The few numbers worth reviewing monthly

The problem with restaurant reporting is rarely a shortage of data. It is that the numbers are defined differently each time they are calculated, arrive too late to change anything, and are read by nobody who can act on them.

## What makes a measure worth keeping

Three tests. If a number fails any of them, it belongs in an archive rather than in a monthly review.

**Someone can act on it.** A measure nobody owns produces discussion rather than decisions.

**It is defined the same way every time.** A number whose definition drifts cannot show a trend, and a trend is the only thing a monthly number is good for.

**It would change a decision.** If the result would not alter anything you do next month, it is interesting rather than useful.

## A workable set

The specific measures depend on the business, but most F&B operations are well served by something close to this.

**Revenue by daypart and channel.** Not total revenue — the split is where the information is.

**Contribution, not just food cost percentage.** The percentage is a ratio; the money is what the business banks.

**Theoretical versus actual food cost.** The single most diagnostic number in the kitchen, and the one most often skipped because it requires a stock count.

**Labour as a schedule against demand.** Labour percentage hides the question that matters: were the right number of people there at the right hour.

**Average transaction value by daypart.** The cheapest growth lever, and invisible unless tracked.

**Repeat rate, however roughly you can measure it.** Even an approximation changes how marketing spend is judged.

**Item mix movement.** Which items moved up and down the ranking, and why.

**One service quality measure.** Ticket time at peak, or complaint volume, or review sentiment — whichever your operation can capture consistently.

That is eight. Most venues can produce them from systems they already have.

## Cadence and format

Monthly is the right rhythm for most of these; weekly for anything that can be corrected inside the month, such as labour scheduling and ticket times.

The format matters more than most people expect. A report that requires interpretation before it can be discussed will not be read. Each number should appear with its previous period, its trend, and one line of written interpretation — not a chart that leaves the reader to infer what happened.

The written line is the part usually missing, and it is the part that turns a report into a decision.

## Thresholds beat targets

A target invites a discussion about whether it was fair. A threshold invites a decision.

"Food cost variance above three points triggers a stock investigation" is more useful than "food cost target 30 per cent", because it says what happens next. Setting thresholds in advance also removes the temptation to explain away a bad month after seeing it.

## Practical recommendations

1. Cut the reporting set to something readable in one sitting. If it takes longer, it will be skimmed.
2. Write the definition of each measure down once, and do not change it without noting that you did.
3. Attach an owner to every number — someone who can act, not someone who compiles it.
4. Add one line of interpretation per measure. The number says what; the line says so what.
5. Set thresholds before you see the results.
6. Review the set annually and remove anything that has not changed a decision in a year.

## In short

Measurement is not a data problem. It is an editing problem. The venues that manage well are usually not the ones tracking the most, but the ones that decided which few numbers matter and then read them the same way every month.