Ask ten restaurants how often they count stock and you will get monthly, monthly, monthly, and then one place that swears by Sunday nights. Ask why, and almost nobody has a reason beyond habit or whatever the last manager did. It is one of those questions that sounds like it should have a standard answer and does not, because the honest answer is that frequency belongs to the item, not to the restaurant as a whole.
A case of tinned tomatoes and a box of fillet steak are both stock, and treating them the same is what makes counting feel like a punishment.
Why “How Often” Has No Single Answer
The reason a single number does not work is that a stock count is a measurement, and measurements are worth taking at the speed the thing you are measuring actually moves.
Dried pasta sitting on a shelf does not do much between Tuesday and Friday. A tray of langoustines does. If you count both on the same monthly rhythm, the pasta gets measured far more often than it needs to be, and the langoustines have had four weeks to go wrong before anyone looks. You end up spending most of the counting effort on the stock least likely to surprise you.
There is a second reason, and it is the one that decides everything in practice. A count you cannot sustain is worth less than a slower count you actually run. Plenty of kitchens have started a weekly regime in January and quietly dropped it by March, which leaves them worse off than the monthly count they abandoned, because now there is a gap in the record too. Worth saying plainly: weekly full counts are not standard practice across most of Europe, and any advice that treats them as the obvious baseline is describing a different market.
So the useful question is narrower than how often you should count. It is which of your stock earns a closer look, and how often that particular stock changes enough to be worth the walk.
The Three Counting Tiers and What Belongs in Each
Most kitchens sort cleanly into three tiers once you stop thinking about the whole store cupboard at once.
The close watch, weekly or thereabouts. Proteins, seafood, dairy, spirits and wine, plus anything else that is expensive per kilo or easy to walk out of the building with. This tier is usually a small slice of your item count and a large slice of your money, which is exactly why it earns the attention. In most kitchens it is somewhere between fifteen and forty lines, and a practised person can get through them in twenty minutes.
The full count, monthly. This one covers everything, the close-watch tier included, timed to the period you report on. This is the count that closes the books and feeds your food cost percentage, which is where your variance figure comes from. It is the backbone, and it does not move.
The slow shelf, quarterly or when something looks off. This covers the cleaning supplies, the disposables, the long-life dry goods and whatever else lives at the back of the store cupboard. These lines are cheap and slow moving, and they almost never turn out to be the source of a problem. Counting them monthly does no harm beyond spending effort that could have gone somewhere more useful.
Bars often want their own rhythm on top of this, because spirits and wine combine high value with the easiest kind of loss to explain away. If that is your situation, counting a bar quickly is a separate craft worth reading up on.
How to Sort Your Own Stock Into the Tiers
You do not need a system to do this, and it only takes one afternoon.
Pull your last full count and sort it by value, which means unit cost multiplied by how much of it you hold. Look at where the money actually sits. In most kitchens something like a fifth of the lines carry three quarters of the value, and that fifth is your close-watch tier before you have thought about anything else.
Then adjust for two things the value column cannot see. The first is speed. An item can be cheap and still deserve a weekly look if you go through enormous quantities of it, because a small percentage of a large number is still real money. The second is temptation. Spirits, premium cuts, and anything portable and desirable belong on the close watch regardless of what the value sort says.
What you should end up with is a short list you could count before service without dreading it, and a long list you touch once a month. If the short list has grown past what one person can do in half an hour, it is too long, and something on it got there through anxiety instead of evidence.
What Changes When a Count Takes One Hour Instead of Eight
None of the above survives contact with a real kitchen unless one other thing is true.
Every argument for counting more often runs into the same wall, which is that counting is slow. When a full count means a clipboard, a walk-in, a spreadsheet the next morning, and someone reconciling it against invoices for another two hours, of course it happens monthly. Nobody is going to do that weekly. The frequency question and the method question are the same question wearing different clothes.
Change the method and the frequency argument dissolves. A count run on a phone, item by item, totalling as it goes, turns eight hours into one or two. At that point a weekly close watch stops being a project and becomes something a chef does between deliveries. The routine itself matters as much as the calendar, because a count taken the same way every time is the only kind you can compare against last week’s.
Settle that before you set any schedule. If your current count is a full day, do not plan a weekly tier. Fix the count first, then let the frequency follow.
The Cost of Counting Too Rarely
The reason any of this matters is detection lag: the distance between something going wrong and you finding out.
Say a supplier reworks a pack size without mentioning it, or a new starter is plating forty grams over on a popular dish. On a monthly count, that runs for up to four weeks before it shows up in a number, and by then it has become a lump of variance you cannot attribute to anything in particular. You know you lost money. You do not know to what. That gap between theoretical and actual food cost is only useful if it arrives while you can still remember what changed.
Shorten the interval on the items where money actually moves and the same problem surfaces in seven days, small enough to trace and small enough to fix. Nothing about the loss changed. What changed is how much of it you had to pay before finding out. Even a few percent of variance adds up to a number worth catching early.
Frequently Asked Questions
How often should a restaurant do a stock count? A full count monthly, aligned to your reporting period, plus a weekly check on high-value fast-moving items like proteins, seafood, dairy and spirits. Slow-moving low-value goods can wait for a quarterly look.
Is weekly stock counting realistic? For a short list of high-value items, yes, and it takes twenty to thirty minutes when the count runs on a phone. A weekly count of everything is a different proposition and is uncommon across most of Europe.
Should every location in a group count on the same day? Yes, if you want to compare them. Counts taken on different days describe different weeks, and any variance you see between sites is partly just the calendar. Groups usually fix a group-wide count day for exactly this reason.
Does counting more often actually reduce food cost? Counting does not reduce anything on its own. What it does is shorten the time between a problem starting and you seeing it, which is what makes the problem fixable while it is still small.
Finding Your Rhythm
The schedule that works is the one that puts attention where your money is and leaves the rest alone. Run the monthly full count that closes your books, add a weekly pass on the short list of items carrying the value, and glance at the slow shelf once a quarter. Then make the count fast enough that the weekly tier is not a negotiation every time.
Stockifi handles the counting on mobile and web, including offline in the walk-in, and works out variance the moment a count closes so the numbers are ready before you have left the building. See how stock counting works.