How Much Coffee a Cafe Goes Through in a Week
Mark, founder of Parly·September 2, 2026·7 min read
The bean order you place from memory
It is Sunday night and the roaster wants the order by Monday at 9. You are on the couch with your phone, and the decision comes down to a feeling: last week four bags of the Peru felt about right, so make it four again. Maybe five, because the weather is turning.
Look up how much coffee a shop goes through in a week and you get a forum thread. Someone runs 40 pounds, someone else runs 12, a third person says it depends. All three are telling the truth about their own bar and none of them are telling you anything about yours. A cafe pulling 400 espresso drinks a week and brewing two batches a day is a different animal from one selling 120 pour overs, even if both call themselves a coffee shop.
Your number is not a benchmark you look up. It is four small numbers you already have, multiplied together. The whole calculation takes about ten minutes and it is the same shape as working out your real milk usage: sales times recipe, not survey times hope.
Every espresso drink is a double shot
Start with the rule that decides most of your bean weight, and get it stated plainly before you multiply anything.
At my shop every espresso drink is a double. Not the cappuccino as a single and the latte as a double. Not a single for a small and a double for a large. Every espresso drink that goes across the bar pulls the same double basket, every time, no exceptions. That is a house rule, and it is the most useful thing you can decide about your bean math, because it turns "how much coffee per drink" into one number instead of a menu of guesses.
If your shop splits singles and doubles by size, you now have two numbers and a ratio between them to track, and the ratio moves every week with what people order. If your shop says everything is a double, your espresso consumption is drink count times one dose. That is it.
The second half of the rule is the dose itself. Weigh it. Not the number printed on the grinder, not what the last barista told you: put a portafilter on the scale and pull five doses back to back. Most shops land somewhere between 17 and 20 grams for a double, and the spread between 17 and 20 is nearly 18 percent of your entire espresso spend, which is a real amount of money to leave unmeasured.
💡 Weigh it, do not remember it
Five doses, one scale, two minutes. Everything downstream in this post is your drink count times that one number, so it is worth being exact once rather than approximate forever.
Work the week from your drink counts
Now the arithmetic. Pull last week's sales from Square by item and count every drink that pulls a shot: lattes, cappuccinos, cortados, americanos, macchiatos, the dirty chai, the espresso tonic. Everything with espresso in it, counted as one double each under the house rule above.
Say that comes to 620 espresso drinks last week and your measured dose is 18 grams. Here is the worked example, and you should run yours with your own two numbers:
| Step | Math | Result |
|---|---|---|
| Espresso drinks sold | 620 drinks | 620 |
| Times the dose | 620 x 18 g | 11,160 g |
| Into pounds | 11,160 / 453.6 | 24.6 lb |
| Into 5 lb bags | 24.6 / 5 | 4.9 bags |
Five bags, and you now know why five rather than "five feels right." More usefully, you know what moves it. Fifty extra drinks next week is 900 more grams, another two pounds, a bag every other week. A dose creeping from 18 to 19.5 grams because the grinder drifted is 930 grams a week you never decided to spend, close to two bags a month leaving the shop with nobody noticing.
Two adjustments before you trust the number. Add the shots that never ring up: dialing in the grinder each morning, the shot you pull and dump when the extraction runs long, staff drinks. Then add the drinks with more than one dose, if you sell any, and count them twice rather than once.
Drip and batch brew are the other half
Espresso is the number people quote. For most cafes it is not the largest one, because batch brew burns coffee by the liter and does it whether anyone is standing at the register or not.
The math is the same shape, with your brew recipe standing in for the dose. Say you brew at a ratio of one part coffee to sixteen parts water, and your batch is 64 ounces of finished coffee. That is roughly 1.9 liters of water and about 118 grams of coffee per batch. Brew six batches on a weekday and four on each weekend day, and the week runs 38 batches, roughly 4,500 grams, right around 10 pounds. Next to a 24 pound espresso week, that is nearly a third of your beans going into a pot rather than a portafilter.
The part that separates a good drip number from a bad one is what you sold against what you brewed. A batch is coffee whether you pour it or dump it at 3 PM, so brewed pounds and sold pounds are two different figures and the gap between them is the most fixable waste in the shop. If your hourly sales say the last two batches of the day pour half a pot between them, brewing to demand instead of to a timer takes real pounds out of the order. That gap has its own worked method, and drip is usually where it is widest.
Turn the weekly number into a bean order
A weekly total is not an order. Beans have two things milk does not: a lead time long enough to hurt and a freshness window short enough to punish an over-order. Ordering to your usage number alone gets you either a shortage on Thursday or four bags going stale in the back.
Three things turn the usage number into a quantity.
- The days the order has to cover. Not a week. The gap between this delivery and the next one, plus the days it takes to arrive. A roaster who ships Tuesday for Friday arrival needs an order sized to nine or ten days of pulling, not seven.
- Your day shape. Saturday does not pull like Tuesday. Split the espresso count by weekday from four weeks of sales and you will see which days carry the weight, and you will size the delivery to the days it actually spans rather than to an average nobody has.
- A floor you do not go under. The point where you place the order, not the point where you run out. That is what a par level is, and for beans it wants to be roughly one delivery cycle of usage plus a couple of days of cushion, because the emergency fix for beans is not a corner store run, it is being off your own coffee for three days.
The freshness ceiling caps it from the other side. Buying six weeks of beans to save on shipping means pulling month-old coffee in week five, and the cup tells on you before the spreadsheet does. Same logic as ordering matcha against its lead time: with anything that degrades, the right order is the smallest one that clears the gap, not the biggest one you can justify.
Do it once, then let the sales do it
Run the four numbers this week: dose, espresso drink count, brew ratio, batches per day. Two of them come from your Square sales, one comes from a scale, and one is your brew recipe. Fifteen minutes and you have a bean number that came off your own bar.
The reason to do it by hand once is that afterwards you can stop. Every espresso drink you sell is already a dose, recorded, timestamped, sitting in your sales history. Map the drink to its recipe and last week's beans compute themselves, the same way a register that only knows what sold turns into shelves that know what was used. What stays manual is the count that checks the math, and a full count of my own shop runs 58 items in about 8.5 minutes on a phone.
Do the espresso half tonight. Pull last week's espresso drink count, multiply by the dose you just weighed, and compare it to the bags you actually ordered. If those two numbers are more than a bag apart, you have found next week's order before you placed it.