Back to Blog

The delicate balance of scheduling

An interactive cafe payroll model for testing when a second or third barista is worth it, why one-hour rushes are so awkward, and how random daily traffic changes the math.

5 min readSean Dokko

CafeTappy case study

You made it through year one. Now the schedule has to prove year two can work.

Imagine you run a coffee shop called CafeTappy. You are a year in. The first year did not crush you, which is already something. But it did not make you feel safe, either.

You understand seasonality now. You know the school calendar, weather, office routines, and weird local events can make the same cafe feel healthy one week and fragile the next. The question you are trying to understand is not simply, "Can I afford another barista?" It is, "How do I schedule enough help for the rush without letting payroll quietly become the thing that breaks the business?"

Operating window

6a-2p

The model below treats CafeTappy as a cafe with an 8-hour day, one baseline person, and helper shifts placed around the moments where sales justify more hands.

Interactive operating model

The staffing rule

The simple rule here is one person per roughly $60 of hourly sales capacity. Below $60, you can usually get by with one person. A second person starts to make sense around $120 per hour. A third person needs about $180 per hour behind it.

Second person needs

10.9 tickets/hr

At $11 per ticket, the two-person line is $120 per hour.

Peak staffing

1 person

8a is modeled at $98 per hour, or 8.9 tickets.

Planned labor

12 hrs

Includes the starter bar helper. Threshold-only staffing would be 8 hrs.

Modeled day

$0

After 41.2% COGS and the planned schedule.

Commentary: the slider that matters most is not always wage. It is volume density. At this curve, the busiest hour crosses the second-person line around $600 in daily sales and the third-person line around $900.

Smart scheduling

Turning an 8-hour cafe day into actual shifts

Smart scheduling starts with the sales curve, then translates it into shifts a person would actually work. CafeTappy is open from 6a to 2p. The first question is not the shift. It is the configuration. A lead barista scheduled alone is a 1-person shop. A lead from 6a to 2p plus a 4-hour bar person is already a different operating model.

Barista configurations

These are not just headcount levels. They are different ways of running the cafe. Super busy cafes can move beyond this into five-person event or destination days, but the fourth person is already a serious signal that the shop understands its volume.

1-person shop

1 person

Owner-driven and very slow

CounterBarLight food prep

One person does counter, bar, and light food prep. This is usually a super slow, owner-driven shop that not many people know about yet.

2-person shop

2 people

The owner has to delegate

Counter and light food prepBar

One person handles counter and light food prep. The other person owns the bar. This is where the owner has to stop being the only engine in the shop.

3-person shop

3 people

A strong shop with peak staffing

CounterDedicated barDedicated food

One person handles counter, one person stays on bar, and one person owns food. This is often a 2-person shop that staffs up on peak days.

4-person shop

4 people

High-volume and intentional

CounterMilk steamerShot pullerBOH

One person handles counter, one person steams milk, one pulls shots, and one works BOH. At this level, you need to understand your volume clearly.

Starter 2-person day

Lead plus a 4-hour bar shift

12 paid hours, $28858.8% of forecasted sales

Floor lead

6a-2p

Counter, handoff, light food prep, and keeping the day moving.

Bar support

8a-12p

A 4-hour bar shift for the morning rush, driven by the minimum helper-shift slider.

Live modeled shift plan

What the current sliders justify

The model below still follows the forecast. If the sales curve does not cross the second-person threshold, it will show a true 1-person shop. When the forecast rises, it adds bar, food, and overflow support as the peak gets denser.

Lead barista

6a-2p

8 paid hours

keeps the cafe open and covers the baseline.

Commentary: this is the scheduling complexity the model is trying to reveal. The one-person shop is not the same business as the 2-person shop, and the 4-person shop is a different thing again. The rush may only last one or two hours, but the schedule has to translate that demand into real roles and humane shifts.

