How to do a stocktake in a restaurant, step by step

A stocktake is not a counting exercise. It is a measurement of the gap between what your recipes say you used and what you actually used, and that gap is where the money goes. Everything below exists to make that number trustworthy enough to act on.

Before you count anything

1. Close the till period. Counting while sales are still landing guarantees a wrong figure.
2. Enter every delivery that has physically arrived, and nothing that has not. A delivery sitting unentered will read as if you lost the stock.
3. Record transfers between sites, bars or kitchens. Untracked transfers are the single most common cause of a variance that turns out to be nothing.
4. Print or open your count sheet in the order you will walk, not alphabetically. Alphabetical sheets make people criss-cross the building and skip lines.

The counting route

Always the same route, every week. Consistency matters more than speed, because a line skipped in the same place every week looks exactly like theft.

1. Dry store — sealed and unambiguous, so it warms you up.
2. Walk-in fridge, shelf by shelf, top to bottom.
3. Freezer. Do this quickly and do it last among the cold areas; nobody counts accurately at minus eighteen.
4. Kitchen line — the fridges under the pass, the sauces, the prepped garnish.
5. Bar, cellar and back bar, left to right, never doubling back.

Two people — one counting aloud, one recording — is faster and materially more accurate than one person doing both. The counter never looks at the sheet, so they never anchor to the expected number.

Part-used items

This is where most stocktakes lose their accuracy. An open 5kg sack of flour, half a wheel of parmesan, a part-used bottle. Choose one method and never mix them:

• Weigh it. The most accurate, and worth it for expensive lines — cheese, meat, spirits.
• Estimate to the nearest tenth. Perfectly adequate for cheap, high-volume items.
• Never guess in halves. 'About half' is where consistent errors hide, because everybody rounds the same direction.

Reading the result

Your report should show, per line, what you should have — opening stock plus deliveries minus what your recipes say you sold — against what you counted, with the difference in cash. Sort by cash value. A forty-unit discrepancy on cordial matters less than four missing bottles of premium gin.

Then look for direction, not size. A line that is short one week is a counting error. A line short four weeks running is a process problem: over-portioning, an unlogged wastage habit, a wrong recipe spec, or someone helping themselves. Only the fourth of those is theft, and it is the least common.

How often

Weekly if you can, monthly at worst. A rough count done every week beats a perfect count done every quarter, because you are looking for trends and a trend needs points on a graph. Most venues that count quarterly are not measuring their kitchen; they are writing its history.

FoxEra takes count sheets from a phone — including photographing the paper sheet your team already uses — and produces the variance report in cash automatically.