Why system stock and real stock always drift apart
Even in the most up-to-date system, the quantity shown for a warehouse item is almost always an estimate, not a verified fact. Every time something leaves the stockroom without being logged precisely — waste thrown out without recording it, a portion more generous than the recipe says, a typo on a received quantity, a small theft — the system number and the real quantity on the shelf drift a little further apart. No single error is serious, but they add up.
The physical count is when these two numbers meet again: you actually count what's on the shelf and in the walk-in, and update the system to match reality. Without periodic counts, "system stock" becomes an increasingly imaginary number, and every decision built on it — when to reorder, what the stock is worth, whether a dish is really losing money — inherits that error.
How often to count
Not everything in the stockroom deserves the same frequency:
- Monthly full count: the entire stockroom, every zone. This is the minimum for a reliable number for the P&L and for stock valuation.
- Weekly rolling count: a subset of high-value or high-turnover products, counted more often to catch drift early before it grows.
- Targeted count after an event: a supplier change, a shift with heavy staff turnover, a suspected shortage — all moments worth counting off-schedule rather than waiting for the next due date.
How to organize the count, step by step
1. Pick a low-activity moment. Before opening or at the end of the day, never during service: counting while the kitchen is still working produces wrong numbers because the stock is moving under your eyes.
2. Freeze receiving during the count. If a delivery arrives mid-count, keep it separate and add it only after closing the count — otherwise you lose track of whether a case has already been counted or not.
3. Organize the count sheet by zone, not alphabetically. Walk-in, dry storage, bar, freezer: count zone by zone following the physical path of the space, so you never skip anything or count the same shelf twice.
4. Count in the same unit the product is recorded in. If the system tracks a product in kg and you count "3 packs," someone has to convert — and that's where the most common transcription errors are born. Better to weigh or count directly in the correct unit.
5. For high-value products, have two people count independently. Not out of distrust: even the most careful count has a margin of error, and on an expensive product that margin is real euros. Two independent counts that match give confidence; if they don't, you know right away a third check is needed.
6. Log the count the same day, not "later." A paper sheet filled out during the round and transcribed into the system hours later, maybe by someone else, is another point where numbers can go wrong.
What to do with variances (don't just overwrite)
When the counted number differs from the system one, the temptation is to update the system to the real number and move on. That's correct — that's exactly what the count is for — but it's worth looking at the size of the variance first:
- A small, recurring variance across many different products is usually normal: small portioning errors, weight loss, rounding. Update and move on.
- A large variance on a single product deserves one more question: did a delivery never get logged? Was there unrecorded waste? Is it a typo from weeks ago only now being noticed? Understanding the cause avoids repeating the same hole next month.
- A systematic variance always in the same direction (always less than expected, never more) on a specific product is the pattern that deserves the most attention: it's the difference between random error and a recurring problem.
The most common mistakes
Counting during service. The number you get isn't the real stock at any precise moment, it's somewhere between before and after — useless for reconciliation.
Forgetting a zone. The bar, external storage, the staff fridge: if a zone never enters the count round, that stock stays "invisible" and keeps drifting from the system.
Not freezing new deliveries. A case arriving mid-count and counted along with everything else skews the number, whether it's included by mistake or forgotten because "it just arrived."
Updating the system without logging the reason. An adjustment with no recorded cause doesn't help you tell whether the same variance keeps happening for the same reason every month.
How BiteBase handles physical counts
A physical count in BiteBase updates stock for every counted product in bulk, replacing the system number with the actually counted one — not an increment or decrement, an absolute value. Every adjustment generates a tracked movement, so later you can still see when and by how much a product's stock was corrected by the last count.
One thing to keep in mind: a physical count doesn't automatically trigger a low-stock alert, even if the corrected number drops below the configured minimum threshold. After a count that reveals a significant shortage on a critical product, it's worth checking thresholds manually instead of expecting an automatic notification.
Frequently asked questions
How long does a full count take? It depends on the size of the stockroom, but for a mid-sized restaurant, budget 1-2 hours with two people, less if the count is well organized by zone with count sheets ready in advance.
Is it better to count by weight or by piece? In whatever unit the product is recorded in the system. Mixing units (counting "pieces" for a product recorded in kg) is the most common cause of transcription errors.
Should the chef be involved in the count, or is the storeroom manager enough? For menu-critical products, yes: the chef more easily spots an anomaly (a product that should be much more or much less than what's found) than someone who doesn't work with that product every day.