An inventory cycle count is a count of a few products each day on a fixed rota. Over a year, every line is counted several times, and the shop never closes. It works because each error turns up within weeks, while the receipts that explain it are still on file.
This guide shows what to count, how many lines a day your rota needs and how to measure accuracy. The worked example is a pharmacy in Durban, in South African rand.
How is a cycle count different from a full stock take?
A stock take counts everything on one day; a cycle count spreads the same work across the year. The USAID Logistics Handbook puts it simply. Count a fraction of the items each month, and every item has been counted by the end of the year.
The same handbook gives the rule for when to count more. If the records don't match the shelf, count more often and fix the record keeping.
Cycle counts don't always replace the full count. If your accountant needs closing stock on one date, you still do a year-end inventory count, with fewer surprises. Stock taking vs stock checking compares the kinds of count.
Which cycle counting method should you use?
Use ABC counting to set the rota, and add zero counts to catch stock that exists only on paper. The other methods suit particular shops or a first trial.
| Method | How it works | Best for |
|---|---|---|
| ABC counting | Count A lines monthly, B lines quarterly and C lines twice a year | Most shops; rank the lines with the 80/20 rule |
| Random sample, constant population | Count the same number of random lines each time; a line can come up again | Products alike in value, such as a spare parts shop |
| Random sample, diminished population | Leave a counted line out until every line has had its turn | Making sure no line is skipped |
| Control group | Count a small group of lines many times over a few weeks | Training counters and testing the routine first |
| Location counting | Count one shelf or bay a day, in order | Shops that keep each product in a fixed place |
| Zero and low counts | Count a line when the records say it is out or nearly out | Catching phantom stock while there are few units to count |
Zero counts earn their place for one reason. A line that shows 3 on record and none on the shelf never triggers a low-stock alert, so nobody reorders it.
How many lines should you count each day?
Lines per day = (A lines × 12 + B lines × 4 + C lines × 2) ÷ counting days a year. The multipliers are the counts a year: monthly, quarterly and twice a year.
For example, a pharmacy in Durban carries 1,200 lines and trades 300 days a year, six days a week for 50 weeks. These are example figures, not a real pharmacy's.
| Class | Lines | Counts a year | Line counts a year | A week | A day |
|---|---|---|---|---|---|
| A, top 10% by sales value | 120 | 12 | 1,440 | 28.8 | 4.8 |
| B, next 20% | 240 | 4 | 960 | 19.2 | 3.2 |
| C, the other 70% | 840 | 2 | 1,680 | 33.6 | 5.6 |
| Total | 1,200 | 4,080 | 81.6 | 13.6 |
Rounded to whole lines, that is 5 A lines, 3 B lines and 6 C lines a day: 14 in all. At about two minutes a line, the count takes 28 minutes before opening.
Over the year, the rota takes about 136 hours, against about 40 hours for one full count of the same 1,200 lines. Cycle counting costs more hours, not fewer. What it buys is a shop that never closes, and an A line that is never more than a month from its last check.
Put theft-prone and fast-expiring lines in class A, whatever their sales value. A line that keeps coming up wrong moves up a class until it stops.
How to perform an inventory cycle count, step by step
A cycle count takes six steps, and the first one happens before anyone counts. Each step closes one way a count can mislead you.
1. Bring the records up to date
Enter today's deliveries, finish held sales and record customer returns before you count. A count against stale records finds differences that aren't there.
2. Count before trading, or pause the lines
Count before opening, or keep the till off those lines for the minutes the count takes. A sale during the count makes the record and the shelf disagree.
3. Count blind, in the selling unit
Count without seeing the recorded quantity, and count pieces if you sell by the piece. A counter who knows the expected number can stop when they reach it.
4. Compare, then recount what is outside tolerance
Compare each line with the record. Recount any line outside its tolerance, with a second person if you can, before you change anything.
5. Find the cause, then adjust
Check the line's movements since its last count: deliveries, sales, returns and transfers. Fix a missing entry at its source, and adjust only what stays unexplained, with a note.
The inventory discrepancy guide maps each kind of difference to its usual cause.
6. Log the result
Write down the date, the lines, who counted and the accuracy. The log shows which lines, shelves and counters keep going wrong, and it belongs in your inventory SOP.
How do you measure cycle count accuracy?
Measure two figures: the share of lines within tolerance, and unit accuracy on absolute differences. A net figure lets a surplus on one line hide a shortfall on another.
Line accuracy (%) = lines within tolerance ÷ lines counted × 100
Unit accuracy (%) = (1 − total absolute difference ÷ total recorded units) × 100
The unit formula is often printed as 1 − (difference ÷ counted) × 100, which gives a negative number as written. The 1 belongs inside the brackets: for 150 on record and 148 on the shelf, (1 − 2 ÷ 150) × 100 = 98.7%.
