Your First 90 Days Running a Cafe, by the Numbers

Month one tells you almost nothing, and that is fine

Two weeks in, you will have a theory about your cafe. Saturdays are the big day. The 9 AM rush is the rush. Oat milk is running away from you. Everyone loves the matcha.

Some of that will turn out to be true. The part that catches new owners is not that their instincts are wrong, it is how confident two weeks of data can make a wrong instinct feel. Fourteen days is fourteen observations, and roughly two of each weekday. A single rainy Tuesday, one catering pickup, one week where the office up the street was closed, and your "pattern" is one event wearing a trend's clothing.

So month one has a narrower job than most advice admits. You are not looking for patterns yet. You are building the record that makes patterns visible in month two, and you are fixing the things that are obviously broken regardless of what the data eventually says.

Here is what month one is actually for:

• Get your catalog honest. Every drink you sell should exist as a real item in the register with its modifiers set up properly, because a sale rung up as "coffee, other" is data you can never recover. This is the single highest-value hour of admin in your first month.
• Start counting on a cadence. Not because the counts will tell you anything yet, but because count number twelve is only useful if counts one through eleven exist. Start here if you have not built the habit.
• Write down your recipes. How many grams in a shot, how many ounces of milk in a 12 and a 16. You will change these, and you want to know what they were when you did.
• Watch your cash, not your averages. In month one the useful number is what is in the account, not what your average ticket is doing.

The one thing worth watching daily is what you ran out of. Every stockout in month one is a par level you have not set yet.

Month two is when the week takes shape

Around day 45 you cross a threshold: roughly six to eight observations of every weekday. That is not statistical certainty, but it is enough to separate a real shape from a loud memory.

The first honest read you get is the shape of your week. Not "Saturday is busy," which everyone believes on day three, but where Saturday actually sits against Thursday, and whether the gap is in transactions or in ticket size. Those are different businesses. A day with fewer customers spending more needs different staffing and different prep than a day with a crowd buying drip. Sales patterns by day of the week is the read to run here, and it is the first place your ordering should follow the data instead of your calendar.

The second read is your real peak hour, by day. Owners are consistently wrong about this, and the reason is honest: a shift feels busy when you were understaffed, not when revenue was highest. Weekday peaks tend to sit early with a commuter crowd; weekend peaks push later. If your Saturday peak is actually 10:30 to 12:30, your opener should have batch drinks ready at 10:15, not at 8 AM when the pressure feels worst.

The third read is the first one that costs you money to ignore: the gap between what your recipes say you used and what your counts say you used. If your sales math says 600 ounces of oat milk a day and your counts say 800, that 200 ounce gap is waste, over-pouring, or drinks leaving without being rung up. Two months of counts is the minimum to see it, and it is the reason month one's boring cadence mattered. This is the mismatch to chase, and it usually pays for the effort several times over.

Month three is when you are allowed to decide things

By day 90 you have twelve to thirteen observations of each weekday and three points per monthly metric. Three points is the first time you can talk about direction rather than position: revenue climbing, oat milk share growing, waste falling since you fixed portioning.

This is the month to make the decisions you have been deferring.

Set par levels from data, not gut. You now know your daily usage per item and your suppliers' real lead times. That is everything a par level needs. How to set the reorder points is the mechanics; the input is your own three months.

Cost your top ten drinks properly. Not estimated, itemized: beans, milk, cup, lid, syrup. Do the ten that carry your volume and you will have covered most of your cost of goods. Work one recipe end to end, then repeat it nine times.

Cut or fix the bottom of the menu. After 90 days you can see which items sell in single digits per week. Some of those are worth keeping because they cost you nothing to carry. The ones that require a dedicated ingredient with a shelf life are the expensive kind of loyalty.

Now reprice, if the math says so. With three months of real costs behind you, a price change is a decision instead of a flinch.

What you still cannot do at day 90 is seasonality. You have one season. The iced shift, the holiday swing, the back-to-school move, none of it is visible yet, and treating a summer trend as your permanent baseline is the classic month-four mistake. Note the shape and wait for the calendar to come back around.

The three numbers to check every week from day one

Long before the data is rich enough for analysis, three numbers are worth a weekly look, and they stay worth it forever.

Sales versus the same day last week. Not versus yesterday, which mostly measures which day of the week it is. Same weekday, week over week, is the comparison that means something in a business with a strong weekly rhythm.

What you ran out of. Every stockout is either a par level that is too low or an order that missed a cutoff. Both are fixable, and both get expensive when they become normal. Ordering from your count instead of your memory is the fix for most of them.

What you threw away. Milk poured out, pastries that did not sell, a batch of cold brew that turned. Write down the item and the amount, nothing fancier. Ninety days of that list is a better waste report than any software will give you in month one.

Three numbers, ten minutes, once a week. Do that from your first week and by day 90 you will have something most new owners do not: a record that can be questioned. What the first 30 days reveals goes deeper on the review itself.

The advice you will get most often in your first months is to trust your instincts. Your instincts are worth trusting about coffee, about hiring, about how the room should feel. About usage and cost and what to order Tuesday, they are guessing, and 90 days of counts is what replaces the guess. Start the record now; the reading comes later.