Peak curve

The peak curve

Cafe traffic usually concentrates around the morning commute window, often 7-10 a.m., with another smaller lunch bump. This chart applies that shape to CafeTappy's modeled average monthly gross receipts: $12,650 per month, or about $490 per open day across 26 days.

Thresholds

1 person: under $120 / hr

2 people: $120 / hr

3 people: $180 / hr

$120 / hr
$180 / hr
$34

6a

1p

$69

7a

1p

$98

8a

1p

$88

9a

1p

$64

10a

1p

$39

11a

1p

$59

12p

1p

$39

1p

1p

Morning concentration is doing the work. Your highest hour is $98.
Extra helper demand appears for 0 modeled hours. The schedule pays for at least 0.
A third person needs roughly 16.4 tickets per hour at your current ticket size.

Employee happiness

The shift problem

Commentary: the graph is why payroll feels so weird in a cafe. Peak periods can justify help for an hour or two, but people are not puzzle pieces. A two-hour shift may make the spreadsheet happy, but it is a silly thing to offer a real employee. That is why the minimum-shift slider turns a short rush into a larger payroll commitment.

Random day demo

Simulate the next day

The operator plans labor before the day happens, optimized around the forecast to reduce wasted payroll while still covering the expected rush. After that, the schedule is hard to change, so each click keeps labor fixed and reveals what traffic did.

Disclaimer: this is just a simulation. It is designed to show the scheduling tradeoff, not predict actual sales or prescribe payroll.
Current CafeTappy: About $500 in daily sales. The shop is alive, but there is not much room for a wrong guess.Planned for about 45 drinks, $495 in sales, 8 labor hours, $2,800 monthly rent, and peak staffing of 1 person.
4 / 7: Stable
1 / 7: Amazing day
1 / 7: Bad
1 / 7: Tragic event
Planned labor is fixed at $192 for the next simulated day. That is 38.8% of this scenario's forecasted sales target. Stable days are shaped to keep labor around 20-40% of sales, while tragic days land around 50-60%. The operator does not know which kind of day is coming. Rent hits every 30th simulated day and subtracts $2,800 from the balance.

Accrued balance

$0

Click the button to start the simulation.

DaySalesLabor costDay resultBalance

No simulated days yet.

Mini day result graph

Each bar is one simulated day after sales, product cost, scheduled labor, and rent. Rent days use the rent color because the fixed monthly bill can overwhelm an otherwise stable day.

$0

Simulate days to draw the day-by-day results.
Stable Amazing Bad Tragic Rent day

Commentary: this is the emotional trap of scheduling. The employee schedule is decided before the day happens, and the best version of the plan is already trying to keep labor lean. Revenue is discovered after customers either show up or do not. The owner is stuck balancing customer wait times, employee happiness, and the hard cash cost of guessing wrong.

Model note

What the model is really saying

This is not a payroll prescription. It is a visibility tool. CafeTappy's model uses $151,796 in annual gross receipts, $62,556 in COGS, and $11,133 in average monthly wages before the extra payroll-tax line.

The peak curve reflects a typical cafe rhythm: a morning rush, a smaller lunch bump, and softer traffic later in the day. The point is not to declare one universal schedule. The point is to make the pressure visible before payroll is committed.

Conclusion

What CafeTappy is really trying to understand

The problem is not that cafe scheduling is mathematically complicated. The problem is that the schedule has to be decided before the day reveals itself. Demand is uneven, the rush is compressed, and employee-friendly shifts are larger than the rush itself.

This model sheds light on that gap by separating three things that are usually tangled together: the hourly sales curve, the smart shift plan, and the random day that actually happens. Once those are visible, the tradeoff becomes clearer. CafeTappy is not just deciding whether another barista is affordable. It is deciding how much uncertainty the business can carry while still being fair to the people working there.

Want to try CafeTally early?

We are looking for a small group of cafe owners who want hands-on setup and are willing to give honest feedback.

Keep reading

View all posts