The Forecast Join

At the top of the chain, a business forecast becomes a number of people. The translation is a single assumption and it is usually orphaned.

What crosses

Sales, covers, transactions or volume, by period.

Converted to a labour requirement by a ratio: covers per server, transactions per hour, units per picker.

Which produces the requirement scheduling works from.

Why the ratio is the weak point

It was set once, often years ago.

By someone who has left, with reasoning that was not recorded.

Under different conditions: a different menu, layout, till system or product mix.

And it is not owned. Operations assumes it came from finance; finance assumes it came from operations.

Meanwhile it determines every staffing number in the business.

Testing it

Compare the ratio's prediction against outcomes rather than against the forecast.

Where the ratio says four people and shifts consistently run badly with four, the ratio is wrong.

Ask the people working it. They know within ten minutes of a shift starting, and nobody asks them.

Review it annually with a named owner, which is an hour and is the highest-leverage hour in this subject.

The granularity problem

Forecasting by the hour and staffing in four-hour blocks discards most of the precision.

Match the forecast granularity to the scheduling granularity and spend the effort saved on the ratio.

A forecast more precise than the shift structure produces confidence rather than accuracy.

Feedback that does not happen

The loop should close: actual demand and actual outcomes feed back into both the forecast and the ratio.

In most operations the forecast is compared against actual sales and the ratio is never revisited at all.

Which means half the model is maintained and half is frozen.

What to measure

Forecast error, by site and day part — bias and scatter separately, because they mean different things.

And ratio accuracy: shifts where the requirement was met and the shift still ran badly, or was overstaffed and quiet.

The second measure barely exists anywhere, and producing it is usually what starts the conversation about the ratio.

Give the ratio an owner

The single highest-leverage hour in this subject.

Covers per server, transactions per hour, units per picker.

Set once, by someone who left, under conditions that have changed.

Name an owner and review it annually against outcomes rather than against the forecast.

Because it determines every staffing number in the business and currently belongs to nobody.

Connect policy and data

The choices in this note can be compared with workforce optimisation tools. Keep the written purpose in control and enable only the information needed at this boundary.