Everything between the first tray out of the oven and the last cup of the day.
A café day has a shape that table-service and delivery software were not drawn around. Stock expires by evening. The queue is the constraint. The same faces return four mornings out of five, and tomorrow's production has to be decided tonight.
Counter Ops is the five areas that shape covers, and they are built to feed each other - what sold sets what gets baked, and what gets baked sets who needs to be on shift.
- Areas
- Five
- Screens to learn
- One, per role
- Counter hardware
- Use what you have
- Runs offline
- Order entry, yes
Four movements, one after the other, every trading day.
This is the whole product in one sentence: count what is in the case, keep the line moving, project tomorrow from today, and catch what is falling behind before it becomes waste.
Today's bake
Left / baked
- Butter croissant18 / 24
- Banana bread5 / 20
- Almond danish11 / 16
- Sourdough loaf9 / 12
- Flat whiteOn tap
Counter queue
~120s to serve
Watching the till for a returning face
Track / mid-morning
Forecast / tomorrow
Close / 17:40
What each part of Counter Ops is responsible for.
Order & POS
Take the order in as few taps as the order deserves
Quick keys for the lines that carry the morning, modifiers that do not bury themselves in submenus, split and part payments, and a queue timer that shows you when the line is slipping rather than telling you afterwards.
- Quick keys
- Split payments
- Queue timing
- Receipt printing
Daily bake inventory
Perishable stock counted properly, for once
Each baked line carries a same-day quantity that only falls, a low threshold you set, and an aging point. When something is behind its normal pace you hear about it with hours to spare, not at close.
- Live countdown
- Low alerts
- Aging flags
- Discount prompts
Loyalty & regulars
The order they always place, already on screen
Repeat customers are matched at the till and their established order staged for confirmation. Rewards accrue against actual repeat behaviour instead of a punch card nobody remembers to stamp.
- Usual order
- Visit history
- Rewards
- Group-wide identity
Staff scheduling
Rosters built against the rush you will actually get
Shifts are planned on the same forecast that sets bake quantities, so cover matches demand instead of matching last month. Attendance and per-shift sales sit alongside the roster.
- Forecast-led rosters
- Attendance
- Shift sales
- Cover gaps
AI demand forecasting
Tomorrow, worked out line by line
Every baked item is projected on its own curve from its own history, the weekday, and comparable trading days. You get a prep list to confirm or override - never an automatic change to what goes in the oven.
- Per-item numbers
- Weekday patterns
- Confidence shown
- Prep list
A general till can sell a croissant. It cannot reason about one.
Restaurant software is largely organised around a table: seat it, take a course order, hold the kitchen, turn the cover. Delivery software is organised around a rider and several marketplaces. Both are legitimate shapes. Neither describes a café.
At a counter the constraint is the queue, and the inventory is hostile - it exists for one day and then it is a loss. Recognition matters more than it does anywhere else, because the same twenty people account for a disproportionate share of a café's month.
So Counter Ops treats a same-day tray as a first-class object with a life span, treats the queue as the thing to optimise rather than cover turns, and treats a returning customer as something the system should already know.
CAPPATERY is deliberately narrow. It covers one operational shape, the café and bakery counter, because that day does not fit software built around table service or delivery dispatch. Cafés and bakeries got their own build rather than a configuration flag.
Practical answers about running this on a real counter.
Do we have to adopt all five areas at once?
What happens to order entry if the internet drops?
How long before the forecast is worth trusting?
Can staff override what the system suggests?
Start where it pays back fastest
Put your bake lines in and the rest of Counter Ops has something to work with.
Order entry and bake counting produce the history that forecasting, recognition and rostering all read from. That is the sensible first week.
- Bake lines and quick keys mapped to how your case is actually laid out
- Existing terminal and receipt printer kept where they are compatible
- Forecasting switched on once there is history worth forecasting from