Last updated . Source: SnagRes availability record, 61,530,416 checks across 284 restaurants through 19 October 2026. How we measure. Open dataset.
Change the hour before you change anything else. Across 465,235,816 logged open slots at the restaurants we track, 41% fall before 6 PM and only 24% land between 7:00 and 8:30 PM, so the prime-time search is looking at a minority of the supply. Then set an availability alert for cancellations, try a party of four, and only after that call the restaurant or walk in for bar seats.
Usually not. It means the online inventory for your exact date, time and party size was gone at the moment you looked. Across the 284 restaurants we watch we have logged 7,048 separate release events, moments when an empty calendar suddenly had something in it. Bar and counter seats are also never listed online, so the widget shows less than the room.
On our 19 October 2026 reading, 26 of the restaurants we track had every two-top taken across their window. Seven of them have never logged a single open slot at all, including The Polo Bar across 644,820 checks and 4 Charles Prime Rib across 645,420. For rooms like those, a cancellation alert or a walk-in is the only realistic route.
More than any other single change. Of the 132 restaurants where we have enough slot data, 53 open most often at 9 PM or later and 37 open most often before 6 PM, while only 28 open most often at prime time. I Sodi has never shown us a table later than 5:15 PM. Torrisi puts 43% of its openings at 9 PM or after. The kitchen is the same at both ends of the night.
At the hardest rooms, a lot, and not in the direction people expect. The average open rate across our list is 49% for a party of two and 50% for a party of four, so four is not harder. Of the 26 restaurants at 0% for two, 10 had availability for four at the same moment. Z & Y reads 0% for two and 77% for four.
Marginally, and not the weeknight the advice names. Counting only the nights each restaurant is actually open, Thursday is the easiest at 53.3% and Wednesday second at 52.5%, against Friday at 48.4% and Monday, the hardest of the seven, at 45.5%. The whole spread is 7.8 points. Raw figures exaggerate it because 52 of the rooms we track are dark on Mondays. Treat the day as a tiebreak and move the hour instead.
By being notified rather than looking. Scheduled releases have a clock, and we have a confirmed drop time for 134 of the rooms we track, but cancellations have none: a seat can free up at 4 PM on a Tuesday and be gone in minutes. A watcher that rechecks around the clock catches the window a person refreshing by hand will miss.
No, and that is deliberate. SnagRes watches Resy, OpenTable, Tock and SevenRooms and alerts you the moment a watched table opens. You book it yourself, in your own account, at the restaurant's own price. TableSwap advertises about $75 per successful reservation, which is the fee-on-a-seat practice New York's 2024 reservation law was written to stop. We never book, so nothing happens to a table without you.
Work down a ranked list, and start with the cheapest move. A "fully booked" screen means the online inventory for your exact date, time and party size is gone at the moment you looked. It does not mean the restaurant is full, and it does not mean it will still be gone in an hour. Across the 284 restaurants we watch in New York, San Francisco, Los Angeles and Chicago we have logged 21,432 separate release events, which are moments when a calendar that had nothing suddenly had something. Sold out is a reading, not a verdict.
The list below is ordered by effort, and the order matters because the first two moves cost you nothing and the last two cost you an evening. Most people do this backwards. They spend a week refreshing the app for a 7:30 PM two-top, never change the question they are asking, and then call the restaurant on the day and discover the room was never the problem.
Three of these levers are measurable, and we have measured all three across 61,530,416 availability checks. Two of them turn out to be much bigger than the advice usually suggests, and one of them is much smaller. That is the useful part of this page, and it is in the sections below.
Set an availability alert, so a reopening reaches you instead of you hunting for it.
Change the time you are asking for. This is the largest lever by a wide margin.
Change the party size. At the hardest rooms this is close to a different calendar.
Join the restaurant's own notify list, such as Resy Notify, which covers one restaurant, one date and one party size on one platform.
Try an adjacent date, but expect a small effect rather than a large one.
Call the restaurant, or walk in, for bar and counter seats that were never listed online.
Check again on the day, when holds are released and no-shows get cleared.
For a small number of restaurants, the sold-out screen is what you will see almost every time you look. On our 19 October 2026 reading, 26 of the restaurants we track had every table for two taken across their booking window. For those rooms, browsing the calendar later is not a plan, because the release is gone within minutes; the way in is being ready at the release or hearing first about a table given back.
