You're welcome to wait in the bar
February 7, 2010 3:29 PM   Subscribe

How do restaurants estimate how long you'll have to wait to be seated?

One article I found mentions "complex formulas" used to come up with estimated wait times, and I'm curious if anyone knows what those formulas might be - or, alternatively, what factors are considered when the front hostess tells you how long the wait will be.

I couldn't find any good answers online, and the wife and I had plenty of time to discuss it last night while waiting an hour for a table - after having been promised a 20-minute wait, of course.
posted by AngerBoy to Grab Bag (27 answers total) 6 users marked this as a favorite
 
I think they know the room and when they get a block of tables, just a rough idea of 'I just sat a truckload of tables and none of them will be leaving soon and there's four couples ahead of you' versus 'the room is full but the seatings have been staggered and there's no one else waiting'. So, the first 45 minutes, the second, ten minutes, both just guesses. I don't think complex formulas are needed; sometimes they're dead on, sometimes they're earlier or later but I don't remember noticing either way more than once or twice in my life.
posted by A Terrible Llama at 3:35 PM on February 7, 2010


I'm familier with a restaurant that will always quote 10-15 minutes when reality is 40+. They lie to keep their tables turning over all night long.
posted by mmdei at 3:51 PM on February 7, 2010 [1 favorite]


Best answer: We have a general idea of how long a turn is...e.g, how long a table takes to complete from sitting down to the check being dropped. At my restaurant, it is a 90 min turn, probably plus/minus 15 min depending. So given what time the table is sat we have an estimate on what time they'll be up out of their seat. We have a sense that if a table just ordered desserts they'll probably be out the door with in 25 minutes (max): the desserts are quick fires, quick to eat, and then assuming no coffee time or chit chat, they ask for the bill and we reset the table once they're up...especially if we know someone is waiting.

Long story short: experience tells us how long people take to eat any combination of courses at our restaurant. Should the quoted time be under, it could be because the table decided to camp, we misestimated, or we couldn't get to the reset quick enough if we were super busy. We don't want to compromise service to our current guests just to hastily seat two more people that may get sub-par service.

In general, we certainly don't try to under estimate and in fact (speaking personally for my self and my colleagues) try to over estimate the quoted time. Having people crowd by the door isn't in our interest, seating people in a timely fashion is! :)
posted by virga at 4:03 PM on February 7, 2010 [6 favorites]


Plenty of restaurants I go to quote 30 minutes and then seat me in 5-10. I've seen a few POS/seating software systems that seem to have enough data to give a good ballpark -- they know exactly how long each table has been there, so they could give a decent estimate -- but usually it's just seat of the pants.
posted by xueexueg at 4:05 PM on February 7, 2010


There really isn't a complex formula, at least in the restaurants I've worked in. And in my experience, it's better to over quote the time than to under quote and that is how every host/hostess was trained. Under quoting can piss people off, but you don't want to over quote by too much because people won't want to wait that long. A competent host/hostess will take the following into consideration when giving you a quote:

1. How busy it is.
2. Lunch or dinner shift (tables turn faster at lunch).
3. Number of people in the party.
4. Number of tables ahead of you.
5. How staffed the restaurant is.

I eat out a lot, and every time I've had to wait for a table, the host/hostess has been pretty much spot on. Although there has been the occasion grossly under quoted time but these things can happen. You can't predict the occurrence of all the tables wanting to camp out at the same time.
posted by MaryDellamorte at 4:06 PM on February 7, 2010 [1 favorite]


I find it hard to believe that these "complex formulas" are used in practice, because it always seems like the host/ess just looks at a piece of paper which lists the people that need to be seated, thinks for a moment, and says a number. Usually this number is right.

Maybe I don't go to the kind of restaurants that are run by computers.

Then again, I suspect "complex formulas" may mean something as simple as "about five minutes for every table in front of you". (Adjust "five" accordingly for the size of the restaurant and how long a typical party takes.) Math is scary!
posted by madcaptenor at 4:21 PM on February 7, 2010


I think this problem would fall in the domain of queuing theory, which is studied by computer scientists. Some of the research seems to be in regard to fast food, but it looks like there are applications involving table service as well. Beyond estimating a particular wait time, this could let you estimate how much seating you should have, or what the peak or average wait times would be for a particular setup.

I have no idea how much realtime simulation computerized host stations actually do, for those restaurants that even have them.
posted by serathen at 4:23 PM on February 7, 2010 [2 favorites]


You could do the simulation very quickly, I think; the hard part would be collecting the data.
posted by madcaptenor at 4:25 PM on February 7, 2010


The better the host/hostess, the better they are at ESTIMATING TIMES BASED ON PAST EXPERIENCE.

