Restaurant Reservations: Reconcile Attendance and No-shows

Fix the service and the roster first

Name the restaurant, the service date and its timezone, the booking window you are capturing and the cutoff time for the closing review. Give the review a version number so a later correction can be traced back to it.

The roster is every genuine reservation booked for that service within the capture window, including bookings cancelled earlier. Removing a cancelled or awkward booking makes the remaining picture look cleaner than it was. Duplicates, test bookings and reservations moved to another service are handled through a written scope note and change log, not by quietly deleting a row. Use anonymous booking references in the shared worksheet and keep guest names and contact details in the reservation system.

Give each reservation one final outcome

For this review, a reservation has one of five outcomes. Attended means at least one guest from the party was actually seated. Cancelled means a cancellation is recorded. No-show means nonattendance was verified under the restaurant’s own written rule. Other verified records a known outcome outside those categories, with an explanation; Unresolved means the outcome has not yet been checked.

These are deliberately simple review categories. A reservation system may use many more statuses during service, such as booked, confirmed, running late or arrived. Write the definitions down before service begins so that the closing review applies them rather than inventing them.

Treat unclear arrivals as questions

Some cases need a person to check the evidence. A party that arrived but was never seated, a very late arrival or a record with missing details should stay Unresolved while that check is open. A verified arrival without seating can then be recorded as Other verified, with zero seated guests and the reason. It is neither missing evidence nor automatically a guest no-show.

Record who resolved each case, what they checked and when. A party seated after arriving late is Attended. Define how late cancellations and other exceptions are classified before reviewing the results, and record any later policy change with its effective date rather than silently reinterpreting past outcomes.

Worked example: ten reservations for one dinner

In this fictional dinner service, the fixed roster holds ten reservations. At the first review six are Attended, two are Cancelled, one is a No-show and one, R10, is Unresolved; none is Other verified. Nine of ten outcomes are verified, so coverage is 90%, and the no-show share is withheld because one outcome is still unknown.

A follow-up check confirms that R10 was seated. The final count is seven Attended, two Cancelled, one No-show and zero Other verified, which adds back to ten. Coverage is now 100%, and the no-show share is 1 of 10 reservations, or 10%. This local denominator includes every reservation in the fixed cohort, including cancellations and other verified outcomes. It is a declared definition, not an industry benchmark.

Count guests separately from reservations

Reservations and guests are different units, and so are tables. Keep the planned party size beside the actual number of guests seated. At the first review the six known Attended bookings account for 14 seated guests; R10 later adds 3, giving 17. The planned total of 26 is not the number of people served.

A partly attended booking stays one Attended reservation. R02 booked four and three were seated: record 3 actual guests, and do not create an extra no-show for the missing person. If more guests arrive than planned and staff verify it, record the larger actual number and note the change. Count each party once.

Keep walk-ins in their own register

In this review, a walk-in is a party seated without an existing reservation. A booked party remains in the reservation roster even if a host used a walk-in action to seat it; reconcile the records so the same party is counted once. In the fictional example two genuine walk-in parties seated 5 guests. The completed total is 17 reservation guests plus 5 walk-in guests, or 22. Walk-ins stay outside the reservation denominator, and none of these counts estimates lost revenue.

Before reporting a final rate or guest total, the reviewer also confirms that the reservation roster and walk-in register are complete against their source records. A blank walk-in register is not evidence that nobody walked in. A reviewed, genuinely empty reservation cohort has no no-show percentage, but a complete walk-in record can still establish seated guest totals.

Correct the source, not just the worksheet

If the closing review finds a wrong status, correct it in the reservation system through the process your team is authorised to use, and keep the earlier version and the reason in the change log. The worksheet records the review; it is not the place to contact guests or change bookings.

Keep exceptions and agreed actions beside the outcomes. A pattern of unclear arrivals may point to a host-stand step that needs a clearer record, not to the guests.

What the reservation-system reference supports

OpenTable's support article on managing reservations separates statuses before a visit, such as booked and confirmed, from statuses during it, such as arrived and seated, lists cancelled and no-show separately, handles walk-ins apart from bookings and keeps a change history for each reservation. This guide uses only those general distinctions. Its workflows, time limits, messages and policies are not applied, and no integration is implied.