Every year, the same scene. You close the shop on a Saturday, bring in the whole team, count until late, and on Monday you discover a gap of tens of thousands of dirhams between theoretical and actual stock. Nobody knows when the gap appeared, or why. You correct it, move on, and repeat the following year.
Cycle counting solves this. Instead of counting everything once a year, you count a small portion of your stock every week, without ever interrupting trading.
Why the annual stocktake comes too late
An annual stocktake gives you a snapshot at a single point in time. The snapshot is not the problem — the delay is. If a data entry error occurred in March, you find it in December. By then the supplier invoice is paid, the customer involved is gone, and nobody remembers anything.
A discrepancy found within seven days is traceable. You pull up the delivery note, ask the person who received the goods, and understand the cause. A discrepancy found eleven months later is a straight loss written off.
The other cost is direct: a day of closure is a day of lost revenue, plus overtime for the team.
The principle: split the stock, not the time
Cycle counting rests on a simple idea. You do not count everything at once — you spread counting across the year according to how important each reference is.
Not all references deserve equal attention. A reference representing 30% of your revenue and turning over weekly needs frequent checks. A reference sold three times a year can be checked once a year.
ABC classification exists for exactly this:
- Class A — roughly 20% of references, accounting for 80% of value. Monthly count.
- Class B — roughly 30% of references, 15% of value. Quarterly count.
- Class C — roughly 50% of references, 5% of value. Twice-yearly or annual count.
These percentages are orders of magnitude, not law. Adapt them to your business: a low-value item that is highly prone to theft belongs in class A regardless of its share of revenue.
Building your counting schedule
Take a business with 1,200 active references.
| Class | References | Frequency | Counts per year | References per week |
|---|---|---|---|---|
| A | 240 | Monthly | 2,880 | 55 |
| B | 360 | Quarterly | 1,440 | 28 |
| C | 600 | Twice-yearly | 1,200 | 23 |
That comes to roughly 106 references per week. At two minutes each, this is under four hours weekly, spread across several days and several people. An hour each morning before opening is more than enough.
Compare that with a full day of closure and an entire team tied up.
The five rules that make the difference
Count before opening or after closing. During trading hours stock moves: a sale in progress during the count creates a discrepancy that is not one.
Count blind. Whoever counts must not see the theoretical quantity. Otherwise, faced with doubt, they will recount until they reach the expected figure. This is the most common bias, and it empties the exercise of meaning.
Separate the counter from the approver. The person entering stock movements should not be the one checking their own entries.
Analyse discrepancies, don't just correct them. Correcting a quantity without finding the cause guarantees the gap will return. Always record the reason: receiving error, data entry error, undeclared breakage, theft, misplaced item.
Never postpone a count. One skipped week becomes two, and then the whole system collapses. Better to reduce the weekly volume and hold to it than to aim too high and give up.
What you measure after three months
Cycle counting produces an indicator an annual stocktake never gives you: the stock accuracy rate, meaning the percentage of counted references whose actual quantity exactly matches the theoretical quantity.
Stock that is 95% accurate or better lets you make purchasing decisions without manual verification. Below 85%, your replenishment alerts are wrong and your team ends up walking the aisles before every order, which cancels out the benefit of the software.
Also track the breakdown of discrepancies by cause. If 60% of gaps come from receiving, the problem is not in the counting — it is upstream, at the point where goods enter.
Where to start this week
You do not need a six-month project to begin.
- Pull the list of your references sorted by annual sales value, highest first.
- Take the top 30. These are your priority class A items.
- Count them Friday morning before opening, blind.
- Record every discrepancy with its presumed cause.
- Repeat next week with the following 30.
After a month you will know whether your discrepancies come from receiving, the till, storage or shrinkage. That information is what has value — far more than the corrected figure itself.
Where software fits
Cycle counting can be run on paper, but it rarely survives more than two months. The schedule gets lost, discrepancy reasons are never consolidated, and nobody calculates the accuracy rate.
Stock management software brings three concrete things: it generates the list of references to count based on their class, it records the discrepancy and its reason at the moment of counting, and it produces the history that shows whether things are improving.
In G-stock, every count creates a traced adjustment movement with its reason and its author. The monthly report shows the accuracy rate by product family and by warehouse, so you know exactly where to focus.