Cycle Counting Is the Audit Process Most Warehouses Get Wrong#
Cycle counting is the practice of counting small subsets of inventory continuously throughout the year. Done well, it keeps inventory accuracy above 98% without shutting the warehouse down. Done badly — or not at all — it leaves operations relying on annual full stock-takes that produce one giant adjustment journal and obscure the operational signals that caused the variance.
This article covers the canonical cycle-counting practice: ABC classification, count cadence, variance investigation, and how to set the discipline up against any inventory system.
The Definition#
Cycle counting is a rolling inventory audit process where small subsets of SKUs are counted on a recurring schedule, with the goal of catching variances early enough to fix the root cause. The whole warehouse never gets counted at once; instead, every SKU gets counted at least once per year, with high-velocity or high-value SKUs counted more often.
The defining test: cycle counting catches a discrepancy within days of it appearing, not months later at the annual count.
Why Cycle Counting Beats the Annual Stock-Take#
The annual stock-take has three operational problems:
It shuts the warehouse down. Most annual stock-takes require closing dispatch for one to three days. For a 5,000-order/month operation, that is 200–500 dispatches not happening on those days — revenue loss plus customer-experience risk.
It catches variances months after they happened. If a receiving error in March creates a variance, the annual stock-take in December catches it. By then, the root cause — the receiving process, the supplier, the shift, the person — is dead-trail. The variance gets adjusted; the underlying problem persists.
It produces one massive adjustment journal. Year-end inventory adjustments of NZ$100,000+ are common. Finance has to write the journal; auditors have to investigate; the P&L takes a hit. None of which would have happened if the variances had been caught in real time.
Cycle counting solves all three: no shutdown, variances caught within days, adjustments small and continuous.
The ABC Cycle Count Method#
The most common cycle-counting structure is ABC, which classifies SKUs by velocity or value:
A items — top 20% by velocity or value. Counted every month (12× per year). High-impact SKUs get the most attention.
B items — middle 30%. Counted every quarter (4× per year). Medium-impact.
C items — bottom 50%. Counted once or twice per year. Low-impact SKUs get the least attention.
The principle: high-velocity and high-value SKUs deserve more attention because their variances cost more. C items can wait; if a slow-moving C item has miscounted by 2 units, it does not move the P&L.
Some operators use ABCD (adding a D tier for very slow movers counted once a year) or sub-divide A items into AA, A1, A2 for very high-velocity SKUs. The principle is the same: count effort scales with item importance.
The classification can be by velocity (units sold per period), by value (cost × velocity), or by Pareto-style impact (top contributors to revenue or COGS). Velocity-based classification is the most common and simplest to maintain.
How a Cycle Count Actually Runs#
The operational flow:
- Count list generation. A modern WMS generates the daily count list automatically based on the ABC cadence. For an operation with 5,000 SKUs (1,000 A-items, 1,500 B-items, 2,500 C-items), the daily count list might be 40–60 SKUs: roughly the daily count needed to hit the monthly/quarterly/annual cadence.
- Physical count on the floor. A picker (or dedicated cycle-count team) walks to each bin and counts what is physically there. Scanner-driven counts (Bluetooth or wired) reduce errors; manual paper counts work but introduce data-entry risk.
- System comparison. The counted quantity is compared to system on-hand. Any variance above a threshold (typically 2% or a fixed unit count) is flagged for investigation.
- Variance investigation. Was it a receiving error? A picking error? A putaway error? A theft pattern? A system bug? Root cause matters more than the adjustment journal.
- Adjustment + audit trail. Variances above tolerance get adjusted with audit trail. Variances below tolerance get logged but adjusted automatically.
- Trend analysis. Monthly reviews surface SKUs with rising variance rates, zones with systemic issues, or shifts with higher error rates.
What "Good" Cycle Counting Looks Like#
The benchmarks for a mature cycle-counting practice:
- Inventory accuracy above 98% across all SKUs, above 99% on A items.
- A items reach 100% count completion every month.
- Variances trend down over time as root causes get fixed.
- Finance signs off on the rolling-adjustment process and skips the annual full count.
- The cycle-count team's investigation closure rate is above 90% within 7 days.
Most operations starting cycle counting come from 85–93% inventory accuracy with annual stock-takes. The discipline consistently pushes them above 98% within 90 days of setup. The hardest part is not the counting — it is the variance investigation discipline.
Cycle Counting Without a Dedicated WMS#
The discipline is system-agnostic. It works alongside whatever inventory system the business runs:
On Xero or MYOB native inventory. The count list can be a spreadsheet derived from item velocity. Counts are entered manually; variances are calculated against Xero / MYOB on-hand quantity. Adjustments flow back to the accounting system.
On Cin7 Core, Unleashed, or DEAR. These platforms support count workflows natively, though the cadence is usually manual rather than ABC-automated.
On a modular ERP with WMS (NetSuite Warehouse, OpsUI Warehouse, Dynamics 365 Supply Chain). Full ABC cadence, scanner-driven counts, automated variance flagging, and finance-signoff workflow are built in.
On a dedicated WMS (Manhattan Active, Blue Yonder, SoftEon, Mintsoft, Logiwa). All of the above plus task interleaving — cycle counts run between picking tasks to maximise picker utilisation.
The depth of automation matters less than the discipline. A weekly manual count on a spreadsheet is materially better than no count at all; a fully automated WMS workflow is materially better than the spreadsheet.
Variance Investigation — The Hard Part#
Counting is easy. Investigating variances is hard. Most cycle-counting practices fail at the investigation step.
The investigation questions:
- Receiving error? Wrong quantity recorded at intake, wrong SKU assignment, or wrong putaway location.
- Picking error? Wrong SKU picked, wrong quantity picked, or item picked but not recorded.
- Putaway error? Item moved to a different bin than the system recorded.
- Returns error? Returned item not recorded back into stock, or recorded wrongly.
- Theft pattern? Variances clustered on specific SKUs, shifts, or staff.
- System bug? Sync error between WMS and ERP, double-counting at a transaction boundary, or unit-of-measure confusion.
Each root cause has a different fix. Receiving errors need supplier or process fixes; picking errors need training or workflow fixes; theft needs HR action; system bugs need vendor escalation. The variance is a symptom; the fix is upstream.
Operations that investigate variances closure rates above 90% within 7 days consistently push accuracy above 98%. Operations that just adjust the ledger without investigating consistently drift.
When Cycle Counting Replaces the Annual Stock-Take Entirely#
Most external auditors accept a documented cycle-counting program as a substitute for the annual full stock-take, provided:
- Every SKU is counted at least once per year
- Variances are documented with audit trail
- Adjustment journals are reviewed and signed off
- Accuracy benchmarks are met (typically 98%+ rolling)
This is a finance and audit conversation, not a software conversation. Cycle counting is a documented process; the auditor needs to see the discipline, not the WMS dashboard.
For deeper coverage of warehouse architecture, see What is a WMS? and Inventory Management Module Architecture.