Why stock accuracy drifts – and why the annual stocktake doesn't fix it

Every warehouse has a version of the same story. A picker reaches a location and the pallet isn't there. A customer orders twenty and the system says eighteen. Nobody is being careless; errors creep in through a hundred small gaps. A mis-pick gets corrected on the shelf but not on the system. A damaged case is binned without a write-off. A return is put away in the nearest empty slot rather than its home location. Goods in are booked against the wrong purchase order line.

The traditional answer is the annual full stocktake. It is expensive, disruptive, and for most operations out of date within a week. Worse, it tells you that you have a problem without telling you why. Cycle counting takes the opposite approach: small, frequent counts built into the working day, so errors surface while the trail is still warm.

What cycle counting looks like day to day

A cycle count is not a mini stocktake. It is a continuous audit. Each day, or each shift, a counter works through a short list of locations and records what is physically there. In a typical ambient operation that might be thirty to fifty locations, taking well under an hour.

Two details matter. First, count locations rather than products. A pallet of the right SKU in the wrong bay is still a pick failure waiting to happen. Second, use blind counts: the counter records the quantity without seeing what the system expects. If the expected figure is on the sheet, people tend to find what they expect to find.

Deciding what to count, and how often

You rarely need to count everything equally. Classify stock by value, throughput and risk, then set frequency accordingly.

• A items – high value, fast moving or customer-critical. Count monthly, or even weekly for your top lines.
• B items – moderate value and movement. Count quarterly.
• C items – slow movers and low-value consumables. Count once or twice a year.
• High-risk areas – pick faces, returns, quarantine, damaged goods and anything near a despatch lane. Count these often regardless of value; this is where errors accumulate.

Timing matters as much as frequency. Count a location after replenishment and before picking begins, not in the middle of a pick wave. Build the schedule into quieter periods so counting never competes with despatch, and aim for full coverage of every location at least once a year, with your busiest areas covered far more often.

Setting tolerances and investigating the gaps

No operation hits zero variance, and chasing it everywhere wastes time. Set tolerances by item: zero for serialised or high-value goods, a small percentage for low-value fasteners sold in bulk.

When a count falls outside tolerance, don't simply adjust and move on. That habit – the tick-and-adjust culture – hides the causes and guarantees the same errors return. Instead:

• Recount the location with a second person before touching the system.
• Check the obvious explanations first: stock in the wrong location, an unrecorded write-off, a return not yet booked in, a unit-of-measure error between cases and eaches.
• Look at recent transactions for that SKU – receipts, picks, transfers, adjustments – to narrow down when the error appeared.
• Adjust only once the cause is understood and approved, and record the reason code.

Making it stick

Cycle counting fails when it is treated as an extra job for whoever happens to be free. Give it an owner, a published schedule and a handful of simple measures: locations counted against plan, location accuracy, quantity accuracy, and the value of adjustments made.

Rotate counters so nobody audits the aisles they pick from, and train them properly on the scanner or count sheet. Technology helps – most warehouse management systems will generate count lists and enforce blind entry – but the discipline matters more than the software. Review results monthly and look for patterns. A cluster of errors in one aisle usually points to a labelling, layout or lighting problem rather than carelessness.

Get it right and the numbers speak for themselves. Operations that count regularly typically hold accuracy above 98 per cent, which means fewer stockouts, less safety stock tied up on the racks, cleaner audit trails and a far calmer peak season. The annual stocktake stops being a cliff edge and becomes a formality.