Thats all it is. I'm sure there are all sorts of formulas that may work for certain restaurants. I wouldn't expect ANYONE to actually do this unless they are a actuary for the WalMart food cafe or even Starbucks.
posted by hal_c_on at 4:40 PM on February 7, 2010


I was a hostess (twenty years ago!) and, yeah, it was as hal_c_on says -- educated 'guesstimates.' Glance around -- all those tables just sat down -- hmm -- they're just getting their food, but, ah, those two are looking businesslike about finishing their coffee and those two are declining dessert, thus: ten minutes? That sort of thing.
posted by kmennie at 5:36 PM on February 7, 2010


Best answer: I've never heard of queuing theory, which sounds interesting, but essentially: Mary Dellamorte and Virga have it.

I've worked in many restaurants, often as host or shift manager. For dinner service, for a party of two, 90 minutes is an extremely common amount of time for a table to be occupied. For a party of four, sometimes it's safer to state 3 hours. On a Friday or Saturday night, people tend to linger a bit more because they may be taking more time with drinks and springing for dessert - things that aren't as common on weeknights.

At my most recent restaurant job, we kept a clipboard at the desk with a map of the entire night shown as a grid. This was about a 60-seat restaurant that regularly filled on Friday and Saturday and stayed full, so reservations became a fine art. On one axis were the tables available, listed by number of seats, like this:

2

2

2

4

4

6

...meaning we had three two-tops, two four-tops, and one six, for instance (of course it is usally a much longer list in most restaurants.

Across the other axis we listed times in half-hour from the start to the end of service: 5 PM to 10.

And to indicate the number of people in the party, we'd write in that number and circle it in the center of their shaded in (reserved) block, so we could see how many individuals we were saving the seats for. We added the last name of the reservation maker and a phone number.

When someone called in, we'd pencil in the blocks representing the time they were likely to need. Again, on a Friday-Saturday, we'd shade in three half-hour blocks for a party of two (90 minutes) and four for a party of four or more (120 minutes).

It was always wise to spend some time with the existing reservations on this grid at the beginning of the night. Until the day itself, people who took reservations on the phone were not necessarily completely efficient at setting up the grid. Sometimes you could find a way to shift a three-block chunk on the grid to another location, freeing up a spare half hour or so. Shifting a party of 2 that someone had shaded into a four-top location could free up two entire seats and accommodate another party that would not have otherwise fit. Sometimes you could do things like move a party of 3 from a larger, true 4-top table to a two-top to which you could easily add a third chair, or a party of 5 from a true 6 to a four with an extra chair, could also change the mathematics. Round tables were a godsend for this, flexing to accommodate anywhere from four to seven people and perhaps even eight on a crazy night.

Every now and then someone shows up with an additional person for which they didn't have a reservation, or show up with one person less. This can necessitate a change in the seating plan, and can unexpectedly free things up. Sometimes a couple who decides to have a drink at the bar while waiting elects to stay at the bar instead of taking a table, freeing up their seats. Sometimes there are no-shows, and at the place I'm thinking of, we would wait 15 minutes, and if they neither showed up nor called, we wrote in "NS/NC" next to their name in the reservation book and remembered that for next time.

On very busy nights, if people were waiting, it was not unheard of for us to 'check in' with tables who had overstayed the 2-hour window and try to move them along. Sometimes people do 'camp,' and are quite unaware that by lingering, they might be usurping the precious resource of table space and making others wait longer.

As others have noted, even if your table is actually clear of people earlier than the time your host has given you as the likely time you'll sit down, they may not actually seat you until the time they've stated. So let's say you walked in at 6:35 and I was the hostess; I might tell you the wait will be "25, 30 minutes." That's because I know that Table 4 is finishing the last two bites of their dessert and should be out the door in ten minutes., meaning the table will be empty at 6:45. However, "empty" does not mean "ready," a fact that frustrates many a confused and hungry person who has never worked at a restaurant and can see empty tables that seem to be available for them. AS the host, I have to build in a cushion for a few eventualities. IF a number of people are finishing their meals and leaving all at once, the waitstaff will be a bit delayed getting around to them all and taking payment. If someone pays in cash and needs change, there will be a slight delay. If the kitchen is backed up, or the rest of the restaurant is busy, the busboys may not get to the table in time to clear it right after the diners get up. If I know I have a 15-minute cushion in there, it allows for any SNAFUS with payment, and allows for the table to be cleared and re-set - by me, if it's really that busy - before I bring you to it. So whatever time you have been given, it probably does contain a 10- to 15-minute cushion consistently. Sometimes nothing goes wrong and you feel as though you got in 'early,' which is all to the good. But sometimes it really is going to take that long, and chances are we don't go into overtime because I anticipated the outer time limit required to get your table ready.

