Seven parties ahead. The host looks at the room, says "about thirty-five minutes," and writes the name down. Forty-eight minutes later that party finally sits, having watched two groups who arrived after them get seated first because they were a two-top and a table for two opened up. The quote was not slightly off. It was wrong by 37 percent, and it was wrong for reasons the host could have calculated in ten seconds.
Where did thirty-five come from? Almost always the same place: parties ahead times five, or the vague memory that "we usually run about half an hour when it looks like this." Neither of those is a model. The first assumes every table serves every party. The second assumes tonight resembles the average night, which on a Saturday it never does.
And the cost of the miss is not just an annoyed table. Over-quote and parties leave the door before joining the list — you never even see them. Under-quote and they join, wait past the number they were promised, and arrive at the table already unhappy, which is the version that shows up in your reviews. Industry waitlist studies consistently put peak-hour walkaway rates in the 8 to 15 percent range at restaurants quoting by feel, and the single largest driver is the gap between the number spoken and the number delivered. So let us do the arithmetic properly.
The Formula, In Two Halves
A wait quote has exactly two components, and hosts almost universally compute a mangled version of the first while ignoring the second entirely.
Quote = (parties ahead needing your table size ÷ seating rate for that table size) + head-of-line delay
The first half is queue time: how long it takes to work through the people in front of you. The second half is the delay before the very first suitable table becomes available. Skip the second and you are systematically optimistic; use total parties instead of matching parties in the first and you are systematically pessimistic. Most bad quotes contain both errors, pointing in opposite directions, which is why they occasionally look right by accident.
Step One: Count by Table Type, Not by Head
A party of two is not waiting behind a party of six. They are waiting behind other parties of two, because a six-top opening does nothing for them. So the first correction is to segment the queue by the table category each party needs — typically two-tops, four-tops, and large-party or combine tables.
Run the numbers on a real list: eleven parties waiting looks catastrophic to a guest walking in. But if six of them need four-tops, four need two-tops, and one needs a large table, then a party of two is genuinely fifth in line, not twelfth. That distinction is often worth 15 to 20 minutes on a quote, and it is the single easiest correction to make.
Step Two: Compute Your Seating Rate
Seating rate is how often a table of a given size becomes available. The math is trivial:
Seating rate (minutes per table) = average turn time ÷ number of tables of that size
Take a dining room with twelve four-tops running a 68-minute turn. Twelve tables divided by 68 minutes means roughly 10.6 four-tops become available per hour, or one every 5.7 minutes. Now the earlier example resolves: seven four-top parties ahead times 5.7 minutes is 40 minutes of queue time — not 35, and certainly not the "five minutes per party" heuristic that would have given 35 by coincidence and then failed completely in a room with a different table count.
Do this once for each table category and write the numbers on a card at the host stand. Four-tops every 5.7 minutes. Two-tops every 4.1 minutes. Large tables every 26 minutes. Those three numbers are the entire model, and they change only when your table count or turn time changes.
Step Three: Add the Head-of-Line Delay
Here is the half nobody computes. Even if you are first in line for a four-top, you do not sit at minute zero — you sit when the next four-top clears. If the earliest four-top just dropped a check, that is maybe four minutes. If every four-top in the room was seated in the last twenty minutes, it could be forty.
A host with a live table status view can read this directly: find the four-top closest to turning, estimate its remaining time, and add it. Without that view, the honest fallback is half your average turn time as an expected value — 34 minutes on a 68-minute turn — which is a large enough number to explain why quotes made without it run so short at peak. This is precisely the value a live floor view provides, and it is why quote accuracy and table status tracking are the same project rather than two.
Bellwether Kitchen — Madison, WI
A 116-seat American restaurant: fourteen four-tops, ten two-tops, three large tables. Weekend waits routinely hit an hour.
Before: the host quoted by counting heads in the lobby. A two-week audit comparing quoted to actual seating times found only 46 percent of parties seated within five minutes of quote, with under-quotes averaging 13 minutes late. Peak-hour walkaways ran 12 percent, and two-tops were the worst-served group because they were being quoted the four-top queue.
After adopting segmented queue math: the team wrote seating rates by table type on a card, added head-of-line delay from a live floor view, padded five minutes, and reviewed quote accuracy weekly. Within-five-minutes accuracy rose to 83% in nine weeks, walkaways fell to 6%, and the average two-top quote dropped 16 minutes while getting more accurate.
Key insight: "We assumed accurate quotes meant longer quotes. For half our guests it meant shorter ones — we had been quoting them somebody else’s line." — Priya Raman, Service Director
Three Corrections That Fix Most Remaining Error
With the base formula running, three adjustments account for the bulk of what is still wrong. Each is worth a few points of accuracy on its own.
