexpected short picks or low or empty locations | At the event under your exception procedure | Waiting for the next scheduled count could delay orders |
ABC analysis needs a defined basis. One conventional approach ranks inventory by annual consumption value: annual units used or shipped multiplied by unit cost. NetSuite documents that calculation for its inventory segmentation process.[2] A high unit price alone does not establish how much attention an item needs.
Add operational judgment to the classification. A low-cost component required for every subscription box can create more disruption than a costly product that rarely ships. Both value and the consequences of an incorrect record belong in the decision.
For a straightforward baseline, plan to verify every in-scope inventory record at least annually, with shorter intervals for higher-risk stock. Treat that as your coverage policy, rather than a universal legal requirement. Include decisions about inactive inventory, returns, damaged goods, and recorded-empty locations so stock does not disappear from the program.
Which factors should change your counting frequency?
Two SKUs in the same ABC class can need different schedules. Start with the class interval, then check how each product moves and what happens when its record is wrong.
How much handling does the stock receive?
Look beyond sales units. Receipts, putaway, replenishment, transfers, picks, and returns all create points where physical stock and system records can diverge. An item moved between several bins deserves different scrutiny from inventory sitting untouched in controlled storage.
The same SKU can also need different treatment at different facilities. A busy forward-pick location and a slow reserve location do not necessarily carry the same exposure.
Does the item have a history of discrepancies?
Repeated shortages, unexplained gains, and incorrect locations are reasons to shorten the interval while investigating. Review the transactions between counts to locate the failure. The ASCM Wisconsin Chapter describes control-group counting as a way to test whether process corrections hold after stock moves.[3]
Counting more frequently can reveal the pattern sooner. Correcting receiving, scanning, replenishment, or return procedures is what prevents the pattern from continuing.
What would an error prevent you from shipping?
Consider the order, not just the missing item. One inexpensive accessory may stop a complete kit from shipping. A product with a long supplier lead time may be difficult to replace after a discrepancy surfaces. Customer commitments and replacement options can justify a shorter interval even when the stock has modest value.
Has demand or the operation changed?
Review the schedule before peak trading, after a major promotion, and when adding a warehouse or changing systems. A quiet-season count history may say little about how inventory performs when temporary staff, more returns, and heavier replenishment enter the picture.
How many inventory records should you cycle count each day?
Calculate the workload from what must actually be verified. One SKU may occupy several bins and warehouses, with additional records for lots or stock status. Oracle supports both item-based and location-based counting, so the counting unit needs to be clear before you plan labor.[4]
For example, your system shows 20 units in a pick bin and 80 in reserve. The physical quantities are zero and 100. You own the correct total quantity, but the picker’s location is wrong. A warehouse-wide SKU total would hide that problem.
Define a count target as the item, location, and other attributes your procedure verifies. Then calculate the annual work for each group.
Annual count events = count targets in each group × annual frequency, added across all groups.
Daily planned count events = total annual count events ÷ available counting days.
Suppose 2,000 unique SKUs create 5,000 separately verified item-location targets. This example assumes one item record per target, without further lot or serial-number divisions.
| Class | Unique SKUs | Count targets | Counts per target per year | Annual count events |
|---|---|---|---|---|
| A | 300 | 1,000 | 12 | 12,000 |
| B | 600 | 1,500 | 4 | 6,000 |
| C | 1,100 | 2,500 | 1 | 2,500 |
| Total | 2,000 | 5,000 | Mixed | 20,500 |
With 250 available counting days, the program requires 82 count events per day. At an assumed seven minutes per event, that is about 9.6 labor hours per day. A 20 percent planning allowance increases the estimate to approximately 11.5 hours.
The quantities, seven-minute duration, and allowance are illustrative assumptions, not industry benchmarks or Speed Commerce operating data. Measure travel, physical counting, entry, recounts, and investigation time in your own operation before assigning staff.
A SKU-only calculation produces a different answer: 300 × 12, plus 600 × 4, plus 1,100 × 1 equals 7,100 annual events, or 28.4 per day. In this example, ignoring multiple locations understates the planned work by a factor of approximately 2.9.
Also distinguish count events from WMS tasks. A single task might contain several item lines. Use the same unit when measuring time and estimating capacity. Oracle’s scheduling documentation combines class population, annual frequency, and working periods, which provides a useful basis for this planning approach.[5]
Finally, 82 per day is an average, not a complete schedule. Each target needs a last-count date, a next due date, and a completion record. Otherwise, a team can hit its daily activity goal while repeatedly overlooking the same inventory.
When should you count inventory outside the schedule?
An unexpected shortage should trigger attention before the next monthly or quarterly count. Microsoft documents cycle count plans alongside quantity thresholds that generate count work when stock falls below a defined level, including empty-location triggers.[6]
Useful exception events include a picker reporting insufficient stock, a bin unexpectedly reaching zero, an unexplained negative balance, or a transfer that does not reconcile. Check the affected location and related stock, then trace the relevant transactions before deciding whether an adjustment is needed.
Schedule targeted verification before a major sales period or after a substantial stock move when the risk warrants it. Leave time to resolve discrepancies before the next wave of orders.
An exception count should receive credit only for the inventory it actually verifies. Counting one pick bin does not establish that reserve stock or another warehouse is correct. Similarly, receiving verification and return inspection should not automatically satisfy a scheduled cycle count with a broader scope.
How should you count while orders keep moving?
Cycle counting can spread work through normal operations, but movement still needs control. Pause relevant transactions in the count area or use a documented process that reconciles movement against the count time. An inventory snapshot taken before a pick cannot be compared blindly with a quantity counted afterward.