That is the case for moving from the calendar to an alert for cancellations, and it is worth knowing which situation you are in before you spend two weeks on the wrong one. A room that opens on 60% of dates and happens to be full on your Friday wants flexibility. A room that has never shown you a table wants patience and an alert.
The counters make the distinction concrete. Monkey Bar sits at 50% for a two-top across 550,180 availability checks, and in the same period we have watched it release tables 273 times on 22 different days. Carbone sits at 0% across 588,600 checks with 581 observed release events. These rooms are not closed. They are opening constantly, in slices too thin and too brief to catch by hand.
The reason we can say that about a specific restaurant is that the list is short on purpose. Table Alert advertises 62,198 restaurants. We watch 284, continuously, which is how 61.53 million checks pile up against a list this size. Nobody covering tens of thousands of rooms can tell you how many times one of them reopened last month.
If you change one thing, change the hour. Across 465,235,816 logged open slots on the restaurants we track, 41% fall before 6 PM and only 24% land in the 7:00 to 8:30 PM band that almost everyone types into the search box first. Another 20% arrive at 9 PM or later. Prime time is not just the hardest slice, it is a minority of everything that ever comes free.
It gets sharper per restaurant. Of the 132 restaurants where we have enough slot data to name a pattern, 53 have their single most frequently open time at 9 PM or later, and 37 have it before 6 PM. Only 28 have it inside the prime window. So at roughly two thirds of these rooms, the time you are most likely to get is a time you are probably not asking for.
Individual cases are blunter than the average. At I Sodi, 100% of the 11,180 open slots we have logged fall before 6 PM, and the latest table we have ever seen there is 5:15 PM. At Monkey Bar, 95% of what opens is before 6 PM, most often 11:45 AM. Torrisi runs the other direction: 78% of its openings are 9 PM or later, most often 10:45 PM. House of Prime Rib in San Francisco is at 97% after 9 PM. Carbone's single most frequently open time is 11:30 PM.
None of those rooms will ever look available to someone searching 8 PM. The kitchen is the same at 5:15 PM and at 10:45 PM, the menu is the same, and the table is real. Widening the hour costs you nothing except the version of the evening you had imagined, and it is the difference between a calendar that is empty and a calendar that has something in it.
40% of everything we have seen open falls before 6 PM.
27% falls in the 7:00 to 8:30 PM band. 15% comes at 9 PM or later.
53 of 132 rooms open most often at 9 PM or later. Only 28 open most often at prime time.
I Sodi: 100% of logged openings before 6 PM, nothing later than 5:15 PM.
Torrisi: 78% at 9 PM or later, most often 10:45 PM.
Book the smallest party you can is the standard line, and our own pages said it until we measured it. We check availability separately for a party of two and a party of four at every restaurant we track. The average share of dates with a bookable table is 49% for two and 50% for four. Four is not harder. If anything it is very slightly easier.
The average hides where the lever actually lives. At most restaurants the two figures sit within a few points of each other and party size does close to nothing. At the hardest rooms it does something dramatic. Of the 26 restaurants showing 0% for a two-top, 10 were showing availability for a party of four at the same moment.
Those are exactly the rooms people give up on. Z & Y in San Francisco reads 0% for two and 77% for four. Monkey Bar, 50% and 47%. Bistrot Ha, 13% and 13%. Torrisi and Semma, 0% and 10%. House of Prime Rib and Oriana, 10% and 7%. If you have searched one of these for a couple for a month and concluded the room is shut to you, the search was answering a narrower question than you thought.
The structural reason is mundane. Dining rooms are built out of four-tops, and seating a couple at one burns two covers, so some rooms protect four-tops for parties of four. It also runs the other way at a minority of restaurants, which is why this is worth checking per room rather than assuming. The full breakdown is on our party-size page.
This is the lever that shrinks when you measure it. Go on a Tuesday is the most repeated advice in dining, and the raw numbers look like they are about to settle something: a Monday check found a table 28.6% of the time against 51.3% on a Thursday, a spread of 22.7 points. Almost all of it is closures. Fifty-two of the rooms we track have never once shown a Monday table, because the room is dark on Mondays, and 40 have never shown a Sunday one. On Friday, exactly one room is in that position.
Compare each restaurant only against the nights it actually opens, and 126 restaurant-weekdays drop out of the comparison while the spread collapses to 7.8 points. Thursday is the easiest night at 53.3%, Wednesday second at 52.5%, and Monday the hardest at 45.5%. Tuesday, the night the whole folklore rests on, finishes fourth at 47.7%.
