Coffee Shop Business Model: The Unit Economics of a Specialty Café

A coffee shop business plan has one job: show that the shop can serve enough drinks, fast enough, at a margin that survives rent and wages. The beans are the least interesting line. A café is a capacity business, and the scarce thing is the barista's time at the busiest hour of the morning.

Plans usually start from the cup: a cost of goods, a price, a margin. That is the easy half. This post starts from the minute instead, because that is where specialty cafés quietly lose money.

Why are published coffee shop margins so far apart?

Because nobody is working from the same dataset. Two published sources show the problem.

A benchmark table from the roaster Bellwether Coffee (page dated 2026) sets targets of 65 to 75 per cent gross margin and 10 to 15 per cent net profit. A piece by the Australian roaster Pablo & Rusty's (February 2025) says that most industry reports show average café margins of around 2 to 3 per cent. Neither names a dataset, and both sell to cafés or serve them.

Both can be true. The first describes a well-run shop. The second describes an average one, after the costs a benchmark table tends to skip. So build the plan on the second number and treat the first as a goal. If your model only works at 15 per cent net, it does not yet work.

What does a cup actually earn?

Drinks carry very high gross margins, but the number is almost always before labour, so it flatters. Bellwether's benchmark lists gross margins of 87 per cent for espresso and Americano, 88 per cent for drip, 76 per cent for a latte, 80 per cent for a cappuccino and 73 per cent for a mocha. Food sits lower: 57 per cent for a breakfast sandwich and 54 per cent for a salad or lunch dish, and a pastry at 50 per cent bought in or 71 per cent baked in-house.

Read these as shape, not fact. Milk is the swing cost: the more milk in the cup, the lower the margin. The useful lessons are:

• A filter coffee is the cheapest drink to make and the cheapest to serve. It is a throughput weapon.
• Milk drinks earn less per cup but are what most guests want.
• Baking in-house raises food margin, but costs oven time and labour that the table above does not count.

For a deeper method on costing and pricing individual items, see food cost percentage and how to price a menu.

Where does the revenue go?

Three costs decide the outcome: product, people and premises. A sensible target for each, from the Bellwether table, is:

Add the midpoints and you have about 30 + 30 + 11.5, so roughly 71 to 72 per cent of revenue gone before utilities, card fees (Bellwether cites 2.5 to 3.5 per cent of sales), insurance, repairs, marketing and equipment finance. That leaves about 28 per cent to cover everything else and still pay you. It is easy to see how 2 to 3 per cent net becomes the average.

Rent deserves its own warning. The Pablo & Rusty's piece, written for Australia, notes that rents there rise by 4 per cent or inflation each year, whichever is higher. Check your own lease terms, but the pattern is common. Your best year for rent-to-sales ratio is year one. Plan for year four.

What is the one formula to learn?

Labour as a share of sales equals the hourly labour cost divided by (transactions per labour hour multiplied by average ticket). It ties the three numbers that actually move together.

Imagine a café where one hour of paid staff time costs 18 currency units, loaded with employer taxes and breaks.

• At 12 transactions per labour hour and an 8-unit average ticket, labour is 18 ÷ 96, about 19 per cent.
• At 10 transactions and a 6-unit ticket, it is 18 ÷ 60, which is 30 per cent.
• At 8 transactions and a 6-unit ticket, it is 18 ÷ 48, about 38 per cent.

Same wage, same shop, three very different businesses. Bellwether's benchmark ties this to targets of 12 or more transactions per labour hour, with the best shops around 15 and poor ones under 10 (their table puts labour at 25 to 28 per cent for the first and 35 per cent or more for the second). That guide comes from a supplier to cafés, but the logic holds in any currency: a small rise in average ticket or transactions per hour does more for the labour line than a pay freeze.

This is why a shop with a lovely room and a slow bar quietly fails. It is also why the average ticket matters more than the headline margin per cup. Target tickets in the Bellwether page are 7 to 10 units or more (in US dollars), so check against your own market.

How do you test the plan against the peak hour?

Test the peak hour, because that is the capacity ceiling. A café does not sell evenly. If a third of the day's sales land in two hours, the machine and the bar set a hard limit on revenue, however good the rest of the day looks.

