Restaurant Cellar Inventory: Frequency, Method, Automation

Cellar inventory is the job nobody wants — repetitive, slow, and a reliable source of argument between the floor and the office. It is also the only exercise that puts a real number on your beverage assets and shows you where the theoretical margin is evaporating. Here is the full method: how often, with what tool, what variance to accept, and how to move to continuous counting so you never lose half a day to it again.

Why a good count changes everything

A bad count costs you three things at once: a balance sheet that does not match reality, a wine list offering bottles you do not have — which the guest discovers at the table — and no way to measure your real breakage, which means no way to fix it. A good count gives you the base for steering profitability, negotiating with distributors on real volumes, and finding the dormant stock tying up capital without producing anything.

How often to count

Three rhythms coexist in the trade, each with its own logic:

• Full monthly count — the norm in most restaurants. Every bottle counted, reconciled against the month's purchases and sales. Gives you a clean close and honest margin tracking. Budget two to five hours depending on the size of the cellar.
• Rolling partial counts — you count a section each week, reds one week and whites the next, with a full count each quarter. Lighter in any given week, but it demands discipline that most operations do not sustain.
• Continuous — every movement, deliveries in and sales out, is tracked in real time through your POS connected to your list. The periodic count becomes a spot check rather than a recount. This is the most efficient mode, and it is what Winevizer puts within reach of mid-sized operations.

FIFO or LIFO?

For wine, the answer is almost always FIFO — oldest bottles out first. It stops references with a drinking window, older vintages and young whites in particular, from sitting past their best. LIFO only makes sense on the few references where age adds value, and even then the arbitrage usually belongs in a separate reserve cellar rather than in your working stock.

What to count with

Paper (unworkable past fifty references)

The clipboard is the historical method. It works for a short list and becomes a source of errors past fifty to eighty references: crossings-out, sheets that go missing, lines counted twice. Nothing feeds back into a central system.

Spreadsheets

The next step. Readable, shareable, exportable. It demands rigour, though: any spreadsheet shared between several people eventually grows competing versions, broken formulas, and archive copies scattered across personal drives. Workable for a mid-sized cellar, not for a group.

Dedicated software

Dedicated software couples the count to a living list: one entry feeds the cellar, the guest-facing list, the printable version and the analytics. With a POS integration, counting becomes marginal — you verify that theoretical stock matches physical stock rather than recounting from zero. Winevizer's cellar is built for lists of fifty to a thousand references.

How much variance to accept

No count is perfect. There is always a gap between theoretical stock, which is calculated, and physical stock, which is counted. The question is which gap you should tolerate. The working rule in the trade:

• Under 1% by value — excellent, your controls are tight.
• 1 to 2% — acceptable, the normal residue of breakage and service loss.
• 2 to 4% — watch it, and find the references that are drifting.
• Above 4% — a real signal. Either your data entry is wrong, or you have a management problem: theft, unreported breakage, or a discrepancy with a distributor.

The open-bottle trap

A bottle opened for the by-the-glass program sits in an intermediate state: no longer full stock, not yet zero while there is wine in it. Most manual counts ignore this and count in whole bottles, which builds a structural bias into every close. Winevizer's by-the-glass program counts remaining pours per open bottle and feeds inventory at the pour, not at the bottle.

Moving to continuous counting

Continuous inventory rests on three pieces:

1. Clean entry of purchases, with deliveries captured in the system either by hand or by reading the invoice.
2. Sales flowing back in real time, which means a POS integration.
3. A periodic control count — monthly, no full recount, just verification.

On Winevizer these are native from the moment your POS supports the integration. At that point the monthly inventory becomes a thirty-minute verification where it used to take three or four hours.

Going further

Inventory is a steering tool, but what it reveals — dormant stock above all — calls for action to recover the capital sitting still. That is the subject of the piece on clearing dead stock.