The version of this effect worth acting on sits inside one restaurant rather than across the city. Of the 79 rooms where we can rank all seven nights, the median gap between a room's easiest night and its hardest is 20 points, the mean is 29, and 27 of them swing by 30 points or more. Thursday is the easiest night at 23 of those rooms and Wednesday at 20, but Saturday is the easiest at 8 of them, so the city ranking is the wrong altitude for the restaurant you actually want.
So treat night of the week as a tiebreak, and look up your room rather than the ranking. If two dates work equally well, take Thursday or Wednesday. If moving the date means giving up the restaurant you wanted, keep the restaurant and move the hour instead. The one date check genuinely worth doing first is whether your room is even open that night, because an itinerary built around a Monday dinner at one of those 52 restaurants has a problem no booking strategy fixes.
You do not catch one by predicting it. You catch it by being told. Releases from the restaurant have a clock: we have measured a confirmed drop time at 134 of the rooms we track, so those you can put in a calendar. Cancellations have no clock at all. Someone's plans change at 4 PM on a Tuesday, the seat falls back into the pool, and it is bookable by whoever sees it first, sometimes for only a few minutes.
That asymmetry is the whole argument against refreshing. A person can sit on a booking page for an hour, maybe two. A watcher can sit on it for a month. Across the tracked list we have logged 7,048 release events on 136 restaurants, and the great majority arrived at times nobody would have chosen to be looking.
SnagRes rechecks each tracked restaurant around the clock across Resy, OpenTable, Tock and SevenRooms, and texts or emails you the moment a table appears. Then it stops. We never book the table, never hold it, and never join a waitlist on your behalf. You book it yourself, in your own account, at the restaurant's own price.
That line is worth stating plainly in this category. TableSwap advertises $25 a month plus roughly $75 for every successful reservation, which attaches a fee to a restaurant seat, and attaching a fee to a restaurant seat is the practice New York's 2024 reservation law was written to stop. An alert sits on the other side of that line, and it costs the restaurant nothing.
Call when the online levers are genuinely exhausted, and be honest that this is the point where our data stops. Everything above is measured from the public online calendar. Bar seats, counter seats, the in-house waitlist and the tables a host keeps in reserve are not in any feed we can read, so we will not put a percentage on them. We would rather leave that blank than invent one.
What we can say is that the inventory a restaurant publishes is not the whole restaurant, and that seven of the rooms we track have never logged a single open slot at all while continuing to serve dinner every night. The Polo Bar sits at 0% with no logged open slot across 644,820 availability checks, 4 Charles Prime Rib across 74,910, and The Corner Store, The Eighty Six, The Ambassadors Clubhouse, Jules and Z & Y across tens of thousands each. Z & Y is the instructive one, because it has never shown a two-top and yet reads 80% for a party of four, so there the answer was a different question rather than the door. Whatever the other six are seating, it is not going through the widget.
Timing the call the way you would time a search helps. The day of service is when holds get released and no-shows get cleared, so a late-afternoon call reaches a floor plan the app has not caught up with. Ask specifically about the bar, the counter and a cancellation list, and be easy to seat. A guest who will take 5:15 PM at the counter is a guest a host can actually help.
It cannot tell you about a restaurant we do not watch. Our list is 284 rooms in four cities, chosen because they are hard, and every number here describes that list rather than dining in general. If the restaurant you want is not on it, a broad directory is honestly the better tool, and we would rather say so than stretch a claim.
The percentages are also a point-in-time reading of the current booking window, dated 19 October 2026, while the counts are cumulative and only grow. Raw snapshots get pruned after a few days; the counters keep going. Both are stated with the sample they came from, every time, so you can tell the difference between a pattern and a coincidence.
What we do claim is narrow and checkable: for these 284 restaurants, we know what the calendar actually does, because we have asked it 61,530,416 times. That is the only version of this page worth writing.
Keep reading: How to get a last-minute reservation · Getting in with no reservation at all · Booking for four is not harder · How to get a prime-time reservation · The easiest night is not Tuesday · The hardest rooms right now, ranked
Browse all restaurant reservation guides
SnagRes checks 284 hard-to-book restaurants across four cities and alerts you when one opens. Watching one is free and no card is needed: watch a table free in New York City.