On the Books

Covers on the books are the only forward-looking number a restaurant has, and almost every decision made before service is made against them. Prep quantities, ordering, and the roster all come out of the book, so the gap between what the book says and what walks through the door costs money in both directions.

How much of the book actually shows up?

Less than it says, and the shortfall is predictable enough to plan around. Three things shrink a book between the morning and the last seating. No-shows remove the party entirely. Late cancellations arrive inside the window where the table cannot realistically be resold. Party-size drift is the quietest of the three: a table booked for six that arrives as four is still counted as six in most reporting, so the covers figure overstates while the reservation count looks perfect.

The book understates in one direction too. Walk-ins are not in it, so a room at 120 on the books can serve 170. Read the book as a floor for guests and a ceiling for tables, and prep against the pattern of the last several comparable services rather than against tonight’s number alone.

Does the book tell me who is coming?

It tells you one name per booking. A six-top is one named guest and five covers nobody can identify, so a night that looks fully attributed in the book is usually half anonymous by the time checks close. The named booker’s history is also only as deep as that platform’s own record of them.

What a team actually wants before service is the shape of the room: which bookings are regulars, which are first visits, which carry a note or a preference somebody should act on, and which have a spend history worth knowing about. That view is built from a guest record assembled across systems, not from the booking record. It is why pre-shift reporting in a restaurant CRM starts from tonight’s book and then attaches everything known about the people in it.

Can I see one book across all my restaurants?

Not inside a reservation platform, once a group runs more than one. Each platform reports the venues that use it, so a group running Resy in two rooms and SevenRooms in a third has three books, three sets of guest records, and no combined view of the weekend. The same guest booking at two of those rooms appears twice, in two systems, with nothing connecting the records.

A layer that reads every book produces the group-level version, and it also makes the number comparable, since each platform counts covers, cancellations, and no-shows on slightly different rules.