Run this check before signing a lease:

1. Estimate the number of drinks one barista can make in an hour at your menu's complexity. Time it with a real recipe, not a guess.
2. Multiply by the number of bar positions your layout allows.
3. Multiply by your average ticket for drinks only, then by the peak hours per day.
4. Compare the result with the revenue you need to cover fixed costs.

If the ceiling is below the need, you have four fixes, in order of cost: shorten the menu at peak, move prep off the bar, add a second grinder or machine, or raise the average ticket with food and retail. Do not fix it with more seats.

The beans set the price of a cup. The bar sets how many cups you can sell.

Seen that way, a "slow" menu is a cost paid at the busiest hour of the day, whatever style reasons lie behind it.

Site choice is part of this arithmetic. A commuter corner with a short, intense peak needs speed and a tiny menu. A neighbourhood shop with a long, flat day can run a slower, more crafted menu and a higher ticket. Neither is better. They are two different businesses, and the plan should say which one you are building before it says anything about the logo.

What do rising costs mean for your model?

They mean you should model the whole cup, not just the bean. A World Coffee Portal report on the US branded market, dated 1 October 2025 and so a year old now, describes record green coffee costs, a 50 per cent tariff on Brazilian imports and persistent inflation. It also reports menu prices for 16oz filter, latte and cappuccino up about 3 per cent over twelve months, and iced coffee up nearly 5 per cent. Only 30 per cent of the industry leaders it surveyed reported positive trading conditions, down from the year before.

Two cautions. First, that covers branded chains in the United States. Second, the same site's UK independents report (published February 2026) counts 13,211 independent shops, up 3.3 per cent in twelve months, with higher cost and price exposure than chains. More shops chasing the same morning guests tightens things further.

Bellwether quotes commodity roasted coffee at about 6 to 8 dollars a pound and quality wholesale at 10 to 14. Take an 18 gram dose, which is about 0.04 of a pound: a 4-dollar gap per pound is roughly 16 cents a double shot (an estimate from those prices, not a Bellwether figure). A sudden rise in the bean price hurts. So does a slower bar: in the first labour example above, 10 per cent fewer transactions per labour hour lifts labour from about 19 to about 21 per cent of sales. Model both swings, and put your attention where your own numbers show the larger one.

How do you build the model in practice?

Use a simple, honest spreadsheet with four tabs and review it monthly.

1. Menu. Cost per item, price, gross margin, and time on the bar in seconds.
2. Day curve. Transactions by hour for a typical weekday and weekend.
3. Labour. Rota against the day curve, with loaded hourly cost.
4. Fixed costs. Rent with annual rise, utilities, card fees, insurance, finance and repairs.

Then run three cases: a base case, a case with 15 per cent fewer transactions, and a case with the coffee price up and menu prices unchanged. The break-even point is fixed costs divided by gross profit per transaction. If it needs more customers per hour than your bar can serve at peak, change the plan before you change the lease. For which numbers to watch each week once you open, see restaurant KPIs weekly, and if a bar or food programme is part of the concept, the café-bar hybrid piece covers where the day curve shifts.

What are the common mistakes, and what fixes them?

These four mistakes recur in café plans, and each has a plain fix.

1. Planning on the target margin. The plan only works at 15 per cent net. Fix: rebuild it at 5 per cent and see what has to change.
2. Counting seats, not minutes. The room holds 40 and the bar serves 30 drinks an hour. Fix: size the bar first and the seating second.
3. A menu that grows. Every added syrup, milk and brew method adds seconds at peak. Fix: for each new item, name the one it replaces, and time it.
4. Treating food as a bonus. Food has lower margins than espresso in the benchmark above, and it needs prep space and waste control. Fix: cost it in full, or sell a short, bought-in range and say so.

None of these is about coffee knowledge. They are about arithmetic done early.

Key takeaways

• A coffee shop business plan is a throughput model. Plan around transactions per labour hour and the peak hour.
• Labour share equals hourly labour cost divided by (transactions per hour times average ticket). Use it to test any menu or layout.
• Published margins range from about 2 to 3 per cent net (average) to 10 to 15 per cent (target). Build on the low number.
• Gross margin per cup is not margin per minute. Simple drinks protect the queue.
• Rent rises over time, so test year four, not just year one.
• Run three cases, including fewer customers and higher coffee costs, before you commit to a lease.