Stop stock from moving, walk the shelves in a fixed order with a sheet that matches that order, write down what's physically there in the units it's stored in, and only afterward compare it with your records. Then chase the differences that cost the most money first. Most counts that go badly went wrong in the first step: things kept moving while people were counting.
The short version
- Pick a time when nothing is being received, produced or shipped. Early morning before production works.
- Tidy first. Half-open cases get consolidated, strays go home.
- Print a count sheet in shelf order, without the expected quantities on it.
- Count what's there, in the unit it sits in: cases, bags, each. Convert later.
- Compare to your records. Sort the differences by dollars, not by units.
- Recount the big ones, find out why, then correct your records with a reason attached.
Before you count
Freeze movement. If a delivery arrives mid-count, it waits by the door with a sign on it. If production has to run, count the production area first and lock in those numbers before anyone scales an ingredient. A count is a photograph. It's no good if the subject keeps walking around.
Get the paperwork caught up. Anything received but not yet entered, anything shipped but not yet invoiced, any batch made yesterday but not recorded. If the records are behind, every one of those will show up as a "variance" and waste your afternoon.
Tidy. Consolidate the three open bags of almond flour. Bring the case of lids back from wherever it wandered to. Count day is when you discover the second, forgotten storage spot for labels, so check the office, the van and the top shelf.
Set the order. Walk the space the way you'd read a page: left to right, top to bottom, one storage area at a time. Dry storage, then walk-in, then freezer, then packaging, then finished goods. Build the count sheet in that same order. A sheet sorted alphabetically while the shelves are sorted by whatever-fit-where is the biggest time-waster in inventory counting.
The count sheet
| Location | Item | Count unit | Count | Counted by | Notes |
|---|---|---|---|---|---|
| Dry, shelf 1 | Flour, bread | 50 lb bag | |||
| Dry, shelf 1 | Flour, bread (open) | lb, by scale | |||
| Dry, shelf 2 | Sugar, cane | 50 lb bag | |||
| Dry, shelf 3 | Vanilla extract | qt bottle | |||
| Walk-in, left | Butter, unsalted | 36 lb case | lot / use-by: | ||
| Walk-in, left | Heavy cream | qt | lot / use-by: | ||
| Packaging | 5 oz bottles | case of 144 | |||
| Finished goods | Choc chip cookies, 4-pack | each | lot: |
(There's a fuller count sheet example here.) Notice what isn't there: the quantity you're supposed to have. That's deliberate. If the sheet says "14" next to flour, the counter's eyes find 14. A blind count makes people actually count.
Opened containers get their own line. Weigh them if the ingredient is expensive (vanilla, chocolate, nuts), and eyeball to the nearest quarter if it's cheap (flour, sugar). Don't spend ten minutes weighing a $6 bag of salt.
While you're standing in front of the dated stuff anyway, write down the lot or use-by date. It costs seconds, and it's how you catch the cream that's about to expire behind newer cream.
For anything valuable, count in pairs: one person counts and calls, the other writes and repeats it back.
After the count: variances
Now bring out your records and compare. The variance for each item is:
Variance = counted − expected, and in dollars, variance × unit cost.
| Item | Expected | Counted | Variance | Unit cost | Variance in $ |
|---|---|---|---|---|---|
| Flour, 50 lb bag | 14 | 12 | −2 | $25 | −$50 |
| Butter, 36 lb case | 6 | 7 | +1 | $144 | +$144 |
| Vanilla, qt | 3 | 1 | −2 | $90 | −$180 |
Sorted by units, flour and vanilla tie. Sorted by dollars, vanilla is the problem, and it's the one where a two-unit difference is plausible as theft, a spill, or a recipe that uses more than it says.
Note the butter. A positive variance isn't good news, it's a record that's wrong in the other direction. Most often it's a delivery that was put away and never entered, which also means the invoice might not have been checked.
(More on the inventory variance formula if you want percentages too.)
Net variance here is −$86 (−50 + 144 − 180). Absolute variance, ignoring signs, is $374. Track the absolute one. The net number flatters you, since errors in opposite directions cancel out while both remain errors.
Recount anything big before you believe it. Then look for the reason: a receipt not entered, a batch not recorded, a unit mix-up (the record is in pounds, somebody counted bags), waste nobody wrote down. Fix the record with an adjustment that has a reason on it. "Count correction" with no explanation, month after month, is how a real problem stays hidden.
How often
A full count at least once a year for your tax return: Schedule C asks for your inventory at the beginning and end of the year, even for small businesses using the simplified rules. Your accountant will want the number and the date. There's a checklist for the year-end count.
For actually running the place, once a year is nowhere near enough. Monthly full counts are reasonable for a small shop. Better still, add cycle counts: every week, count just the ten items where being wrong hurts most, meaning the expensive ones, the ones you run out of, the ones that vary. Fifteen minutes a week keeps the important numbers honest between full counts. There's a cycle count form here.
Where this stops working
Paper sheets get typed into a spreadsheet afterward, which is a second chance to make mistakes and means you learn about the $180 vanilla problem two days later, when nobody remembers anything. Phones and tablets would fix that, except the walk-in and the back of the storeroom are exactly where the Wi-Fi gives up. And a count is only as good as the freeze. In a business that can't stop for a morning, stock moves while you count it.
How TaroStack does it
TaroStack counts on a phone or tablet, or on printed sheets if you prefer paper, and it keeps working with no signal. You count in the walk-in, and it sends itself when you're back in range. Count sheets follow your storage locations. You can count a whole site or just a section, and the app suggests what to count first based on what being wrong would cost and how much each item moves, which is your weekly cycle count chosen for you.
Before you post a count, you see what every variance is worth in dollars, so the recount happens while you're still standing there. Adjustments are recorded as corrections with a reason and a name, and they can be undone. You can scan barcodes with the phone's camera or a Bluetooth scanner, and units convert on their own, so counting bags against a record kept in pounds isn't a problem.