- Quote peak hours from the 75th percentile, not the average. Turn times have a long right tail: the lingering table, the split check across six cards, the dessert-and-two-coffees crowd. On a slow night those average out. On a packed Saturday they stack, because a slow table blocks the queue behind it. If your mean turn is 68 minutes and your 75th percentile is 84, use 84 after 6:30 p.m. This one change typically eliminates most systematic under-quoting.
- Subtract held tables for reservations. A four-top held for a 7:30 booking is not in your seating rate at 7:05. If three of your fourteen four-tops are spoken for in the next half hour, your effective count is eleven, and your seating rate slows from 4.9 to 6.2 minutes per table. Hosts who forget this quote a room they do not actually have.
- Compute large parties as discrete events. Flow math breaks down when the denominator is three tables. If you can seat one party of eight per hour and two are already waiting, the truthful quote is 90 to 120 minutes, not the 25 that a smoothed average would produce. Look at the specific tables that can hold them, quote the actual next-available time, and offer a bar alternative outright.
What ties these three together is that each replaces an average with a distribution. Averages are exactly the wrong tool for a peak-hour queue, because the entire problem is that peak is not average. If you want to sanity-check your turn-time inputs against your own room, a table turnover calculator you can run on your own numbers is a faster first pass than a spreadsheet.
See Why Restaurants Are Switching to KwickOS
KwickOS quotes waits from your actual floor: live table status feeding head-of-line delay, seating rates computed per table size, reservation holds subtracted automatically, and peak-hour turn times taken from your own percentile data rather than a house average. Every quote is logged against the actual seating time, so your accuracy improves on its own.
Start Your Free Trial →How Much Should You Pad?
Once the arithmetic is honest, one judgment call remains: what number do you actually say out loud?
The case for padding is well established. A guest seated at 27 minutes against a 30-minute quote experiences a small win; the same guest seated at 27 minutes against a 25-minute quote experiences a small failure. Identical wait, opposite feeling. Five minutes of padding buys that win on most parties at almost no cost.
But the case against over-padding is just as strong and gets ignored. Guests comparison-shop the door. A 55-minute quote when the honest number is 40 sends people down the block, and you never learn it happened because they never joined the list. Inflated quoting is the most self-concealing mistake in front-of-house operations — it looks like a slow night rather than a lost one.
The working rule: compute honestly, add five minutes, cap total padding at ten. If you find yourself wanting to add fifteen, the problem is your turn-time input, not your confidence. And whatever you choose, be consistent, because inconsistent padding destroys the accuracy data you need to improve. Making the number visible to waiting guests helps too — a posted, updating estimate is easier to trust than a spoken one, as covered in guest wait time display boards.
Measuring Whether Your Quotes Are Any Good
None of this matters if nobody checks the result. The measurement is straightforward and takes about ten minutes a week.
- Log two timestamps per party: the time quoted and the time actually seated. Any waitlist system records both automatically; on paper, write the quote next to the name.
- Compute the within-five-minutes rate. What share of parties sat within five minutes either way of the quote? Below 60 percent means your inputs are wrong. Above 80 percent is a well-run door.
- Check the direction of the misses. Late misses hurt far more than early ones. If most of your error is late, raise your peak turn-time input before touching anything else.
- Break it down by party size. Aggregate accuracy hides the classic failure where two-tops are quoted well and four-tops badly, or vice versa. Segment or you will optimize the wrong number.
- Break it down by hour. Almost every restaurant has one bad hour — usually the shoulder where reservations and walk-ins collide. Fix that hour specifically rather than adjusting the whole model.
Run that review four weeks in a row and the quote stops being a personality trait of whoever is hosting and becomes a property of the restaurant. That is the real goal: a number that means the same thing on Tuesday with your new hire as it does on Saturday with your veteran. Shortening the underlying wait is a separate project, covered in reducing restaurant wait times, and the front-door strategy that feeds all of it is in restaurant queue management strategies. Worth noting too: a meaningful share of quote requests never reach the host stand at all — they come in on a phone line nobody can answer at 7:15, which is its own solvable problem for restaurants drowning in rush-hour calls.
The Bottom Line
A wait quote is arithmetic, not intuition, and the arithmetic is small enough to fit on an index card. Count the parties ahead who need your table size. Divide by how fast tables of that size turn. Add the minutes until the first one actually frees. Use peak percentiles instead of averages, subtract tables held for reservations, treat large parties as discrete events, add five minutes, and never add fifteen.
Do that and two things happen at once, which is unusual in restaurant operations. Half your guests get a shorter quote than they used to, because they were being charged for someone else’s queue. And the quotes get more accurate, because they are finally derived from the room rather than the host’s memory of last Saturday. Fewer people leave your door, fewer people sit down annoyed, and the number you say out loud starts meaning something.