Cycle Counting: how to implement it without stopping the warehouse

Cycle counting is the practical way to keep stock accurate without closing the warehouse. Instead of one full count at year end, you count a handful of locations every day. Discrepancies surface in days, you fix the cause, and the operation never stops.

This guide shows how to build the process: what to count first, how often, who counts, how to calculate accuracy, and what to do when the numbers do not match.

What cycle counting is

Cycle counting is a continuous inventory audit. You count small samples of references or locations at scheduled intervals so that, across the year, the whole warehouse is verified, some zones more often than others.

It belongs to the same model as the perpetual inventory system: the record updates with every movement, and counting confirms that the record still matches reality.

Step 1: classify products by rotation (ABC)

Not everything deserves the same attention. ABC analysis ranks references by value moved or number of movements and sets how many times a year each group gets counted.

The percentages are a starting point, not a rule. If you have low-value references that keep running out, move them up a class. The criterion is risk, not just euros.

Step 2: size the daily count volume

The maths is simple. Add the planned annual counts per class and divide by working days:

With 2,000 references split into 400 A, 600 B and 1,000 C, that is 4,800 + 2,400 + 1,000 = 8,200 counts a year, about 33 per working day. One person handles that in little more than an hour with a handheld terminal.

Step 3: choose the count trigger

• By calendar. A reference joins the list when its class-driven date comes up.
• By opportunity. The system asks for confirmation when a location hits zero during picking. That is the cheapest moment to count.
• By exception. Whenever there is a pick discrepancy, an odd return, or a manual adjustment, that location goes into the queue.
• By zone. One full aisle at a time, useful in warehouses with dense locations.

Step 4: count blind

One rule changes results: whoever counts should not see the expected quantity. If the terminal shows "48 expected", the human tendency is to confirm 48 without counting. Blind counting means the operator enters what they see, and only the system compares.

Second rule: recount before any adjustment. A first discrepancy is a hypothesis, not a fact.

Step 5: measure accuracy properly

The indicator that matters is not "how many units are missing overall". It is accuracy per location:

If you counted 200 locations and 12 had a discrepancy, you are at 94%. A mature operation lives above 98%. Track the indicator per zone and per shift, because that is where the problem shows: it is rarely the whole warehouse, it is almost always one aisle or one process.

Step 6: treat the cause, not just the number

A stock adjustment closes the symptom. What prevents repetition is recording the reason:

• Receiving error (wrong quantity or wrong reference)
• Picking from the wrong location
• Product without a SKU or with an unreadable barcode
• Unrecorded internal movement
• Shrinkage, damage, or theft

After a month, the reason map tells you where to invest: training, labelling, or a review of the location layout. It is also how you cut stockouts caused by phantom stock.

Mistakes that ruin the process

• Always counting the easy references. Hard-to-reach ones usually carry the most error.
• Counting while the zone is moving. Freeze the location during the count or the numbers will never close.
• Letting lists pile up. A deferred count is a useless count, because the picture has already changed.
• Using counts to grade people. If the operator fears the result, the error gets hidden.
• Never running a full count. A complete physical inventory still makes sense when you change site or change system.

How LogisticsWMS runs cycle counting

In LogisticsWMS, counts are created per location and executed on the terminal, recording who counted and when. Stock regularisations carry a reason code, which keeps every adjustment auditable and turns history into diagnosis.

The Ticks assistant covers the analytical side: ABC classification from real sales rotation, detection of references with recurring discrepancies, and replenishment suggestions. Questions like "which locations had the most adjustments this month?" get a direct answer, with no exports.

Frequently asked questions