I really enjoyed hosting. The reservations book was like a very complex game of Tetris that had to be played in real time with real people as the funny-shaped pieces.

So it does interest me that there's a math to this - and it makes sense. There are a lot of variables existing within a world of hard limits - time, staff availability, and number of chairs.
posted by Miko at 6:19 PM on February 7, 2010 [18 favorites]


For a party of four, sometimes it's safer to state 3 hours\

Ack! Hello no. Make that 2!
posted by Miko at 6:20 PM on February 7, 2010


P.S. I understand Open Table runs on POS systems and now does this sort of thing automatically - probably calculating wait time, too.
posted by Miko at 6:23 PM on February 7, 2010


In addition to estimations like "table fifteen was seated at six-thirty, so they should be done by eight or so," it's also part of the job of a good server to tell their host/ess things like "table nine is chatty and will probably have dessert, so they'll probably be here for another half hour at the least," or "the people at table eighteen never get dessert or coffee, and their table will probably be available in ten minutes or so." You also have to plan longer times for larger groups, or reservations for special occasions like birthdays.
This is, of course complicated by reservations, and the need to shuffle them to maximize table usage. That is, while we may have reserved table one for your party of four for six thirty, if you're not there when you're supposed to be, and group of four wants your table (and table two will be ready as soon as Mr. Somebody stops chatting and pays the bill and leaves), we might give the people the table that's waiting for you, and give you the table that'll be available shortly. Usually problems in time-estimating happen due to switcheroos like this - especially when there's no one dedicated to seating, and two severs both plan to give the same table to two separate groups without writing it down, or if someone misreads a reservation-change, or something.
Estimates are also a mess when flustered, inexperienced servers just guess - I hate when a server obviously doesn't know what's going on on the floor, and just spits out an answer rather than finding out how many people are waiting to be seated, how long people have been there, etc.
(Sorry if this answer leaves you even more confused!)
posted by Bergamot at 6:37 PM on February 7, 2010 [1 favorite]


Actually, what Miko said is pretty much perfect.
posted by Bergamot at 6:40 PM on February 7, 2010


In actuality, they likely quote from experience - they know when people were seated and how long people are expected to stay; they may even consult the booking sheet to see when a table will likely open up. Most likely once the place is full, there's a known wait-time that pretty much always applies, or at least only varies across the week and/or evening, so they might be able to say "it's saturday 8pm, your wait time will be X" based on their experience of that restaurant and its usual customer dynamics.

There are formulae based in rigorous statistics: google for "queueing theory" and M/M/D to get an idea of it. No way the waiters are using them though.
posted by polyglot at 6:52 PM on February 7, 2010


Like Miko, I worked in many restaurants, at all different price points, and I will opine that the experience she outlines is an extreme outlier. Most restaurants do it all by feel, with zero training and zero thought behind it.

The thought process, at most, is:

"How long's the wait? Hmm, well, last time it was like this, things seemed to take about X minutes, and when I quoted that figure to people, they didn't really complain. So ... X minutes."
posted by Cool Papa Bell at 7:16 PM on February 7, 2010


Sometimes there are no-shows, and at the place I'm thinking of, we would wait 15 minutes, and if they neither showed up nor called, we wrote in "NS/NC" next to their name in the reservation book and remembered that for next time.

What would happen next time? Would you refuse to take their reservation?
posted by grouse at 8:12 PM on February 7, 2010


Also, waits don't just happen, they build up. It starts with a ten minute wait. If those people aren't seated before someone else arrives, it goes up to a 15-20 minute wait. You use the wait times of people ahead of you to calculate your own wait.
posted by abc123xyzinfinity at 8:13 PM on February 7, 2010


Both of the (higher-end) restaurants I hostess at go by what Miko said.
posted by pintapicasso at 2:42 AM on February 8, 2010


Best answer: For a restaurant with lots of tables and steady flow of customers, the host/hostess keeps a list of people's names, the number of seats they want, and the time they arrived. As people are crossed of the list, the host writes the time at which they were seated. Now you know how long they waited. When a new person shows up, you say, "the party just seated waited x minutes", unless the party that just arrived or is being seated is very large. If the waiting party is large, you say, "it'll be a longer wait for your party of 8 since we'll have to clear two tables. If the line is short enough, the estimate becomes, "there is only a few tables ahead of you, it won't be long." The point is, a good host/hostess never gives an exact time because they can't predict the future and it will only piss off the customer when the estimate is wrong.
posted by Tristram Shandy, Gentleman at 6:57 AM on February 8, 2010


I will opine that the experience she outlines is an extreme outlier. Most restaurants do it all by feel, with zero training and zero thought behind it.

