Demand Forecast

Staff the Week Your Business Will Actually Have.

Most rosters are copied from last week and corrected by memory. Shiftavo can do the arithmetic instead: you tell it what you count — covers, transactions, checked-in guests, parcels — and how much labor one of those takes. It learns the shape of your weeks from your own history, projects the next 54, and turns the projection into required hours per position, per quarter hour, per location. Turning it on is a switch; leaving it off changes nothing about how you plan today. Demand forecasting is part of Plus. The 30-day trial runs on Pro, so you can feed it your own history before deciding.

How It Works

Four Steps From Numbers to Hours.

01

Say what you count.

A driver is one measurable thing — covers per service, transactions per hour, occupied rooms per night. You set how finely it is counted, which locations it applies to, and what being short costs you compared with idling. That last number decides how defensively the forecast staffs, and the form tells you the resulting service level before you save.

02

Feed it the actuals.

Type them in, import a CSV, or have your POS or booking system push them over the API. Mark the days that lie — a closure, a street festival, a one-off promotion — so they distort the history instead of teaching it.

03

Set the labor standard.

How many minutes of a position one unit needs, plus a fixed setup, plus the smallest and largest crew that role ever runs with. Prefer thinking in units per hour? Type that instead and it converts.

04

Read the plan, override where you know better.

One row per day, per location: what the forecast expects, the required hours it implies, and the peak headcount. Type a day total you believe more and it is spread along that day's real shape — and it survives the nightly rebuild. Open a day to see the shape hour by hour, alongside what each candidate method would have answered and what moved the number.

What It Is and Isn't

Arithmetic You Can Check.

No black box

Every number can be traced back.

The forecast blends a handful of ordinary statistical methods and weights them by how each one has actually performed on your data. The day detail names what moved the number — the weekday pattern, the season, last year, a marker you set. There is no model you have to take on faith, and nothing is generated by a language model.

Honest about accuracy

It scores itself against doing nothing.

A Forecast vs. Actual report and a Demand Accuracy report compare every prediction with what happened, including against the naive baseline of assuming last week repeats. If the forecast is not beating that baseline for a driver, you will see it rather than guess it.

Off by default

Nothing changes until you switch it on.

Demand forecasting is a company setting. With it off, the coverage grid and automatic assignment use the fixed headcount ladder on each position, exactly as before. With it on, they use the forecast where a labor standard resolves and fall back to the ladder where none does.

Where It Shows Up

The Forecast Reaches the Places You Already Work In.

Coverage grid

Required becomes an evidenced number.

Where a labor standard resolves, the required figure in the coverage grid comes from the forecast instead of a static headcount you typed once. Quarter-hour buckets fold into the hour as their peak, because staffing has to satisfy the busiest quarter, not the average one.

Shift generation

The curve becomes draft shifts.

The week's requirement can be turned into draft shifts in one pass — from your own catalogue of shift shapes where you keep one. See Shift Profiles & Generation.

Automatic assignment

Headcount targets follow the forecast.

Each shift can target its share of the forecast for the hours it covers, with concurrent shifts dividing the number rather than each claiming all of it. The run report then words its gaps against the forecast — two of the three it asks for — instead of against a number nobody remembers setting.