Cycle Count: Inventory Record Accuracy, Count Frequency by ABC Tier, and the IRA Formula
Compare line accuracy with absolute quantity variance and keep physical counts distinct from receiving checks.
Jainul Vaghasia/Published /Updated /8 min read
Use the definition
Turn procurement terms into an operating system.
This reference page should help you understand the concept first. When the term affects purchasing execution, LineNow connects it to live POs, supplier replies, receiving, and accounting handoff.
Cycle counting is an inventory management practice in which a subset of SKUs is physically counted on a rotating schedule — rather than halting operations for a wall-to-wall physical inventory count. The goal is to maintain Inventory Record Accuracy (IRA) continuously, so the quantities in the system match the quantities on the shelf at any given moment. In a closed-loop procurement system — one where inventory alerts, purchase orders, supplier replies, receiving, and accounting handoff run in one connected record — accurate inventory records are load-bearing: bad count data corrupts every downstream replenishment calculation.
Quick answers
What is a cycle count? A partial physical count of a rotating subset of SKUs, performed during normal business hours without halting operations. The counts work through the full catalog over a defined period, so every item is verified at an interval appropriate to its value and velocity.
What is Inventory Record Accuracy (IRA)? The percentage of counted units whose system record matches the physical count within an acceptable tolerance. Formula: IRA = (1 − |Σ variance units| / Σ book units for counted items) × 100. An IRA of 97% means approximately 3 in every 100 counted units showed a discrepancy.
How does cycle counting differ from a full physical count? A full physical count covers every SKU in every location simultaneously — typically requiring a shutdown of receiving, shipping, and sometimes sales. Cycle counting covers a fraction of SKUs per day or week in rotation, so operations continue uninterrupted and accuracy is maintained throughout the year rather than recovered once annually.
Why do A-items get counted more often than C-items? Because the cost of an IRA error scales with annual usage value (AUV). An A-item discrepancy that goes undetected for three months can corrupt a reorder point, trigger a false-confident system count that hides an actual stockout, or cause an unnecessary emergency order. A C-item with a ten-unit ghost-stock error has minimal operational or financial consequence.
Read before ordering
A dense operator briefing for teams that need sharper buying, cleaner supplier follow-up, and fewer expensive surprises.
The IRA formula
The standard unit-level formula:
IRA = (1 − Σ|variance| / Σ book units for counted items) × 100
Where:
Σ|variance| = sum of the absolute unit discrepancy on each counted line (|system quantity − physical count|)
Σ book units for counted items = total units counted across all lines in the cycle
A simpler line-level variant:
Line IRA = (lines with zero variance / total lines counted) × 100
Unit-level IRA is more sensitive: one line with a ten-unit error matters more than five lines with one-unit errors. Line-level IRA can mask systemic problems in high-volume SKUs by treating a ten-unit gap identically to a one-unit gap.
Worked example
A specialty grocer cycle-counts 50 SKUs in the produce section:
SKU
System qty
Physical qty
|Variance|
Roma tomato (case)
12
10
2
Baby spinach (bag)
24
24
0
Lemon (each)
48
44
4
Jalapeño (lb)
30
27
3
(46 other SKUs)
506
509
2
Totals
620
614
11
IRA = (1 − 11 / 620) × 100 = 98.2%
All 11 units of variance came from 4 SKUs. The remaining 46 SKUs matched exactly — but unit-level IRA is still below 99% because the four exceptions are unit-weighted rather than line-weighted. On a 50-line count, this would show as 92% line IRA (4 lines with variance / 50 total) or 98.2% unit IRA. Neither measure is inherently more conservative: here the unit score is higher. Line accuracy shows how widely errors occur; the absolute-unit variance score weights their quantity. Report both and also inspect value and critical-item exceptions.
Count frequency by ABC tier
ABC classification — ranking every SKU by annual usage value (AUV) — drives the count schedule directly. Count what you can least afford to get wrong, most often.
ABC class
Recommended count frequency
Annual count rotations
Rationale
A
Weekly or monthly
12–52×
High AUV; a stockout or ghost-stock unit at this tier carries direct revenue and replenishment impact
B
Quarterly
4×
Moderate AUV; quarterly verification keeps IRA within tolerance without over-investing count labor
C
Semi-annually
2×
Low AUV; even a multi-week IRA error is inexpensive relative to the count labor cost
A practical schedule for a business with 200 active SKUs (40 A / 60 B / 100 C):
A-items: count 8 SKUs per week → full A-tier rotation in 5 weeks, 10+ rotations per year
B-items: count 5 SKUs per week → full B-tier rotation in 12 weeks, 4–5 rotations per year
C-items: count 4 SKUs per week → full C-tier rotation in 25 weeks, 2 rotations per year
Combined, this is 17 SKUs per week — about 3 per workday for a team that spends the last 20 minutes of the day counting. No shutdown. No disruption. Continuous accuracy.
