Before you count: pick the period and the unit
Compare like with like. A Tuesday lunch and a Saturday dinner do not belong in the same average: group your numbers by type of service (weekday lunch, weekday dinner, weekend) and by month.
Also note what skews the picture: a closure, building work, a heatwave, an event in the street. A number without context misleads more than it informs.
Your sources are simple: the booking system for covers, no-shows and channels, the till for revenue. If you keep a paper diary, count four representative weeks rather than the whole year.
Covers per service
The base number: how many people did you serve, on average, per type of service? Read it over time (this month against the same month last year) and across services (Tuesday lunch against Thursday lunch).
What it shows:
- weak services, which cost almost as much to open as a full one;
- seasonality, to plan buying and rotas;
- the effect of a change, a new menu, new opening hours.
This number feeds two weekly decisions straight away: ordering stock from your bookings to cut food waste, and building your staff rota around your bookings.
With gonnabook.it, the calendar shows the covers booked each day and the month’s total, which gives you a first reading without any maths.
Occupancy
Occupancy compares the covers you served with the seats you could offer.
| Number | Formula | What it measures |
|---|---|---|
| Covers per service | Covers served ÷ number of services | Volume |
| Occupancy | Covers served ÷ (seats × services) | How well the room is used |
| No-show rate | No-shows ÷ expected bookings | How reliable bookings are |
| Online share | Online bookings ÷ all bookings | The weight of each channel |
| RevPASH | Revenue ÷ (seats × opening hours) | What one seat earns per hour |
A made-up example: a 40-seat room, open for one dinner service, serves 30 covers. Occupancy for that service is 75%. If tables turn, it can go above 100%, and that is fine: a seat was used twice.
Watch the table trap: four people at a table for six fill 100% of your tables but 67% of your seats. Look at both if your tables are large. Our guide to table turns and service durations explains how the two numbers work together.
No-show rate
The no-show rate compares bookings that did not turn up, without warning, with the bookings you expected. Leave cancellations out: a cancellation, even a late one, gives you a chance to resell the table. A no-show does not.
Read it by service and by party size. A rate that climbs on Saturday nights or for tables of six calls for a different fix than a flat rate. Our guide to reducing restaurant no-shows lists the levers for what you see: a reminder the day before, written terms, a card guarantee on your busiest nights.
Online share
Where do your bookings come from? Online, by phone, in person, walk-ins? The split tells you two things:
- time on the phone: every booking taken by hand interrupts the team;
- which channels work: your website, your Google profile, social media, or none of them.
If the online share is low, the cause is often simple: the booking link is hard to find. Check where it appears before looking any further.
RevPASH
RevPASH (revenue per available seat-hour) is revenue divided by the number of seats multiplied by opening hours. It was proposed by researchers at Cornell University to bring hotel-style capacity management to restaurants.
Another made-up example: 40 seats, a four-hour dinner service, 1,600 in takings. RevPASH is 1,600 ÷ (40 × 4) = 10 per seat per hour, in your currency.
Why it helps: it combines occupancy, table time and average spend. A service can be full and still earn little if tables stay three hours for one course. Comparing RevPASH hour by hour shows the dips, often early in service, and the slots where you turn people away.
How to read them together
No number works alone. A few useful combinations:
- Low occupancy and low RevPASH on a service: a service to fill, or to close. Our article on filling weeknights has ideas.
- High occupancy and average RevPASH: tables stay a long time; table times and slots deserve a look.
- No-shows concentrated on one night: a targeted guarantee rather than a blanket one.
- Low online share and a ringing phone: the booking link is not visible enough.
What to change in January
A review only matters if it leads to a decision. Pick two or three changes, not ten, and write down the number that will tell you whether each worked:
| Finding | Possible action | Number to watch |
|---|---|---|
| A service that is always weak | New offer, new hours, or close it | Covers per service |
| Tables for six used by four | Rework the plan and combinable tables | Seat occupancy |
| Too many Saturday no-shows | Reminder the day before, guarantee that night | Saturday no-show rate |
| Few online bookings | Link on your Google profile and website | Online share |
Run the same numbers again at the end of March. Three months is enough to see whether a decision holds.
Frequently asked questions about restaurant numbers
How do I calculate restaurant occupancy?
Divide the number of covers served by the number of seats multiplied by the number of services in the period. A 40-seat room that serves 30 covers in one service is at 75% occupancy. If tables turn during service, occupancy can exceed 100%, since the same seat is used more than once.
What is RevPASH?
RevPASH, revenue per available seat-hour, is revenue divided by the number of seats multiplied by opening hours. Proposed by researchers at Cornell University, it combines occupancy, table time and average spend, and shows the hours when a seat earns little or, on the contrary, when you are turning guests away.
How do I calculate the no-show rate?
Divide the number of bookings that did not turn up, without warning, by the number of bookings expected in the same period. Leave out cancellations, even late ones, since they still give you a chance to resell the table. Then read the rate by service and party size, to aim your fixes where they matter.
Sources
- Sheryl E. Kimes et al., Restaurant Revenue Management: Applying Yield Management to the Restaurant Industry, Cornell Hotel and Restaurant Administration Quarterly, 1998, Cornell University (eCommons), definition of RevPASH.
- Sheryl E. Kimes and Jeannette Ho, Implementing Revenue Management in Your Restaurants: A Case Study with Fairmont Raffles Hotels International, Cornell Center for Hospitality Research, 2019 (measure the starting point before changing settings).
- The figures in the examples (a 40-seat room, 30 covers, 1,600 over four hours) are illustrations, not statistics.