Skip to content

Service manager

The service manager workspace answers one question continuously: is service going well right now, and if not, where?

Tab What it holds
Overview The state of service at this moment
Attention Things that need a decision, ranked
Team Who is working, and what they have been doing
Quality Signals that service is slipping
The service manager's overview of the branch right now.
Service right now.
The service manager's attention queue.
What needs a decision.

A small set of numbers, chosen because each one implies an action. Seven cells, and every one of them opens the queue it is about:

Signal What it tells you
Ready waiting Food made and not taken out — waiter capacity
Partial serve Orders where some of the food went out and some did not
Payment Orders blocked on payment — cashier capacity
Released Sessions with unresolved follow-up
Kitchen late Tickets past their time — kitchen capacity
Incidents Open service incidents
Urgent unread Conversations nobody has answered

Over the grid sits one word for service as a whole: Calm, Busy, High pressure or Critical. It is the branch’s service health, not any one department’s — there is no “kitchen load” summary anywhere in the workspace, and Overloaded / Normal / Idle belongs to a different screen: it is the Team tab’s filter over individual waiters.

Read the cells together. Ready-waiting high while nothing is late in the kitchen is a floor problem, not a kitchen one — and the fix is a waiter, not a cook.

The queue of things that need you, ranked, each with the responsible role and a suggested action.

Types you will see:

  • Kitchen delayed
  • Guest complaint
  • Override request
  • Unanswered conversation
  • Incident
  • Staff issue
  • Robot fallback

Severity runs Critical, High, Medium, Low. Work critical first — that is the whole point of the ranking.

Suggested action is guidance, not an instruction. You know the room.

Patterns rather than single events — the things that mean service is degrading even when no one has complained yet:

  • Long ready-to-serve delay
  • Long pending-payment delay
  • Long kitchen delay
  • Repeated waiter no-response
  • Active complaint conversation
  • Cancelled order
  • Unresolved released session
  • Service recovery incident

Each card carries two cells: Recovery owner — the roles involved — and Suggested recovery, the next move Sitora would make. These are for conversations after service, not mid-rush firefighting.

Waiters, and only waiters. This is floor coverage — who is on, what they have been doing, and how their work is distributed — not a roster of everybody in the branch. Cooks, cashiers and couriers are not here.

Each waiter carries a load state, and the filter above the list is those states: All, Overloaded, Normal, Idle, each with its count. This is the one place those three words appear.

Two things to keep in mind:

  • It is derived from real operational records, not from a clock-in system. It reflects what people actually did in the app.
  • It is not a payroll ledger. Earnings figures are estimates. Useful for a conversation about workload; not a wage slip.

Opening an incident shows a fixed grid of six facts — Assigned role, Created by, Opened, Last activity, Messages, Resolved — and the conversation underneath. You record what you decided and why; the reason is part of the record.

It does not suggest a recovery. That is a Quality-card cell, not an incident one — if you want Sitora’s suggestion, it is on the Quality tab.

Released session review is the manager’s view of sessions the floor could not close. Use it when a session has been sitting too long, or when a waiter needs a decision above their level.

The only thing a manager can pause is the robot fleet, and only where a branch runs one. The fleet strip on the overview shows the vendor, whether the fleet is online, and — when the backend allows the verb — Pause and Resume. It is a real consequence, not a display toggle, so it asks for a reason and asks you to hold.

What it stops is narrow, and the sheet says so: New robot offers stop immediately for this branch. Robots already on a task finish it. Staff serve flows are unchanged. Guests keep ordering, the kitchen keeps cooking, and waiters keep serving.

Pausing order intake is not a manager verb. Stopping the branch from taking new orders belongs to the branch admin — see Branch admin. A manager who needs intake stopped asks one.

Ready-waiting is climbing. Food is being made and not taken out. Put someone on the floor, or take it out yourself. Check Released too — some of it may belong to guests who left.

Kitchen is overloaded. Expect delays and get ahead of them: talk to the floor about what to tell guests. If it is unsustainable, intake has to stop — ask a branch admin, because you cannot stop it yourself.

Urgent unread is not zero. Someone is waiting for an answer. Unanswered conversations turn into complaints — this is the cheapest problem on the screen to fix.

A guest complaint arrives. Open it from Attention, act, and record what you decided. The record is what makes the same complaint next week answerable.

A robot fell back. A serving robot could not complete a leg. Someone has to carry it. Handle the food first, then look at whether it keeps happening.

Someone’s access needs changing. That is a branch admin action, not a manager one. See Branch admin.

  • Confirm payments. Cashier only.
  • Edit history. Corrections are new records.
  • Change staff access. Branch admin or owner.
  • Change plans or subscriptions. Owner, in Sitora Biz.