Set a tolerance for each class. For example, A lines must match exactly, B lines within 1 unit and C lines within 2 units. Treat that as a starting point, not an industry rule.
For example, here is one morning's count of 14 lines at the Durban pharmacy, with the differences at cost in rand.
| Line | Class | Records | Counted | Difference | Within tolerance? | Value (R) |
|---|---|---|---|---|---|---|
| Paracetamol 500 mg, 24 tablets | A | 150 | 148 | −2 | No | −60 |
| Antihistamine, 30 tablets | A | 36 | 42 | +6 | No | +510 |
| Vitamin C 1,000 mg | B | 40 | 39 | −1 | Yes | −120 |
| Hand sanitiser, 500 ml | C | 90 | 84 | −6 | No | −240 |
| Plasters, box of 20 | C | 75 | 77 | +2 | Yes | +50 |
| Nine other lines | A, B, C | 640 | 640 | 0 | Yes | 0 |
| Total | 1,031 | 1,030 | −1 | 11 of 14 | +140 |
- Line accuracy: 11 of 14 lines within tolerance, or 78.6%.
- Unit accuracy: the absolute differences add up to 17 units, so (1 − 17 ÷ 1,031) × 100 = 98.4%.
- The net figure: 1 unit short and R140 over, which looks almost perfect. The errors behind it are worth R980 at cost.
Two of the three misses are A lines, so both go to a recount the same day. The common rule of thumb of 90% accuracy is too loose for A lines. It would leave one best seller in ten with a figure you can't reorder from.
Track line accuracy each week, by shelf and by counter. Inventory accuracy belongs with your other inventory KPIs.
Common cycle counting problems, and the fix for each
The common problems sit in the schedule and the count sheet, not in the counting itself. Each problem below has one fix.
| Problem | Fix |
|---|---|
| The till sells a line while it is being counted | Count before opening, or pause those lines |
| Some lines never come up | Keep a list of lines with the date each was last counted |
| Two branches counted as one | Count each branch on its own sheet |
| The same lines are wrong every month | Trace the cause instead of adjusting again |
| Counters see the expected number | Hide the quantity column before printing the sheet |
| Cartons counted as pieces | Count in the unit you sell, and check the unit on the sheet |
How to run cycle counts in ArmPOS
ArmPOS makes a count sheet for one category at a time and corrects stock with a dated note. A daily rota takes six steps.
- Set the counting order. Fill each product's Internal Location (Rack/Shelf), and the Internal Location Report lists products shelf by shelf (Warehouses and locations).
- Make the day's sheet. Open Products › Count Stock and click Count. Choose the Date, the Warehouse and a Category, to count one group at a time (Count and adjust stock).
- Count blind. Download the Excel sheet from the file column and hide its Quantity column before you print it.
- Correct what stays unexplained. In Adjustment › Create Adjustment, enter the difference in Qty and choose Addition or Subtraction. Add a Note such as "Cycle count, week 12, pain relief". Qty is the difference, not your count.
- Keep losses apart. Broken or expired goods go through Damages › Create Damage, not an adjustment (Record damaged stock).
- Read the trend. Each week, the Stock Adjustment Report totals what you added and removed with its cost. The Negative Stock Report lists lines to count first (Stock reports).
In the adjustment, find each product by its code or with the scan button. ArmPOS reads barcodes, not RFID tags.
See what ArmPOS includes for pharmacies and other shops in South Africa.
Frequently asked questions
How often should you do a cycle count?
Count a few lines every trading day. Each A line then comes up monthly, each B line quarterly and each C line twice a year.
Does cycle counting replace the year-end stock take?
Only if your accountant or auditor accepts the cycle count records as evidence of closing stock. If not, you still count everything on one date, with fewer surprises.
What is a good inventory accuracy rate?
Every A line exact, and the rest within the tolerance you set for their class. Track your own figure count by count, because the trend tells you more than any published benchmark.
What are the benefits of cycle counting?
Errors turn up within weeks, while their paperwork is still on file, and the shop never closes for a count. Repeated shortfalls on one line show theft early, and reorder alerts become reliable because the A lines are right.
What is a blind count?
A count made without seeing the quantity the records show. The counter writes down what is on the shelf, and the comparison comes after.
What is the difference between a cycle count and a stock check?
A stock check is any partial count, often of lines chosen on the day. A cycle count is a stock check on a fixed rota that reaches every line within a set period.
Work out your lines per day with the formula above, and count the first 14 before opening tomorrow. To print each day's sheet by category and correct differences with a note, create a free ArmPOS account.
