A cycle count form lists the handful of items you're counting this week, with columns for what you counted, who counted it, what the records expected, and the difference in units and dollars. The schedule behind it comes from sorting your items into A, B and C by how much money goes through them: A items, the few that are most of your spending, get counted every week, B items monthly, C items once a quarter. For a small kitchen that's about five items a week instead of everything at once. Below is the sort, the schedule and the form, filled in, free in Excel.
The short version
- Sort items by what you spend on them in a year. The top ones that add up to about 80% of the money are A.
- Count A items weekly, B monthly, C quarterly. Same day, same time, before production.
- Fill in Expected afterwards, in the office, so the counter isn't steered by it.
- Chase the differences that cost money, and write down what you found next to the number.
- TaroStack picks what to count for you, from what being wrong would cost. More on that below.
Step 1: sort your items into A, B and C
Take each item's yearly usage times its cost. Here's Windward Roots, the made-up taro business I use in these examples, with eleven items:
| Item | Used in a year | Unit cost | Value used | Share | Running share | Class |
|---|---|---|---|---|---|---|
| Taro paste | 2,246 kg | $3.84 | $8,626 | 37.4% | 37.4% | A |
| Bread flour | 8,254 lb | $0.50 | $4,127 | 17.9% | 55.3% | A |
| Butter | 991 lb | $4.00 | $3,962 | 17.2% | 72.5% | A |
| Coconut milk | 624 cans | $2.25 | $1,404 | 6.1% | 78.6% | A |
| Grated taro | 468 kg | $2.90 | $1,357 | 5.9% | 84.5% | B |
| Sugar | 1,321 lb | $0.91 | $1,202 | 5.2% | 89.7% | B |
| Instant yeast | 132 lb | $5.50 | $726 | 3.1% | 92.8% | B |
| Labels | 9,360 | $0.06 | $562 | 2.4% | 95.3% | C |
| Brown sugar | 344 lb | $1.10 | $378 | 1.6% | 96.9% | C |
| Bread bags | 7,488 | $0.05 | $374 | 1.6% | 98.5% | C |
| Kūlolo trays | 1,872 | $0.18 | $337 | 1.5% | 100.0% | C |
Four items are 79% of the money. That shape is typical, and it's why counting everything equally wastes most of the effort: a miscounted pack of labels costs 6 cents, a miscounted tub of taro paste costs dollars.
The cut-offs are a judgment call. I've used 80% for A and 95% for B, a sensible place to start. Put an item in a higher class if it's small in dollars but stops production when it runs out, or if it tends to walk off.
Step 2: the schedule
A items weekly, B monthly, C quarterly:
| Class | Items | Counts a year each | Counts a year |
|---|---|---|---|
| A | 4 | 52 | 208 |
| B | 3 | 12 | 36 |
| C | 4 | 4 | 16 |
| Total | 11 | 260 |
That's 260 counts a year, five a week on average. In practice it looks like this for the first weeks of the quarter:
| Week of | Count these |
|---|---|
| Oct 5 | Taro paste, bread flour, butter, coconut milk, grated taro |
| Oct 12 | The four A items, sugar |
| Oct 19 | The four A items, instant yeast |
| Oct 26 | The four A items, labels |
Five items is ten minutes or so. Pick a day and a time before production starts, and keep it.
Step 3: the form
| Date | Item | Class | Location | Counted | Expected | Variance | Unit cost | Variance $ | What we found |
|---|---|---|---|---|---|---|---|---|---|
| Sep 26 | Taro paste | A | Walk-in | 24.9 kg | 26.80 | −1.90 | $3.84 | −$7.30 | Recipe uses 7.5 kg a batch, not 7.2: weighed one |
| Sep 26 | Bread flour | A | Dry store | 91 lb | 91.27 | −0.27 | $0.50 | −$0.14 | |
| Sep 26 | Butter | A | Walk-in | 50 lb | 51.45 | −1.45 | $4.00 | −$5.80 | Greased pans: add 0.25 lb to each batch |
| Sep 26 | Coconut milk | A | Dry store | 60 cans | 60 | 0 | $2.25 | $0.00 | |
| Sep 26 | Grated taro | B | Walk-in | 11.2 kg | 11.00 | +0.20 | $2.90 | +$0.58 |
Net, the week is $12.65 short. Ignoring the signs, $13.81 was wrong, almost all of it in two items. And the last column is the one that pays: the taro paste gap turned out to be a recipe that understates what a batch really uses, which means every cost and every reorder that depends on that recipe was off too. One cycle count found it.
Counted is filled in on the floor. Expected is filled in afterwards, from the records, by someone else if you can manage it. That's what keeps the count blind. The variance formulas and how to judge which gaps matter are in the inventory variance formula, and the full-count version of this form is the inventory count sheet example.
The spreadsheet
Download the cycle count form. It has three tabs: Items, where you type usage and cost and the class works itself out; Schedule, a quarter of weeks; and Count form, which prints one page wide. It's a plain Excel file with no macros, and it works in Google Sheets, LibreOffice and Numbers.
The class formula doesn't need the list sorted. "Running share" adds up every item worth at least as much as this one:
Running share =SUMIF($E$2:$E$12, ">="&E2) / SUM($E$2:$E$12)
Class =IF(G2<=80%, "A", IF(G2<=95%, "B", "C"))
The cut-offs and how often each class is counted sit in three cells beside the table, so changing them re-sorts everything.
Does cycle counting replace the annual count?
For running the kitchen, mostly yes: every item gets counted several times a year, and the expensive ones every week. For your books, ask whoever does them whether cycle counts are enough or whether they want a full count at year end too; the year-end inventory count covers that one. If you do both, the year-end count goes quickly, because the numbers have been right all year.
Where this stops working
Two things wear it down. The first is Expected. To fill in that column, someone has to work out what the records say for each item, which means every delivery, batch and sale since the last count added up by hand, for five items, every week. The second is the schedule itself: usage changes, a new product makes sugar an A item, and the classes from January are wrong by June unless someone re-sorts them. A schedule that nobody updates turns back into counting whatever's easy.
How TaroStack does it
TaroStack does the sorting and the scheduling for you, and goes a step past A, B and C. It suggests what to count next from what it would cost to be wrong, times how much has happened to that item since anyone last looked, so an item that's moved a lot rises up the list even if it's cheap, and an item that's never been counted is flagged on that alone. The schedule moves with your kitchen without anyone re-sorting a spreadsheet.
The Expected column disappears too. Every delivery received, batch recorded and sale made has already been adding and subtracting, so when you count on a phone, in the walk-in with no signal if need be, TaroStack shows each difference and its dollar value before you post it, while you can still go and look. From your own batch history it checks whether a recipe yields what it claims, and which ingredient is off, which is the taro paste finding, made without waiting for a count to stumble on it.
Counting is on every plan, from $49 a month; the recipe checks come with recipes and batches on Standard at $99. The first 30 days are free, and your items import from this spreadsheet. If you'd like a hand setting up, ask, and we'll do it with you.
