Counter Ops / Staff scheduling

Rostering against the rush you are going to get, not the one you had last month.

Café staffing errors are expensive in both directions. Understaff the nine o'clock peak and the queue gets long enough that people leave. Overstaff a dead Tuesday afternoon and you have paid three people to restock a case that nobody is emptying.

Scheduling here reads the same demand forecast that sets bake quantities. If the model expects a heavy Friday morning, that shows up in the suggested cover as well as in how many croissants to bake.

Roster basis: Forecast rush profile
Granularity: Half-hour bands
Attendance: Clock in at the counter
Reporting: Sales per shift

Cover against demand

The roster and the bake list come from the same forecast.

Two decisions that are usually made separately by different people on different days, using the same underlying expectation of how busy tomorrow will be.

Suggested cover, Friday

Demand weight

07:00 - 09:00Opening and first rush
2 on counter, 1 on bake
09:00 - 11:30Peak weekday morning
3 on counter, 1 on bake
11:30 - 14:00Lunch, food-led
2 on counter, 1 on kitchen
14:00 - 16:30Quiet, prep window
1 on counter, 1 on prep
16:30 - 18:30Second rush and close
2 on counter

Suggested cover. You publish the roster, and any change you make is remembered as a preference rather than overwritten next week.

Shift / 09:00 - 11:30

On counter3 of 3

Queue held under 386% of band

Sales this shifton pace

Each shift carries its own sales and queue record, so cover decisions can be reviewed with evidence.

Why half-hour bands

A café rush does not respect a four-hour shift block. The difference between quarter past eight and quarter to nine is often the difference between two people coping and three being needed. Bands that narrow let you start someone at half eight rather than paying them from eight.

What it covers

Rostering, attendance and the shift record in one place.

Forecast-led shift suggestions
Suggested cover per half-hour band, derived from the demand profile for that weekday and that counter.
Availability and preferences held
Who can open, who cannot do closes, who is trained on the bake. Suggestions respect these rather than producing a roster you have to unpick.
Cover gaps flagged before publishing
A band with expected demand and nobody assigned is surfaced while the roster is still a draft.
Roles matter, not just headcount
Three people on shift is not cover if none of them can work the bake. Suggestions are role-aware.

Clock in and out at the counter
Attendance is recorded on the same terminal staff already use, with no separate device or app to chase.
Sales attributed per shift
What was sold, what the queue looked like, and how much waste was recorded during that shift.
Rostered against actual hours
Planned cover sits next to recorded attendance, so the gap between the two is visible rather than assumed.
Your overrides are learned
If you consistently add a fourth person to Saturday mornings, the suggestion stops fighting you about it.

A limit worth stating

This suggests rosters. It does not manage people.

Scheduling software has a habit of drifting into something else - productivity scoring, ranking staff against each other, using attendance data in ways nobody agreed to when they clocked in. We have kept this narrow on purpose.

What the system does is work out how much cover a band of time is likely to need, respect what your team can and cannot work, and record what actually happened during a shift. What it does not do is rate your staff, rank them, or make a recommendation about anybody in particular.

Shift-level sales exist so you can tell whether three people on a Saturday morning was the right call. They are a fact about a period of trading, not a performance verdict about the people who were standing there.

Questions

Scheduling, answered.

Does this handle payroll?

No. It records rostered and actual hours per person, and that record exports cleanly for whoever runs your payroll. We do not calculate pay, deductions or statutory contributions.

What if the forecast is wrong and we are short-staffed?

The queue timing on the counter shows it happening rather than after the fact, and that day feeds back into the forecast. Persistent under-suggestion for a band is something you can correct directly, and the correction sticks.

Can staff see their own rota?

Yes, published rosters are visible to the people on them, along with their own recorded hours. They cannot see other people’s attendance records or shift sales.

We run one counter with two people. Is this overkill?

Probably, on the rostering side. Independent cafés more often use the attendance record and the shift sales view and leave the suggestions alone. That is a reasonable way to use it.

Staff scheduling

Cover the nine o'clock peak properly and stop paying for the two o'clock lull.

The forecast that decides how many croissants to bake already knows how busy tomorrow morning is. Rostering from the same number is the obvious next step.

Suggested cover in half-hour bands, role-aware
Attendance recorded on the counter terminal you already use
Hours and shift sales exportable for payroll