It's Sunday night. The spreadsheet is open, you're three cups of tea in, and you're trying to remember whether that half-tub of jam at the back of the fridge counted as full or not. The kids are in bed. Monday's orders aren't done yet. And tomorrow morning you'll open up, take one look at the number, and file it away never to be looked at again.

If that sounds familiar, you're not doing stocktakes wrong. You're doing the wrong stocktake.

Why monthly stocktakes quietly fail

A full, bottom-to-top count once a month looks thorough on paper. In practice it's the worst of all worlds.

It takes hours, so you put it off and do it late.
By the time the number lands, the month is already over. You can't change anything.
Because it's so rare, any single miscount swings the result. A forgotten case of oat milk in the cellar can make your food cost look 3% worse than it is.
Most of the count is long-tail stock that barely moves the needle.

You end up with a figure you don't trust, covering a period you can't influence, built from a count you dreaded doing. No wonder it gets ignored.

The honest alternative a lot of cafés land on is no stocktake at all. Just eyeball the bank balance and hope food cost percentage is behaving. That's not a system. That's a prayer.

The Pareto cut: your top 20 lines

Here's the bit nobody tells you. In almost every independent café, 20% of your ingredient lines drive roughly 80% of your food spend. Count those weekly, count everything else monthly, and you get most of the signal for a fraction of the effort.

Your top 20 lines usually look something like this:

House coffee beans
Whole milk (and oat milk, these days probably at the top)
Butter
Cheese (whichever block goes on everything)
Flour
Eggs
Chicken or the main protein
Bacon
Bread or sourdough from your baker
Sugar
Your hero cake or pastry ingredient

Your list will be different. The point is: these are the lines that hurt when the price moves, and the lines that vanish when there's waste or theft. They're also the lines that appear in the most recipes, so a small recipe drift shows up quickly.

How to pick your 20

Two ways, and you want both.

Highest spend by line. Pull the last three months of invoices. Rank ingredients by total £ spent. Top 20, done.
Highest frequency across recipes. If you've already built your recipes in CostingBrik, you can see which ingredients appear in the most dishes. These are your exposure lines.

The overlap between the two lists is where you focus. A line that's expensive and in half your menu is the one you want eyes on every week.

The 15-minute Friday count

Friday morning works for most cafés. The week's trading is mostly done, the weekend order is about to go in, and you've got a clear head before service. Pick your day and stick to it.

Here's the format that works:

One sheet. One page. Your 20 lines, pre-printed, in the order you walk the kitchen.
Consistent units. Count in the unit you buy in. Kilos of flour, litres of milk, whole blocks of cheese plus the open one weighed.
Same person, same route. Pick the path through your storage that makes sense and walk it the same way every week. Fridge, freezer, dry store, coffee station.
No heroics. If a jar is roughly half, write "0.5". You're after trend, not forensics.

A clipboard, a biro, and a tea towel list beats nothing. A phone with a shared sheet is better. A connected system that pulls invoice prices automatically is better again, but the principle is the same. Don't let tool shopping stop you starting.

Fifteen minutes, once a week. That's the bar.

What you do with the number

This is the part most cafés miss. A count on its own is just a number. The value comes from comparing two things.

Theoretical COGS. What your recipes say you should have used, based on what you sold. Sales mix × recipe cost, across the week.
Actual COGS. What actually left the shelves. Opening stock + purchases - closing stock.

The gap between the two is your truth-telling number. It captures:

Waste (the croissants that went in the bin on Sunday)
Theft (yours, your staff's, suppliers short-delivering)
Miscounts (the forgotten case in the cellar)
Recipe drift (the new chef pouring 40ml of milk when the recipe says 30)
Pricing errors (you're selling something at the old cost)

What a healthy variance looks like

For a well-run café counting 20 lines weekly, expect actual COGS to run roughly 1-3% higher than theoretical. That gap is normal waste, spillage, and the fact that recipes are never perfectly followed.

Under 1% and you're either not counting carefully or not capturing all the waste. Over 4-5% and something's up. Over 8% and you've got a problem that's costing you real money every week.

Where the tools help (and where they don't)

You can absolutely run this on paper. Plenty of good operators do. But a couple of places the software earns its keep.

Theoretical COGS is the hard bit to do by hand. You need recipe costs kept current as supplier prices change, multiplied by your actual sales mix from the till. That's what CostingBrik does. Recipes stay priced, recipe costing stays accurate, and the theoretical number comes out the other side without you rebuilding a spreadsheet.
Actual COGS needs live stock movement. Proper stock tracking - opening count, receipts from invoices, closing count, variance per line - is on the roadmap as StockBrik. For now, a weekly count of your top 20 into a simple sheet gets you 80% of the way there.
Price changes need catching. When a supplier sends a price increase, your theoretical COGS shifts the moment those invoices are processed. That's the whole point of keeping recipes live.

The tool isn't the system. The system is: count 20 lines, every week, same day, compare to what you sold. The tool just makes the maths honest.

When the gap gets ugly

A growing variance is one of the earliest warning signs you'll ever get. Before the P&L comes in, before the accountant's quarterly chat, before the bank balance starts looking thin.

If the gap has been widening for three or four weeks running, walk the kitchen and ask:

Has a recipe changed without the costing being updated?
Is there a new staff member pouring by eye?
Has the waste bin got fuller, or the fridge got fuller of half-used tubs?
Did a supplier quietly change pack size?
Is anything walking out the back door?

Any of these are recoverable once you can see them. The monthly stocktake hides them for weeks. The weekly count on 20 lines surfaces them almost immediately.

The takeaway

Stop trying to count everything once a month. Start counting the 20 lines that matter, every Friday morning, in 15 minutes. Compare what you should have used to what you actually used. Watch the trend.

Friday, not Sunday night
20 lines, not 200
15 minutes, not 2 hours
Every week, not every month

That's the system. It's not clever. It's not exciting. It just works, and you'll actually do it, which is the only metric that matters.

The jam at the back of the fridge can wait until month end.

Ed O'Brien has run Hunters Cake Company for 17 years across cafés in Witney, Burford, and a bakery in Carterton, Oxfordshire. He's building Brikly - modular tools that give independent café owners the same data the big chains have, without the big chain price tag.