3.5Inventory accuracy and cycle counting
The essentials in 30 seconds
- Every calculation assumes accurate stock: below 90% accuracy, everything works on sand.
- Cycle > annual for managing: small portions continuously, ABC cycle, errors near their cause.
- Gap = figure + cause: fixing the number is never enough, the process must be treated.
Running mini-case — Nadia discovers her system lies
Nadia draws 25 items at random: 4 are wrong — receipt counted from the delivery note instead of reality, undeclared breakage. Her system triggers too late and orders for nothing. She launches ABC-weighted weekly cycle counting, investigates each gap within 48 hours and labels every location. In one quarter, her 3.3 thresholds become accurate again.
Final episode of cases 3.1-3.4 — Nadia secures accuracy before optimising.
1. Stock accuracy: the silent assumption
All your calculations (reorder point, safety, turnover) assume system stock equals physical stock. Yet receiving errors (counted 100, received 98), picking errors (taken 12, booked 10), undeclared breakage, theft, unrecorded movements and confused units silently accumulate gaps.
Consequences: surprise stockouts (system thought some remained), useless orders (thought none remained), undeliverable promises. Accuracy (IRA, Inventory Record Accuracy) is the share of correct SKUs (within tolerance); target 95-98% — below 90%, your calculations work on sand.
Below 90%, improving accuracy comes before any sophisticated optimisation: the best model on false data produces false decisions.
2. Annual vs cycle counting
The full annual count — counting everything, operations stopped — mainly meets an accounting requirement (balance-sheet value to verify; duties vary by country and status, see accountant). As a management tool it is poor: one snapshot per year, taken in a rush, causes undetectable twelve months later.
Cycle counting — counting continuously in small portions, ABC-weighted cycle (A several times/year, C once) — is far superior: no shutdown, errors caught close to their cause (this week's gap comes from a recent, explainable and fixable error).
The golden rule: do not just fix the figure — find the process cause (receiving? picking? entry?) and treat it, otherwise the gap returns forever.
3. Prevent at source
Add prevention: dedicated, labelled locations (disorder is the father of gaps), every movement recorded in real time (breakage too, samples too, loans too), limited stock access (a store open to all is never accurate).
When volume justifies it, scanning of movements (see 4.4): barcodes remove most entry errors and keep stock reflected in real time.
Reliable stock is no large-company luxury: it is the prerequisite for everything else. Fund it as such.
🔎 Concrete example — The miscounted carton that skews everything
Receipt of 100 units quickly counted as 100, booked as 100 — actually 98 received. The system triggers too late, surprise stockout. Weekly cycle counting on this family would have caught the 2 % gap that same week.
Fix: recount lots at receiving and book the quantity actually received, not the delivery-note one.
Prevention: scan at receiving and dedicated locations: the gap falls below 1 %.
Section vocabulary
- Inventory accuracy (IRA)
- Percentage of items where physical matches system (within tolerance); target 95-98% — below 90%, all calculations are wrong.
Simple example: Nadia counts 25 items: 21 accurate, i.e. 84% — below the bar, accuracy first.
What it is not: optimising thresholds while the system lies one item in six.
Reflex: test IRA on 20-30 random items before any tuning.
- Full annual count
- Periodic full count (often at year-end), mainly for accounting purposes; duties vary by country.
Simple example: Nadia keeps the annual count for the balance sheet and validates duties with her accountant.
What it is not: expecting fine month-by-month gap steering from it.
Reflex: separate the accounting duty from the daily management tool.
- Cycle counting
- Continuous counting in small portions (A several times/year, C once); detects errors close to their cause, without stopping operations.
Simple example: Nadia counts a few items each week, A items first, without stopping the warehouse.
What it is not: recounting everything once a year in a rush without seeking causes.
Reflex: schedule an ABC-weighted weekly cycle in a simple spreadsheet.
- Inventory gap
- Difference between counted and system stock; treated in two steps: fix the figure AND find the process cause.
Simple example: Nadia fixes 98 instead of 100 then traces back to receipt counted from the delivery note.
