Dwell Time

Dwell time is the number the whole book is built on. Slot length, pacing caps, quote times, and turn targets are all derived from how long a party actually sits, and most restaurants run on an assumption inherited from whoever configured the reservation platform rather than on a measurement.

How is dwell time measured?

From seating to table cleared, in principle. In practice the answer depends on which system you ask, and the two available answers differ by a lot. A reservation platform records the seating and the clear on the host’s floor screen, so its figure includes the minutes before the first order and the minutes after payment. A point of sale measures check open to check close, so its figure starts when a server rings the first item and ends when the check settles, cutting both ends off.

The same party can therefore be a 110-minute occupancy and an 80-minute check. The point-of-sale number is usually the easier one to export, so books get built on the shorter figure, slots get set too tight, and the room runs behind from the second seating onward. When comparing dwell figures between venues or across a year, confirm which clock produced them.

Why is average dwell time the wrong number to plan with?

Because the average hides the parties that actually break a service. If the house average is 95 minutes and one party in six runs past 140, a book paced on 95 will fall behind early and stay behind, since there is no slack anywhere in the grid to absorb the long ones. Planning on a high percentile rather than the mean costs a little capacity and buys back the waits.

The figure also needs breaking out before it means anything. Two-tops at lunch and six-tops at Saturday dinner are not remotely the same number, and neither are the first seating and the last. Dwell time by party size and by service period is a usable planning input. One house-wide average will be wrong in both directions at once.

Does dwell time vary by guest?

Yes, and it repeats reliably enough to be worth recording. Some guests are gone in 70 minutes and some settle in for three hours, and the same party tends to behave the same way on the next visit. That makes dwell a guest attribute rather than only a table attribute, and it is invisible until seatings and checks are joined to a person. A restaurant CRM that matches checks to reservations produces a dwell history per guest as a by-product of the join.

What it changes is practical. Seat the known long-dwellers where a delayed turn costs the least, stop counting on the 8:45 behind them, and quote the waitlist against how the room actually behaves tonight. It also reframes a phenomenon usually discussed as a complaint: a guest who lingers is a known quantity rather than a surprise, and a book built around them holds up.