How to Take Inventory at a Coffee Shop, Fast

Ten minutes, not forty

A recent full count at my shop was 58 items in about 8.5 minutes, from a phone. That is not a stopwatch stunt. It is Monday, Wednesday, and Friday, and my manager has run the last 26 of 26 of them without me in the building.

The reason that matters is not the time saved, though nine minutes three times a week adds up. It is that a count nobody dreads is a count that actually happens. Every cafe I know that stopped counting stopped because the count took forty minutes, ate the pre-open, and produced a number nobody looked at afterward. Speed is what makes the habit survive contact with a Tuesday.

The forty-minute count is not slower because the counter is slow. It is slower because of four fixable things: the room is disorganized, the sheet is in the wrong order, the units are ambiguous, and the counter is making decisions instead of counting. Fix those and the same 58 items take nine minutes.

Fix the room before you fix the count

You cannot count fast in a room you have to search. Twenty minutes of moving things, once, is the highest-return work in this whole article.

One home per item. Every item lives in exactly one place. The moment oat milk lives both in the reach-in and in the back stack, every count of oat milk becomes a hunt, and hunts are where the minutes go.

Face the labels out. Sounds fussy. It is the difference between reading a shelf and picking up every box on it.

Break the backstock line of sight. If you can see the full depth of a shelf from where you stand, you can count it without moving anything. Stacks two deep on a low shelf mean kneeling, and kneeling is thirty seconds per shelf.

Group by supplier where you can. Not strictly, and not at the cost of the walk, but the closer your physical shelves match your order sheets, the less translation happens later.

Do this once, properly, and it holds for months. Do it never, and you will pay for it three times a week forever.

The walk, zone by zone

The single biggest speed lever is the order of the list. Count in the order you physically walk the room, one loop, no backtracking. Alphabetical lists and category lists both force you to cross the shop repeatedly, and crossing the shop is the whole cost.

My loop, six zones:

1. Bar reach-in. Milks, cream, half and half. The fastest zone and the one that changes most, so it goes first while attention is highest.
2. Bean and tea shelf. Bags, tins, concentrate bottles.
3. Syrups and specialty. Whatever sits next to the tea, physically.
4. Paper storage. Cups, lids, straws, napkins, filters. Cases and sleeves, so it counts fast despite being the biggest group.
5. Under the sink. Cafiza, soaps, sanitizer, gloves, bags.
6. Pastry case. Received that morning, counted against the standing order.

Six stops, one direction, ending at the front where you started the day. The sheet is in that exact sequence, which is what a real cafe count sheet is for: the sheet is a route, not a catalog.

The four rules that keep it honest

Speed only counts if the number is right. Four rules do all the work, and every one of them exists because breaking it cost me an order.

Count in the unit you order in. If you buy oat by the case of twelve and count by the bottle, someone will write 3 and someone else will read 3 cases. Pick the unit that goes on the purchase order, write it next to the item, and never carry two units for one item. This is the single most common source of a count that does not match reality.

Same time of day, every time. Before open, always. A count taken at 7 AM and one taken at 4 PM are not comparable, and if your counts drift across the day, your usage math is comparing a full morning shelf to a picked-over afternoon one and calling the difference consumption.

Partials get a rule, not a judgment. An open bottle of oat is not 1 and not 0. Decide once: round down, or count halves, and write the rule at the top of the sheet. Consistency beats accuracy here, because your usage math cares about the change between two counts, and a consistent bias cancels out while a coin flip does not.

Do not fix anything while counting. You will see a nearly empty jar of cafiza and want to go put it on an order. Do not. Stop counting and you lose the loop, the rhythm, and four minutes. Write the number, keep walking, decide at the end. Counting and ordering are two jobs, and doing them at once is what turns nine minutes into forty.

Where the minutes actually go

The total is measured. The split below is my own reconstruction of where it goes, and it is close enough to plan around:

Paper is a third of the count because it is a third of the items, not because it is hard. Everything else is under 90 seconds a zone. Nine seconds an item is not fast typing; it is what happens when you never search and never decide.

A phone beats a clipboard for a specific reason worth naming: the number goes in once. Clipboard counts get transcribed later, and transcription is both a second pass over the same 60 items and the place where a 6 becomes a 60. If you count on paper, at least type it in standing at the shelf, not at the end of the day from memory of your own handwriting.

Then hand it off

The count is the most delegatable job in the shop, and holding onto it is a habit worth breaking. My manager has run the last 26 of 26. Nothing in the loop above requires an owner: the room is organized, the sheet is a route, the units are written down, and the partial rule is at the top of the page.

What does require you is the decision at the end. The count produces numbers; the numbers hit par levels; the pars say order or wait. Keep that. Give away the walk. The mechanics of handing it over without losing accuracy are worked through in delegating the count, and the short version is: walk it with them twice, then stop walking it.

Run your first timed count this week. Set the sheet in walk order, count before open, and note how long it took at the bottom. Do it again Wednesday. The second one will be faster than the first by a minute or two, and the fourth will land somewhere near nine. Once the count costs you nine minutes instead of forty, the argument about whether to count at all is over, and the only thing left to decide is what the numbers tell you to order.