How long should a stock count actually take?
A well-run count should not take a full day for most businesses. Time it properly and the numbers are usually far lower than people assume, because the reason a count takes a day is almost never the counting itself — it is everything around it.
Counting itself is the quick part. What eats the day is everything else: re-walking aisles to re-count items the first person was unsure about, being pulled away to help a customer, and then two or three hours at a desk typing numbers into a spreadsheet that already holds most of them.
Break the day down before you decide it is unavoidable:
count day cost = (people counting x hours) x (loaded hourly cost)
+ hours spent correcting discrepancies
+ hours spent typing the results into the system
+ hours lost to staff being pulled away to serve customersThe last line is usually the biggest and the one nobody mentions, because an assistant counting stock cannot also serve the counter.
Why does a stock count turn into a whole day's work?
Four things reliably cause it.
Everything is counted in one pass, by one team, on one day. Counting every bin and every shelf on the same day means the shop either closes or the count is rushed, and the corrections come later anyway.
The count sheet is separate from the system. The count happens on paper, a spreadsheet is built from the paper, and then the spreadsheet is typed into the stock system. Three copies of the same number, three chances to be wrong.
A discrepancy stops the count. The moment two people disagree on a quantity, the whole process halts for an investigation. Without a rule that says how to resolve it in under two minutes, a single item can eat an hour.
Reports are built from scratch every time. The stock-on-hand report, the reorder report and the variance report are each assembled by hand, usually at month end, usually by the most senior person, usually late.
What is a cycle count, and how is it different?
A cycle count is the same activity spread across the year instead of compressed into one day. Each counting session covers one category, one aisle, or one group of fast-moving lines, and it happens on a fixed repeating cycle.
A workable pattern for most businesses:
| Item type | Share of your stock lines | How often to count |
|---|---|---|
| High value, high movement | A small minority of lines | Weekly or fortnightly |
| High value, slow movement | Small minority | Monthly |
| Bulk, low value, low movement | The large majority | Once or twice a year |
| Dead or obsolete stock | Small | Once a year, then written down |
You will notice that the lines that get counted most often are the ones that have caused the most argument and the most shrinkage. Count those properly and the big annual count stops being a crisis.
Spread the work across the month rather than the week, and keep each session to a fixed length — ninety minutes, then stop, even if unfinished. A partial count that gets done is worth more than a complete count that does not.
How do you measure what your stock take really costs?
Do the calculation above, then add the cost of the stock errors the count was supposed to find. Every wrong stock figure produces one of three problems: stockouts on lines you had, excess stock you cannot move, or a variance that quietly erodes your margin.
Calculate it like this:
cost of stocktake = (hours lost) x (loaded hourly cost)
+ value of write-off of obsolete stock found
+ margin lost to lines that were out of stock while on the shelfThe middle line is often the one that surprises people. A single line of slow-moving stock that nobody reorders and nobody discounts can sit there for years.
Which items should be counted first?
Sort your lines by movement, value, and how often they have been wrong. Anything scoring high on all three goes into the weekly count.
Practical signals:
- Lines that appear on the reorder list repeatedly and then sit unsold.
- Lines where the physical count and the system have disagreed more than once.
- Anything you would be genuinely upset to discover missing.
- High-value lines where a small quantity error matters: fittings, brackets, spares, consumables.
- Stock held on a vehicle or on site rather than in the store, where it is the easiest to go unrecorded.
How do you stop the count being typed twice?
The fix is to make the counting device the input. A phone scanning barcodes as you walk the aisle produces the count in the system, with no paper in between.
If that is not possible yet, the interim version still removes a step: print a count sheet from the system showing what it believes is there, and have the counter mark the actual quantity and condition next to it. A variance list of only the differences is far smaller than a full count, and it is the list that matters.
Then set the rules before you start, not during:
- Anything uncertain is counted again immediately, by a second person. Two counts, no discussion.
- A difference above a set threshold triggers a recount, not an argument.
- Nothing is adjusted in the system until the physical count for that aisle is signed off.
- Every adjustment carries a reason code: damage, expiry, theft, miscount, supplier error.
- The reason code is what turns next year's count into a shorter one.
What reports can be produced instead of built by hand?
Most manual reporting is a filter over data the system already holds. Ask, for each report, what decision it supports, and who reads it. If the answer is that nobody reads it, stop producing it.
Reports worth keeping usually fall into three groups:
| Report | What it tells you | Frequency |
|---|---|---|
| Stock on hand by line | What is actually there | Continuous, never hand-built |
| Reorder suggestions | What to buy, and when | Weekly, with a named buyer |
| Slow-moving and obsolete lines | What to discount, bundle or write off | Monthly |
| Stocktake variance by reason code | Where the losses actually come from | After each count |
| Stock versus sales by category | Whether the range is right | Monthly |
| Free stock and committed stock | What you can actually promise a customer | Continuous |
What do you do when the count and the system disagree?
Treat every discrepancy as information rather than as an accusation, and record the reason. Most of the time the pattern tells you which fix matters: miscounts point to a location problem, theft to a process problem, supplier errors to purchase order handling, and damage to storage.
Publish the variance by reason code, so the same category does not quietly eat the same value every count. If shrinkage keeps landing in one category, more careful counting is almost never the answer — it is a change to how that category is stored, labelled and checked.
The companion templates include a count sheet layout, a cycle count calendar, a variance investigation form and a one-page report specification you can hand to whoever currently builds your reports by hand.
If stock counts and month-end reports are eating a day a month, it is worth working through which part of it to fix first. Book a free strategy session with Agentcy.