The numbers a multi-unit operator should see every week

These are the metrics that change decisions. Anything that would not alter what you do this week belongs in a monthly review, not on the dashboard.

• Prime costFood plus total labor as a percentage of sales. The single most important number in the business. Most full-service operations target the low sixties; quick service runs lower. Track it weekly, never monthly.
• Food cost percentageCost of goods over sales. Useful mainly against your own theoretical, not against industry benchmarks, because menu mix makes cross-brand comparison meaningless.
• Labor cost percentageTotal labor including taxes and benefits over sales. Split management from hourly; they move for entirely different reasons.
• Food varianceActual food cost minus theoretical. This is the waste, theft and portioning number, and it is where the recoverable money usually hides. Anything over one point deserves investigation.
• Sales versus last yearSame-store, same-period. The honest measure of whether the business is growing rather than just busy.
• Transaction countGuests, not dollars. Sales can rise on price alone while traffic falls, which is a slow problem that looks fine on a P&L for two quarters.
• Average checkSales divided by transactions. Read alongside transaction count; rising check with falling traffic means you are pricing out your base.
• Sales per labor hourThe productivity number. Tells you whether you are scheduling to demand and travels honestly across locations of different sizes.
• Hourly turnoverRolling twelve months by position. High turnover shows up in food cost and guest scores months before anyone connects the two.
• Guest score and volumeRating average and the number of reviews. The trend matters far more than the absolute figure.
• Off-premise shareCatering, delivery and takeout as a percentage of sales, tracked separately by channel because their margins differ enormously.
• Third-party marginWhat you actually keep after commission. Frequently a negative contribution and frequently unexamined.

A number in isolation means very little. Every metric on the dashboard should be shown against three references.

1. 01

   Last year, same period

   Controls for seasonality, which is the largest single distortion in restaurant data.

2. 02

   Your other locations

   The most useful comparison you have. Two stores running the same menu four points apart is a systems problem you can act on this week.

3. 03

   Your own theoretical

   What the model says the number should be. The gap between theoretical and actual is the operational opportunity, and it is specific enough to assign.

Budget is deliberately absent from that list. Comparing to budget tells you whether the forecast was accurate, not whether the operation is good.

The technology matters less than the discipline. We have seen excellent reporting run from a spreadsheet and expensive platforms nobody opens.

• One screenIf it scrolls, it is a report, not a dashboard. Twelve numbers, all visible at once.
• Weekly rhythmSame day, same time, every week. Monthly is too slow to correct anything; daily produces noise and anxiety.
• Managers see it firstGeneral managers should review their own numbers before their supervisor calls. Reporting used only to catch people becomes reporting people learn to game.
• Color with a reasonGreen and red should reflect defined thresholds, not a feeling. Define what red means for each metric in advance.
• Automate the collectionIf assembling the dashboard takes someone half a day, it will lapse within two months. Pull from the POS and back office directly.
• One owner per metricEvery number has a name against it. Numbers owned by everyone are corrected by no one.

The meeting is where a dashboard becomes management. Thirty minutes, same agenda, no exceptions.

1. 01

   Start with variance, not sales

   Sales is the number everyone wants to discuss and the one they can least control this week. Start with food variance and labor productivity.

2. 02

   Compare stores openly

   Put every location on the same screen. The best-performing manager explains what they are doing; this is the fastest transfer of practice available to you.

3. 03

   Pick two things

   Two actions for the week, each with a name and a number attached. Not a list of nine observations.

4. 04

   Check last week first

   Open by reviewing whether the previous two actions moved their numbers. Without this the meeting becomes a weekly recitation.

5. 05

   End on the guest

   Review score trend and any pattern in complaints. It keeps the meeting from becoming purely financial, which is how service quietly degrades.

Tracking food cost without theoretical. Knowing food cost is thirty-one percent tells you nothing without knowing it should be twenty-eight. The variance is the actionable number; the absolute is not.

Monthly reporting only. By the time a monthly P&L arrives, the period it describes is five weeks gone and nothing can be corrected. Weekly is the minimum useful cadence.

Averaging across locations. A group average hides both the store bleeding money and the one to learn from. Always show stores individually.

Measuring labor in dollars. Labor dollars rise with sales, which makes them uninformative. Percentage and sales per labor hour are the real measures.

Ignoring third-party contribution. Delivery volume looks like growth on a sales line and is frequently unprofitable once commission and packaging are counted honestly.