How do you forecast restaurant covers?

It does not need to be exact

A restaurant forecast is not trying to predict revenue to the cent. It needs to tell you whether tomorrow is a quiet service or a big one, because those need different prep, different staff and different orders.

Here is an example. A dinner forecast of 120 covers and one of 220 covers lead to two different kitchens. Getting within a band is enough to set the day up well.

What goes into it

Start with history: covers by daypart for the same day of the week over the last several weeks, and the same week last year if you have it. Then add what you already know: bookings on the books, large parties, private events.

Then adjust for what changes demand near you: a game or concert, a public holiday, school holidays, rain on a patio night, a promotion you are running. Write down why you moved the number, so you can check later whether the reason held.

Split it by daypart

One number for the whole day hides the shape of it. A busy lunch and a slow dinner need different people at different times. Forecast lunch, dinner and any other service you run on their own.

Make it get better

Keep forecast and actual side by side, every day, for every location. Once a week, look at where the gap was largest and why. Often it is one daypart, one location or one kind of day that keeps being missed.

Then use it. Prep, the rota, cut times and orders should all start from this number. A forecast that nobody plans from does not get better, because nobody notices when it is wrong.