What it is not: fixing the figure and filing away without investigating within 48 hours.
Reflex: for each significant gap: figure fixed + cause noted + process action.
- Unknown shrinkage
- Unexplained gaps (theft, undeclared breakage); reduced through organisation: dedicated locations, recorded movements, limited access.
Simple example: Nadia logs breakage and samples and restricts access to the store.
What it is not: leaving the store open to all with unlogged movements.
Reflex: label each location and log every movement, even small.
In practice — Measure and secure your stock accuracy
- Test your accuracy: draw 20-30 SKUs at random, count physically, compare to system — calculate your IRA.
- Set up simple cycle counting: a few SKUs each week, ABC-weighted cycle (A several times/year).
- For each gap: fix the figure then trace the cause (receiving? picking? undeclared breakage? entry?) and fix the process.
- Strengthen prevention: labelled locations, every movement recorded (breakage and samples included), limited access.
Ready-to-say script — launching cycle counting without stopping work (1 min)
"Our test shows 84% accuracy: our thresholds work on sand. I propose simple cycle counting — a few items each week, A items first — without stopping work. Each gap: we fix the figure and seek the cause within 48 hours."
IRA figure + gentle method + 48-hour rule: cycle counting lands without blocking.
Copy-paste template — cycle / gap card
WEEK OF: _______________
ITEMS COUNTED (ABC): _______________
OBSERVED IRA: _______________
GAP (item + quantity): _______________
PROCESS CAUSE (receipt / picking / entry / breakage): _______________
PREVENTION ACTION: _______________VSE / small budget
Start with no software: 5 items per week in a spreadsheet, investigation within 48 hours on each gap.
Algeria SME
Count what was actually received, not the delivery note; labelled locations even in a back store to hold accuracy.
Agri-food
Couple cycle counting with dates: check quantities AND expiries to avoid dormants and undeclared breakage.
Scan-justified volume
Scan receipt and picking to kill entry errors and reflect stock in real time.
✓ Best practices
- Start cycle counting even with no software: a spreadsheet and a weekly schedule are enough.
- Investigate each significant gap within 48 h: afterwards, the cause fades.
- Limit stock access: a store open to all is never accurate.
✗ Common pitfalls
- Settling for the annual count: you fix month-old symptoms without understanding causes.
- Fixing the figure without treating the cause: the gap rebuilds the very next week.
- Not recording small movements (breakage, samples, loans): their sum ruins accuracy.
Key takeaways
- All your calculations assume ACCURATE system stock: accuracy (IRA ≥ 95 %) comes before any optimisation.
- Annual = mainly accounting requirement (poor snapshot); cycle = management tool (errors close to cause, no shutdown, ABC cycle).
- A gap is treated in two steps: fix the figure AND treat the process cause — otherwise it returns.
- Prevention: dedicated, labelled locations, all movements recorded in real time, limited access, scan when volume justifies it.
48h Challenge — apply tomorrow
Like Nadia: draw 20 items at random, count them and compute your IRA. Launch cycle counting: 5 items per week, 1 gap investigated within 48 hours.
- I can define IRA and its target (95-98%) without hesitation.
- I measured my IRA on 20-30 items.
- I scheduled ABC cycle counting with investigation within 48 hours.
Frequently asked questions
Why is my stock always wrong despite counts?
Because counting fixes symptoms (figures) without touching causes (process): receipt counted from the delivery note instead of reality, mis-booked picking, ghost movements (breakage, samples, loans), unit mix-ups and disorder (unfindable item counted as zero). Cycle counting detects close to the cause and enables the mini-investigation that identifies the faulty process — run over a quarter, the Pareto of causes points to the first project.
Cycle or annual: which to choose?
Both, but for different reasons. Annual meets the accounting duty (balance-sheet valuation); cycle steers management (continuous accuracy, no shutdown, ABC-weighted). If you had to keep only one for performance, keep cycle: it is what makes replenishment maths right.
What is the stock reliability indicator and its target?