It's true most of my work fell into the narrow category of smaller (60-80 seat) owner-operated restaurants that were booked wall to wall on weekend nights. I think it was really because of those pressures that such fine-tuned systems were developed.

What would happen next time? Would you refuse to take their reservation?

The curve there wouldn't be quite so steep. What actually happened is people who did this repeatedly began to develop a reputation. NS/NC is pretty serious - a lot of people arrive late, but usually they'll call or eventually show up. That's hard enough to deal with. But reserving a table on a Friday night and totally blowing it off? That's pretty bad. So if this happened once or twice with a given name, we might actually star the name in the book to indicate some level of sketchiness. If they weren't there on time next time, we wouldn't be as likely to hold their table for the full 15 minutes we might normally extend, but give it away to someone who was waiting and in the room already. When reserving on the phone, those people might also get a sterner statement from us - "It's going to be very busy tonight, so please call us if your plans change."

it's risky not to take the reservation at all - I can't honestly think of anyone who was flat out blacklisted - but it was quite common to weigh those reservations with much less importance, and consider them tentative unless the people were standing there in front of you. In other words, the space they had reserved would be seen as much more shift-able in the host's eyes.
posted by Miko at 7:14 AM on February 8, 2010 [1 favorite]


Best answer: ...it occurred to me that the kind of restaurant makes a difference. The restaurants I'm talking about were mostly filled by reservation on weekend nights and sometimes Thursdays and Sundays, too. However, if you're thinking about a lot of chain-type restaurants - Olive Garden, Chili's kinds of places - or something else that's more casual in atmosphere and menu, like pubs or pasta/pizza places, they might not deal that much in reservations, but rely instead on walk-in traffic. And that's where I could see the 'wait time' being far more subjective, because probably the amount of time it takes a group to dine is far more variable (like, a table getting wings and beers will take a different amount of time than a table getting starter, entree, and dessert).

Also, sometimes those places have a much larger seating capacity than a finer dining restaurant will. That means, again, the variability is greater. Where you might not be able to count on three of three four-tops all turning over within 90 minutes to create three new openings, you can probably count on three of twelve four-tops turning over within the time frame expected. There's more "churn", as it were, with more tables and with more ad-hoc seating that isn't planned to start on the half-hour or quarter-hour, so a certain number of openings will always be somewhere in the pipeline.

In those settings I can see giving a pretty vague wait time and winging it a bit more. In that case it's more of a pure queue, first come first served as things become available, and less of a jigsaw puzzle of maximizing space use efficiency with many pre-set reservations.
posted by Miko at 7:26 AM on February 8, 2010 [1 favorite]


I worked at a chain-type restaurant, and our estimate was really as simple as "There are Y names on the waitlist, that will be about X long." This did not take into account anything even mildly complicated like number of people in the party. The only thing we did that deviated very far from that was to make longer estimates for large parties depending on how many of our big tables were full.
posted by freezer cake at 10:48 AM on February 8, 2010


if you're thinking about a lot of chain-type restaurants - Olive Garden, Chili's kinds of places - or something else that's more casual in atmosphere and menu, like pubs or pasta/pizza places, they might not deal that much in reservations, but rely instead on walk-in traffic. And that's where I could see the 'wait time' being far more subjective

At the Olive Garden where I worked, we generally quoted 15 minutes per name on the waiting list, although I would add or subtract time to that depending on the "feel" of how busy things were. In general, though, I agree with virga, above; I'd be more likely to overestimate the wait time since it was less likely to piss people off.

Of course, people get pissed no matter what; I would often quote a wait of like 40 minutes, and then after 20 minutes, would have customers being all, "We've been waiting for 40 minutes already! And you told us it would be only 15 fifteen minutes!!!" and I'd be all, "Uh"
posted by Greg Nog at 1:12 PM on February 8, 2010


Greg Nog: fifteen minutes per name? Really? I'm totally pulling numbers out of my ass, but let's say you can turn over a table every two hours. Then fifteen minutes per name seems to mean you have about (two hours)/(fifteen minutes) = eight tables. Unless you worked at a very small Olive Garden, that seems surprising. But then again, I've never worked in a restaurant, so I wouldn't know.
posted by madcaptenor at 6:59 AM on February 9, 2010


Whoa, yeah, I just realized you're right. Sorry, it's been a few years. I think it was five minutes per name, actually, which coincedentally works much better with your math, above. I apologize for my misremembering!
posted by Greg Nog at 1:18 PM on February 9, 2010


« Older Home Cheap Home   |   Rip and Burn Multiple Movies to a DVD using New 0... Newer »
This thread is closed to new comments.