The annual physical inventory is a familiar routine: close the warehouse for a weekend, count everything, find differences, adjust, and hope the numbers hold until next year. A cycle count program replaces that one big event with small counts every day or week. Every item gets counted on a schedule, and errors turn up while there is still time to find out why they happened.
Why count in small pieces
- Errors are found sooner. A receiving mistake found two weeks later can still be traced to the delivery and the supplier. Found eleven months later, it is just a write-off.
- The counts are more careful. One person counting a short list at the start of a shift pays more attention than a tired crew late into a weekend count.
- You keep shipping. Counts fit between picks instead of replacing them.
- You learn where the problems are. Variance by item and by location over time points to causes. One annual number does not.
Step 1: Rank your items
Not every item needs the same attention. A common method is ABC ranking, by SKU velocity or by value:
- A items: the small group of items that makes up most of your sales or stock value. Count them often, for example monthly.
- B items: the middle group. Count less often, for example quarterly.
- C items: the long tail. Count once or twice a year.
Where you draw the lines is your choice, and the example frequencies are a starting point, not a rule. Move items into the A group for other reasons too: high value per case, easy to steal, a history of variance, or perishable enough that a stock error quickly turns into waste. The article on reading your own velocity reports covers how to sort items by movement.
Step 2: Work out the daily load
Turn the frequencies into a daily count list. A hypothetical example with round numbers:
| Group | Items | Counts a year each | Counts a year |
|---|---|---|---|
| A | 200 | 12 | 2,400 |
| B | 600 | 4 | 2,400 |
| C | 1,200 | 2 | 2,400 |
| Total | 2,000 | 7,200 |
Over 250 working days, 7,200 counts is about 29 a day. Time a sample of real counts in your own building before you commit to a schedule. If the daily load is too heavy, lower the B and C frequencies before you touch the A group.
Step 3: Count without stopping work
- Count at quiet times, such as before the pick wave starts or after the trucks have left.
- Count blind. The counter should not see the expected quantity, or they will tend to find it.
- Handle stock in motion. A pick, receipt or move for the item between the count and the record changes the answer. Count when the item is not being picked, or make sure movements during the count are recorded and allowed for.
- Count by lot for perishables, with the dates. The right quantity of the wrong lot still breaks FEFO and traceability.
- Check every location for the item, including overstock, staging and returns, before calling a difference a variance.
Step 4: Recount, then adjust
Set a tolerance. A small difference on a cheap item can be adjusted on the spot. A larger difference, or any difference on a high-value item, gets a recount by a second person before anything changes.
Record a reason for every adjustment: receiving error, picking error, unrecorded damage, unit confusion (case versus each), theft, or unknown. "Unknown" is allowed, but if it becomes the most common reason, the process for finding causes is not working. See inventory adjustments for how to keep write-offs under control.
Step 5: Fix causes, not just counts
An adjustment corrects the number. The cause will keep producing new errors until someone deals with it. Each month, look at variance by item, by location and by reason:
- One item keeps coming up short: check how it is received, how it is stored, and whether its case and unit setup is right.
- One area keeps showing errors: check the labeling and the layout.
- Variance clusters after certain deliveries or on certain shifts: look at that process, and coach rather than blame.
Does it replace the year-end count?
Sometimes. Whether you still need a full physical count at year-end is a question for your accountant, your auditors and your lenders. Some accept a well-run cycle count program with good records; others still want a full count. Ask them before you change anything. Either way, a year of cycle counting means the year-end count starts from records that have been checked all year.
Keep the program alive
Cycle count programs tend to fade when they depend on paper and good intentions. What helps:
- A count list generated for each counter each day, so nobody has to decide what to count.
- Counts entered on a phone or handheld at the location, not copied from paper later.
- A short weekly report of counts done against counts planned, and a manager who reads it.
If counts are missed two weeks running, find out why. The load may be too high, or counting may be the first thing dropped whenever the dock gets busy. Either way, fix the schedule rather than letting the program lapse.
How APFoodz handles it
In APFoodz, counts are done in the warehouse app alongside the rest of the floor work, and stock is tracked by lot (inventory and lots).