Optimize the management of yourstocks and inventories

Why the stock count never matches

What the variance actually says

• A small, stable variance — the system is reliable. You can start steering by it, and a sudden move becomes a usable signal.
• A large variance on a few references — almost always a badly calibrated recipe or a wrong yield on those specific products, not a counting problem.
• A diffuse variance across everything — a flow is missing: receipts not recorded, productions not validated, or sales not attached to recipes.
• A variance widening over time — unrecorded loss, breakage or shrinkage. The only one of the four that really is a problem on the floor.

Every business that counts its stock knows the variance: the count does not match what was expected. The usual reaction is to suspect the count. That is rarely the right lead — in most cases the variance mostly measures the absence of a reliable theoretical stock to compare against.

Theoretical stock is built from three flows: what comes in (supplier receipts), what goes out in production, and what sales consumed through recipes. If any one of the three is missing, the variance becomes a number with no possible interpretation.

Count quickly, count often

What makes a count usable

• A consistent counting unit — counting in purchase units while storing in production units produces a purely arithmetic variance. The conversion belongs on the reference, not in the head of whoever is counting.
• A fixed point in time — counting while production runs mixes two states of the stock. The date and time of the count are part of the data.
• High-value products first — ten references often represent half the value tied up. Counting those weekly beats counting everything quarterly.
• Capture on the spot — copying a count sheet into a spreadsheet adds a source of error to an operation that already has enough.

The quarterly full count is an expensive ritual producing a snapshot that is out of date the next day. The practice that works is the opposite: count few references, but often, prioritising the ones that carry value — stock value is almost always concentrated in a minority of products.

A rolling count by zone or family, done from a phone at the end of service, takes minutes and ties up nobody. Its value is not completeness but frequency: a variance seen after three days can still be explained, one seen after three months cannot.

What stock drives downstream

The three dependent decisions

Valuing stock: which method, and why it moves the result

Separating waste from theft and from error

• The real value of the asset — stock is cash tied up. Its valuation feeds profitability indicators directly, and an approximate valuation distorts the result as much as a margin error.
• What valuation influences — the month's food cost, the asset value on the balance sheet, and the turnover ratio. Three figures management reads, and a method changed quietly moves all three.
• Turnover, the underused indicator — how many times stock renews over a period. Slowing turnover flags overstock before cash flow notices.
• Record waste as it happens — breakage, dropped product, failed preparation, unsold stock binned. Recorded waste leaves stock with a reason; unrecorded waste becomes unexplained shrinkage and clouds the diagnosis.
• Look at reasons, not just amounts — ten kilos binned as unsold and ten kilos binned after a cold-chain break cost the same and call for opposite actions.
• What to order, and how much — the purchasing requirement is planned production minus available stock. Without the second term you order out of habit and store cash in the walk-in.
• What you can produce — a production plan that ignores available stock schedules shortages or remakes what already exists.

Reliable stock is not valuable in itself; it is valuable for what it lets you decide. Three decisions depend on it directly, and go badly without it.

On top of that sits the regulatory dimension: stored batches, their use-by dates and their origin are what food-safety traceability requires you to retrieve, and that information lives in the stock or nowhere.

Two businesses holding the same physical stock can report different values depending on the method. At latest purchase price the value tracks the market but turns volatile; at weighted average it smooths the swings but lags an increase. The choice matters less than the consistency: switching method mid-year makes two periods incomparable.

A stock variance has three possible causes, and confusing them leads to treating the wrong one. Material waste is a production or forecasting problem: it is fixed by adjusting quantities produced or yields. A recording error is a procedural problem: it is fixed by making goods-in or validation more reliable. Unexplained shrinkage, once the first two are ruled out, is the only one that is a people question — and it is far rarer than intuition suggests.

Frequently asked questions about stock management

Do we have to inventory everything to start?

How are products weighed at the point of use handled?

Are use-by dates tracked?

What about multiple sites?

How long before variances become reliable?

Can we manage several storage locations on one site?

Do we need scanners or specific hardware?

One full cycle: a baseline count, then in and out flows over a period, then a second count. Before that second count the variance mostly measures how imprecise the starting point was, not real consumption.

Yes — dry store, chiller, freezer, cellar. That is what makes rolling counts practical: you count a zone, not a whole site.

No. A phone is enough for counting and for recording waste. Dedicated hardware earns its place on high goods-in volumes, not to get started.

No, and it is counterproductive. A baseline count on the high-value families is enough to establish a starting point. Secondary references come in through rolling counts.

Through the recipe: stock is issued on the net weight from the card, adjusted for yield. That is more accurate than weighing at every gesture, and it asks nothing of the kitchen.

Yes, at batch level. That is what lets you issue the oldest batches first and spot what is about to expire before it does, rather than discovering it as you throw it away.

Each site has its own stock, with traced transfers between them. Group-level consolidation gives the total value tied up and surfaces imbalances — one site out of stock while another overstocks the same reference.