Exception trigger: Any SKU that fails a count beyond the tolerance band (see next section) should be re-counted immediately and investigated before the next scheduled rotation. A pattern of repeated exceptions in the same location or SKU signals a systemic problem — receiving short-falls, pilferage, unit-of-measure confusion, or a supplier pack-size change not captured in the item record — rather than normal statistical variance.
Set count tolerances explicitly
Define exact-match or tolerance-based line accuracy, plus a separate absolute unit or value variance. Set review thresholds by criticality and financial impact. A portfolio average can hide a serious error in one important item, so retain the individual count exceptions.
Cycle count vs. full physical count
Cycle count
Full physical count
Scope
Subset of SKUs, rotating
All SKUs, all locations, simultaneously
Frequency
Continuous (daily or weekly)
Annual or semi-annual
Operational disruption
None — runs during normal hours
Requires partial or full shutdown
IRA trajectory
Continuous small corrections; steady-state accuracy
Large catch-up, then slow drift until the next count
Best for
Maintaining steady-state IRA throughout the year
Establishing an accuracy baseline from scratch; pre-audit reconciliation
Most businesses that implement cycle counting still run an annual full physical count as a reconciliation baseline. The practical difference is that by the time the annual count runs, 95%+ IRA means the count is a verification exercise rather than a discovery exercise. The number of exceptions is small, and the corrections are minor.
Why IRA is load-bearing for procurement
Every replenishment formula takes inventory quantity as an input. When that quantity is wrong, the formula produces a wrong answer — and the operator has no signal that anything is amiss.
Reorder point.ROP = (consumption rate × lead time) + safety stock. If on-hand is inflated by ghost stock, the system believes the reorder point has not been reached when the actual shelf quantity is below the trigger. The order fires late, or not at all.
Safety stock. Safety stock is sized to cover demand variability during lead time. If the system-reported on-hand overstates actual stock by 10 units, the effective buffer is 10 units smaller than the model assumes. The margin against a stockout is thinner than the math shows.
Days of inventory on hand.DOH = on-hand / consumption rate. If on-hand is overstated by 15%, DOH says "11 days of supply" when the shelf has 9 days. The stockout window opens silently.
Inventory turnover. Ghost stock inflates the average inventory denominator (turns = COGS / average inventory at cost), suppressing the calculated turnover ratio and making a well-managed catalog appear over-stocked. Decisions made against that metric — reducing safety stock, consolidating SKUs, slowing replenishment — are made against a false number.
GMROI. Ghost stock in the denominator depresses GMROI, potentially misidentifying productive SKUs as poor performers when the real problem is a count error.
This is the garbage-in, garbage-out problem for procurement: the most sophisticated forecasting methods cannot overcome a starting quantity that is wrong. IRA is the quality gate that makes every downstream formula defensible.
The receiving-count connection
Most cycle count programs treat counting and receiving as separate activities. Structurally they share the same physical action: verifying that the quantity of a specific item in a specific location matches a document.
At the moment a purchase order is received, the receiving record captures:
What the PO said should arrive
What the supplier confirmed would arrive (from the supplier-reply-updated PO state)
What physically arrived at the dock or storeroom
The difference between items 2 and 3 is a receiving variance. Resolving it immediately — capturing the actual received quantity before posting — updates inventory with the physically correct number. This is structurally identical to a cycle count on the items covered by that PO.
Accurate receiving improves the transaction record, but counting an incoming delivery is not a count of all existing stock. A weekly receipt cannot verify what remains on the shelf or detect losses since the last count. Keep physical cycle counts and receiving checks as complementary controls.
Three-way matching — verifying the supplier invoice against the purchase order and the goods receipt note — requires a real receiving count as input. A receiving record without an actual count is not a GRN; it is a signature confirming that goods arrived without verifying what arrived. The quantity match between the invoice and the counted receipt is what makes three-way matching a genuine control rather than a paper exercise.
Apply this to a real purchasing record
LineNow's purchasing workflow connects purchase orders, supplier replies, receiving and accounting handoff. In a demonstration, inspect a partial receipt, a later physical count and an unexplained variance; a receipt count does not verify every unit already on the shelf.
Use the result to agree the fields, decision owner and exception process. A linked purchasing record supplies evidence for this analysis; it does not by itself prove a particular dashboard, financial outcome or automatic approval policy.
