Insights and analytics
The analytical layer on top of the agent: why a person got a shift, how the load falls across weeks, who is drifting, where coverage is thin, and whether breaks are being taken.
What this layer is for
Core answers "write my week". This is the layer that answers "and tell me what is going on" — the difference between a schedule that exists and a schedule you can defend.
Everything here is derived from your own operational data. There is no benchmark against other companies and no demand forecast; the numbers are about your team, over the weeks you have run.
Why this person, on this shift
Any assigned shift can be asked to explain itself, and the answer cites sources rather than narrating: this person was available, holds the required certification, costs what they cost, and had a lighter load than the alternatives over recent weeks.
It is the single most useful thing to have when somebody challenges a rota — including when the person challenging it is you, three weeks later, wondering what you were thinking.
Fairness across weeks
The signals the agent already uses when drafting are available to read directly, over a six-week window: weekend load, late finishes, time over cap, and each person's share of the total hours.
The full dossier gives you the per-person breakdown — useful in a one-to-one, and useful when somebody is convinced they always get the closing shifts. Sometimes they are right, and it is better to know.
Reliability and attendance
- Reliability per member — how consistently they show up, confirm, and work what they were given.
- Attendance patterns — the drift you would otherwise notice six months late: the Mondays, the creeping late starts, the clustering of sick days around scheduled weekends.
- Swap reciprocity — who covers for others versus who only ever hands shifts away.
These are conversation-starters, not verdicts. Pair them with the member notes on someone's profile, which are manager-only and never visible to the person.
Coverage, cost and breaks
- Daypart coverage — where you are thin and where you are overstaffed, by part of day rather than by week total.
- Scheduling trends — how hours, cost and headcount have moved over time.
- Labour budget — set a target and see the week measured against it as you build.
- Meal-break compliance — whether the breaks your rules require are actually being taken, which is a different question from whether they were scheduled.
- Event staffing — ask for a suggested shape for a known busy day and get a proposal grounded in comparable past days.
Most of these are also reachable by asking the assistant in plain language, which is usually faster than finding the panel.
Getting the numbers out
Analytics you cannot export are analytics you cannot use in a payroll run or a board pack. See Payroll and exports for the payroll period export and the schedule export, and API and webhooks if you want the data on a schedule rather than on demand.
Common questions
- Do I need Pro for the AI to schedule fairly?
- No. The agent uses the cross-week fairness signals when it drafts on every paid plan. Pro is what lets you read and export them, and adds the per-shift explanations, the people signals and the proactive briefing.
- How far back does the analysis look?
- The fairness window is the previous six weeks. Trend analysis goes back as far as your data does.
Related
The AI scheduling agent
What the assistant can actually do, how a week gets drafted, why nothing is written without your say-so, and how to talk to it so it gets the week right first time.
Automatic drafts and the Sunday briefing
Three scheduled jobs run on Sunday so Monday is not a blank page: template recurrence, a seeded draft of next week, and a manager digest that reaches you in Slack or Teams.
Compliance rules
Five rule types, two severities, and one decision that matters more than the rest: which of your rules stop a publish and which only warn.
Payroll and exports
Export a pay period as CSV in the shape Gusto, QuickBooks or ADP expects, dump a week's shifts, or subscribe a personal calendar feed.