How Often Should a Coffee Shop Take Inventory
Mark, founder of Parly·September 2, 2026·6 min read
The short answer
Three full counts a week, on Monday, Wednesday, and Friday. A 60-second look at five perishable items every morning on top of that. Once a week is the floor, and daily full counts are a mistake almost every time.
That is what I run: about 60 items, counted three mornings a week, plus the morning eyeball on milk and pastry. A recent full count took 8.5 minutes for 58 items, so the whole weekly cost of that cadence is under half an hour.
The rest of this is why those are the right days, and how to adjust them for a shop that orders on a different clock than mine.
Your supplier calendar sets the cadence, not a rule of thumb
Counting has exactly one job: to tell you what to order before the window closes. So the cadence is not a hygiene habit or a discipline exercise. It is set by the days you can place orders and the days the truck arrives.
Work it backward.
- My dairy cuts off in the late afternoon for next-day delivery, no Sunday truck.
- My paper and alt-milk supplier cuts off mid-morning for next-day delivery, six days a week.
- My roaster takes orders in a weekly window.
- Matcha and chai run five to seven business days.
Given that, a count has to land before each cutoff I intend to use. Monday, Wednesday, and Friday covers every one of them with a day to spare, and each count has a different job:
Monday shows the weekend damage. Saturday and Sunday burn through more than any two weekdays and there was no delivery on Sunday, so Monday's count is the one most likely to find something at par.
Wednesday checks whether Monday's order was right and still leaves time to fix Thursday and Friday. Without it you find out on Friday, which is too late to be gentle about it.
Friday sets up the weekend before the suppliers who do not run Sunday go quiet. This is the count that prevents the Saturday emergency run, and it is the one shops skip most often, because Friday is busy.
If your suppliers deliver twice a week rather than daily, two counts is right. If you have one roaster and no perishables, weekly is genuinely enough. The pattern to copy is not my days, it is the method: list every supplier's cutoff and delivery days, then place counts so that every cutoff you use has a fresh count behind it. That mapping is the same work behind never missing a supplier window.
✦ The test for your own cadence
If a count ever tells you something you needed to know yesterday, you are counting too rarely. If two consecutive counts have never once changed an order, you are counting too often.
The five items that earn a daily look
Full counts three times a week does not mean you ignore the shelf on Tuesday. It means the daily look is short and specific.
Five items, roughly, and the list is the same in most cafes:
- Your default milk. The fastest-moving thing in the building, and the one item where running out closes most of the menu.
- Your second milk. Whichever alt milk carries your swaps.
- Pastry. Counted against the standing order at delivery, because the discrepancy is only visible in the first hour.
- Your busiest cup size. Almost always the 16 oz cold cup, and it fails on a Saturday, never on a Tuesday.
- Whatever ran out last month. Rotating slot. If something surprised you, it earns a daily look until you trust the par again.
That is a 60-second glance on the way past, not a count session. You are not recording a number for the usage math; you are checking that nothing is about to fail before the next real count. Write down anything below par and keep moving.
Everything else can wait for Monday. Cafiza does not need a daily check. Neither do napkins, filters, gloves, syrups, or trash bags. They move slowly enough that three counts a week is high resolution, and every one you add to a daily list is time you take from a count that mattered.
Why counting everything daily fails
The standard advice, borrowed straight from restaurant operations, is to count high-value perishables daily. It is written for a kitchen with proteins on a two-day shelf life and a walk-in full of prep. Applied whole to a cafe it is wrong, in three specific ways.
It costs more than it returns. Nine minutes a day, six days, is 54 minutes a week to learn what three counts already told you. On a slow item, six data points a week and three data points a week produce the same order.
It degrades. A daily count that a manager resents becomes a copied number. I have watched shops where the "daily count" was last week's numbers retyped, and that is worse than not counting, because now the wrong number carries the authority of a record. A cadence you keep honestly beats a cadence you keep on paper.
It hides the signal in noise. Cafe usage varies enormously by day. A Saturday-to-Sunday delta and a Monday-to-Tuesday delta are different animals, and stacked daily they look like chaos. Counting on the same three days each week gives you comparable intervals, and comparable intervals are what turn counts into a usage rate you can set a par level from.
| Daily full count | Three counts plus a daily look |
|---|---|
| 45+ minutes a week | Under 30 minutes a week |
| Copied numbers by week three | Short enough to stay honest |
| Noisy day-to-day deltas | Comparable intervals |
| Nobody reads the output | Every count feeds an order |
What has to be true for three to be enough
Three counts a week only works if the counts are trustworthy and the gaps are covered.
The gaps are covered by your sales. Between Monday and Wednesday you did not stop knowing what left the shelves; every drink you sold poured a known quantity of a known ingredient. That is the difference between a cafe and a stockroom: your register already holds a running record of usage, if the recipes and the modifiers behind it are mapped. The count is the correction, not the whole record. When the counted depletion and the sales-derived usage disagree, that gap is your waste, and it is the number worth chasing.
The counts are trustworthy if they are fast, consistent, and in the same units every time. Nine minutes, before open, walk order, same partial rule. The mechanics are in the walkthrough, and they matter more than frequency: two clean counts beat five sloppy ones, every time.
Pick your days this week
Open your last month of supplier confirmations and write down three things per supplier: cutoff time, lead days, delivery days. It takes fifteen minutes and most owners have never had it on one page.
Then put counts in front of the cutoffs you actually use. If everything you buy is next-day, Monday, Wednesday, and Friday will fall out of the exercise on its own. If your roaster and your dairy are your only two suppliers, you will find two days is enough and stop doing a third for no reason.
Then set the daily five. Write them on the sheet as their own short list, at the top, so the morning look takes a minute and not a decision.
The cadence you can hold for a year beats the ambitious one you abandon in March. Start at three, keep the sheet in walk order, and adjust only when a count tells you something too late.