How much milk does a coffee shop use? Do the math

To find how much milk your coffee shop uses, multiply the number of each milk drink you sell by the ounces of milk in it, add them up, and add about 10 percent for steaming loss. Divide by 128 to get gallons. The averages you find online are somebody else's shop. Yours takes one afternoon with a sales report.

Below is the full method with worked arithmetic. The numbers are an example cafe, not Stella's, so swap in yours.

Why the averages online don't fit you

Search for how much milk a coffee shop uses and you'll find figures like "a few gallons per hundred drinks." It sounds useful and it is nearly useless, because milk use moves with at least four things you control:

• Drink mix. A shop that sells mostly drip coffee uses a fraction of what a latte-heavy shop does.
• Size. A 16 oz iced latte carries more milk than a cortado by a factor of three or more.
• Alt milks. Oat, almond, and whole milk get ordered, priced, and spoiled separately, so one blended gallon figure hides the part that matters.
• Day of week. The Saturday rush is a different shop from Tuesday.

At Stella we go through about 30 gallons of whole milk a week, and I would be wrong to hand you that number as a target. A smaller shop might use half and a busy one several times more. What carries over is the formula.

The formula: drinks times ounces

For each milk drink on your menu:

milk used (oz) = drinks sold x ounces of milk per drink

Then:

total milk (oz) = sum over all drinks x 1.10

The 1.10 is a steaming and rinsing allowance for the milk that stays in the pitcher, gets purged, or goes down the drain. Ten percent is my starting assumption. Replace it after you compare a week of sales against your counts. That gap is exactly what the usage variance report is built to show.

Ounces per drink is the number people guess. Don't. Pour a 12 oz latte the way your barista does, then measure the milk with a scale (a fluid ounce of milk weighs a little over an ounce) or a measuring cup. Do it for each size of each drink. It takes fifteen minutes and it is the most valuable fifteen minutes in this post.

Pulling drink counts from your POS

You need quantity sold per drink and size for a stretch of days. Every major POS can export that. The report names vary, so look for an item-level sales report that you can filter by date and export:

• Square: the item sales report in your dashboard, filtered to your date range and exported as a CSV. Sizes show as variations of each item.
• Clover: item sales under reporting in the web dashboard.
• Toast: the product mix report, which lists items sold with quantities for a period.

These reports work whether or not you ever connect a tool to your POS. That matters for Toast in particular, because our Toast integration is built and waiting on Toast's partner approval, so I won't pretend it is available to connect today. The CSV export works now.

Pull at least two full weeks, including two weekends. One week can be skewed by weather or a holiday.

If you later connect your POS, you can skip the exports. ShelfCount syncs sales on Square about every 15 minutes, and Clover connects with a dashboard API token. The Square, Clover, and Toast guides each say what syncs. For milk, the part that matters is mapping each menu item to a recipe so a sale turns into ounces. See building recipes and mapping your menu.

Worked example: a 250-drink day

Here is an example cafe on a normal weekday. These are assumed numbers for illustration, not a benchmark.

Add the 10 percent allowance:

• 1,720 x 1.10 = 1,892 oz
• 1,892 / 128 = 14.8 gallons a day

Now split by milk type. Say 30 percent of milk drinks are made with oat, which you can read straight from your modifier sales:

• Oat: 1,892 x 0.30 = 567.6 oz. At 32 oz per carton, that is 17.7 cartons.
• Dairy: 1,892 x 0.70 = 1,324.4 oz, or 10.3 gallons.

That split is why the single "gallons" figure misleads. You order oat by the case of cartons and dairy by the gallon, and the two have different shelf lives and different prices.

Weekday versus weekend

Weekends rarely match weekdays, so don't multiply one day by seven. Take the same example cafe and say Saturday sells 340 drinks and Sunday 300, against 250 on weekdays. If the mix is the same, milk scales with drinks:

• Saturday ratio: 340 / 250 = 1.36, so 10.3 x 1.36 = 14.1 gallons of dairy
• Sunday ratio: 300 / 250 = 1.20, so 10.3 x 1.20 = 12.4 gallons of dairy
• Weekday: 10.3 gallons of dairy

Weekly dairy: five weekdays at 10.3 plus 14.1 plus 12.4 comes to 78.0 gallons. A weekly average of 11.1 gallons a day would underestimate your Saturday by about 3 gallons, which is a stockout at 2 p.m. with a line out the door.

The mix usually shifts too, not just the volume. Weekend customers tend toward bigger drinks, more iced, more alt milk. Run your numbers from a weekend report to check.

Turning it into a par

Usage only matters when it becomes an order. The par level post covers the full logic. The short version is par equals daily usage times days between deliveries, plus a buffer.

Say your milk vendor delivers Monday and Thursday morning. Dairy gallons, from the weekday and weekend figures above:

• Monday to Thursday (3 days: Mon, Tue, Wed): 3 x 10.3 = 30.9 gallons
• Thursday to Monday (4 days: Thu, Fri, Sat, Sun): 10.3 + 10.3 + 14.1 + 12.4 = 47.1 gallons

Add a 15 percent buffer:

• Monday delivery par: 30.9 x 1.15 = 35.5, call it 36 gallons
• Thursday delivery par: 47.1 x 1.15 = 54.2, call it 55 gallons

Those are order-up-to levels. If Thursday's count shows 6 gallons left, the suggested order is 55 minus 6, which is 49. Check that your walk-in can hold 55 gallons, and that the milk's shelf life after delivery covers the four days. Milk has a ceiling a shelf-stable item doesn't have: carrying more than you'll pour before the date just converts a stockout into waste.

You don't have to do this arithmetic by hand every time. The free par level calculator takes daily usage, delivery gap, and buffer and gives you the number. A count sheet with milk on it, counted before each delivery, gives you the on-hand to subtract. In ShelfCount you can set a different par for each day of the week, and the suggested order uses today's number. The setup is in vendors and purchase orders.

Checking your math against reality

A formula is a hypothesis. Test it. Count your milk at the start and end of a week, add up the invoices in between, and compute actual use:

actual use = opening count + received - closing count

Say you opened the week with 30 gallons of dairy, received 40 and 52, and closed with 41. Actual: 30 + 92 - 41 = 81 gallons. The formula said 78.0, so you are 3 gallons or about 4 percent over. The 78.0 forecast already includes the 1.10 allowance, so a 4 percent gap sits within normal noise and is worth watching, not chasing. If actual use had come back at 95, something is off. Either the drink mix differs from what you assumed, the ounces per drink are higher than you measured, or milk is leaving in ways the register never sees: staff drinks, spills, a warm cooler. The first two you fix by measuring. The last two are the reason you do this.

Close that gap and your food cost numbers become ones you can act on. That is the same comparison the usage variance report makes per item for the whole menu, with theoretical usage from sales on one side and your counts on the other. Milk is usually where I start, because it is both the biggest line and the one that moves fastest.

If you're still doing this in a spreadsheet, that's fine, and it's how I started. When you want sales to turn into ounces automatically, ShelfCount for coffee shops is built around exactly this loop at $50 a month flat, and the pricing page has the details. There is a 14-day free trial.