Use blind counts so the counter does not see the expected quantity. Confirm the item, location, and unit of measure before recording the result. Recording 12 cases as 12 individual units produces a different balance from recording the units inside those cases. Microsoft’s warehouse documentation describes blind counting and prompts for different counting units.[7]
When a variance appears, recount it, verify recent movements, and identify the cause. Keep the original result, any recount, the approved correction, and the reason code. That history helps distinguish a counting mistake from a recurring inventory problem.
How do you know whether your counting schedule is working?
Review coverage alongside accuracy. Clean results from easy-to-count bins do not compensate for overdue high-risk locations. A monthly review is a practical management starting point, with an earlier review when fulfillment problems appear.
| Measure | What to track | What it tells you |
|---|---|---|
| Count completion | Scheduled targets completed by their due dates ÷ targets due | Whether the promised frequency is being delivered |
| Overdue coverage | Number and age of overdue targets by class and warehouse | Which stock is going too long without verification |
| Confirmed exact-match rate | Verified records matching the defined system record before adjustment ÷ verified records | How accurate the records checked were |
| Absolute variance value | Absolute quantity difference for each record × stated unit cost, summed across records | Financial exposure without gains cancelling out losses |
| Repeat discrepancies | Records with recurring confirmed errors over a stated period | Whether process corrections are holding |
| Resolution time | Time from confirmed discrepancy to approved correction and documented action | How long identified problems stay unresolved |
Define what a match means. Exact quantity, correct location, and correct lot or status may all mean something. If you allow a tolerance, report within-tolerance results separately from exact matches. Measuring accuracy after correcting every balance conceals the errors the program found.
An exact-record match rate also differs from a unit-weighted or dollar-weighted measure. Two warehouses can report the same percentage while measuring different things. Ask for the denominator and the matching rules before comparing results.
Use confirmed repeat errors, service failures, and changing activity to shorten intervals. Consider lengthening an interval after the process has been corrected and accurate results persist through representative activity. For example, you might require three consecutive clean follow-up checks before reviewing a temporary weekly schedule. That is a possible local rule, not an industry standard.
Risk-selected counts describe the records selected. They do not automatically provide an unbiased estimate of accuracy across the entire warehouse. Keep results separated by warehouse, risk group, and count type so the interpretation stays useful.
Can cycle counting replace a full physical inventory?
Sometimes, but a counting calendar alone is insufficient. Reliable records, completed counts, documented variances, and effective controls impact that decision.
PCAOB AS 2510 recognizes circumstances where well-kept perpetual records and reliable counting methods can make an annual physical count of every item unnecessary. Its requirements apply in the relevant audit context; finance and the applicable auditor should determine what evidence your business needs.[8]
When records are broadly unreliable, a full physical count may help establish a dependable baseline before ongoing counts take over. NetSuite discusses that approach when records or counting procedures are weak.[9] There is no universal accuracy percentage that decides the issue for every warehouse.
What should you expect from a 3PL cycle counting program?
Ask a provider to explain what “daily cycle counting” means for your inventory. The answer should identify your stock’s scope and due dates, rather than only describing how much counting occurs somewhere in the building.
Request a documented schedule covering the relevant warehouses, pick and reserve locations, and required lot or status records. Ask how the provider tracks completed coverage and overdue counts. A site-wide accuracy figure gives limited insight into whether your products have been checked.
Agree on exception triggers, recount procedures, and adjustment authority. You should know who investigates a confirmed discrepancy, what requires approval, and what information reaches your team. Ask for reporting that explains the change instead of presenting a corrected balance without context.
Clarify pricing as well. Scheduled counts, additional requested counts, recounts, and full physical inventories may receive different treatment under the agreement. Resolve those details when setting the service scope.
A useful provider discussion connects the schedule to your product mix and order activity. Fast-moving variants, customer-specific kits, and inventory spread across facilities may need different controls from stable reserve stock.
How can Speed Commerce help you manage inventory?
A growing catalog creates more inventory records, more movement, and more opportunities for discrepancies to affect customers. A workable counting program connects physical verification with receiving, storage, replenishment, and order fulfillment.
Speed Commerce supports ecommerce and B2B fulfillment for brands managing those demands. Explore our inventory management guide, learn more about inventory tracking, or review the broader steps in an inventory audit.
Talk with Speed Commerce about inventory visibility, counting coverage, and discrepancy handling for your product mix and order volume.
References
1. Kardex. Cycle Counting in the Warehouse: Methods, Frequency, and Best Practices. Sample frequency table.
2. Oracle NetSuite Applications Suite. Inventory Optimization Preferences. ABC Category Preferences.
3. ASCM Wisconsin Chapter. Fred Kindelberger. Quality LEAN Cycle Counting: Accuracy That Lasts. December 10, 2021.
4. Oracle Fusion Cloud SCM. Redwood: Perform Cycle Counts Using Locators. Release 25D.
5. Oracle Fusion Cloud SCM. Cycle Count Creation. Define Schedules and Approvals. Release 25D.
6. Microsoft Learn. Define cycle counting. Counting thresholds and cycle count plans.
7. Microsoft Learn. Cycle counting. Counting work, blind counts, and units of measure.
8. Public Company Accounting Oversight Board. AS 2510: Auditing Inventories. Paragraphs .10 and .11.
9. NetSuite. Inventory Cycle Counting 101: Best Practices and Benefits. Physical inventory and baseline counting.
