Underbook data · 16 snapshots, September 22 to October 3, 2026
At 61% to 81% of hotels, some booking site listed the stay 5% or more below the going rate
We recorded 6,663 hotel prices in 358 cities on every booking site Google Hotels lists, in 16 snapshots from September 22 to October 3, 2026, each for a two-night stay. Pooled, 4,822 of them (72%) had a listing 5% or more below the going rate, the price most sites agree on. Where room names were compared (14 snapshots), at 2,505 of the 4,181 hotels with such a listing (60%) none of those listings named a room, so check the room before you book. Each snapshot, the method and every CSV are below.
The short version, updated October 3, 2026
16 snapshots, 358 cities, 6,663 priced hotels. In every snapshot, 61% to 81% of hotels had some site listing the stay 5% or more below the going rate. On our original room check, a cheaper listing whose room name matched the room at the going rate was rare: 2.9% to 15.7% of hotels; the stricter check below finds more. Details for each snapshot, and a week-later repeat, are below. The newest snapshot in a few findings: this week in hotel prices.
The free extension (Chrome, Edge, Firefox) runs this check by itself on the hotel page you are on and says whether a cheaper listing names the same room. It sends the hotel name, the town, your dates and party size to our price service (privacy).
A stricter room check, October 1, 2026
Our original check compared only the single cheapest listing's room with the room at the going rate, and it counted loose matches (for example "Standard Room" with "Triple Room"). The new check is stricter about what counts as the same room name, and when the cheapest listing names no room it also checks the other listings 5% or more below the going rate.
Over the 14 snapshots of September 25 to October 1, 2026 that compared room names (5,830 priced hotels, 4,181 with a listing 5% or more below the going rate): the original check found a matching room name at 515 hotels (8.8%); 140 of those do not hold on the stricter check. The stricter check, looking at every listing 5% or more below, finds one at 952 hotels (16.3% of priced hotels, 22.8% of those with a cheaper listing). That is a floor: the stored rows keep one room pair per hotel.
At 2,505 of the 4,181 (59.9%), no listing 5% or more below the going rate named a room at all, so no room check is possible. A matching name is strong evidence, not proof: check the room before you book. The per-hotel CSVs keep the original check's column; the Underbook extension and price checker use the stricter check.
We then tested the stricter check on 300 fresh hotels with every room each site listed (October 1, method written down before the first search): of 194 with a listing 5% or more below the going rate, 53 (27%, likely between 22% and 34%) had one whose room name matched. A separate AI model, shown only the two room names, judged 42 of 50 sampled matches (84%) to be the same room, short of the 95% we set as the bar for calling a name match proof. So treat a matching name as a strong hint and check the room before you book. With the stricter rules in use since October 1, a second blind round on 186 fresh name matches from the October 2 repeat judged 167 (90%) the same room; most of the rest were judged unsure because one name says more than the other (a view or pool access), not a different room. Still short of 95%, so the advice stands.
The first snapshot, September 22, 2026: 426 hotels in 25 cities
One afternoon, 426 hotels, 25 cities, the same two-night stay in October 2026, every booking site Google Hotels lists. Here is how often a cheaper listing exists, how much cheaper it is, who has it, and the one caveat that decides whether it is a real saving.
Is there a cheaper listing for your hotel? Check every site Google Hotels lists for your dates, and whether the cheaper one names the same room.
Correction, September 23, 2026
The first version of this page called 45 of these listings "confirmed same-room savings". The matcher behind that label only checked that the cheaper site named a room, not that the room matched the one other sites were pricing, and a spot-check found shared-dorm beds among them. We now describe the 45 as what they are: cheapest listings that named a room. The matcher in the Underbook extension was fixed the same day (it now compares room names), and the second snapshot below reports how often the room names matched. Every other number on this page is unchanged.
Caveat, October 1, 2026: some prices were listed for travelers in other countries
Every search in this study asked Google Hotels for prices for a traveler in the US, in US dollars. The booking sites' links show that some answers came back priced for a visitor in another country instead (64 countries; the UK, Brazil and Canada most often). In the two runs that kept those links (October 1), 134 of 394 answers in the US-city run and 145 of 427 in the international run (34% of both) were priced for a visitor elsewhere. A site that prices by the visitor's country can show a US traveler a different price: of 6 such listings whose booking page we could read, 2 (Agoda) showed a higher price than ours, by 5.6% and 11.7%, and 4 showed our price.
Of the 578 hotels those two runs count as having a listing 5% or more below the going rate, 179 (31%) had such a listing as their cheapest, and for 156 of them no listing priced for a US traveler was left. Earlier runs did not keep the links, so they cannot be checked; they were searched the same way, and in a separate test of 300 hotels, 88 of 280 answers (31%) came back priced elsewhere. The figures on this page are unchanged and include these listings. From the next run on, each run records which country each answer was priced for, and the week-later repeat planned for October 2 reports its figures both for all answers and for answers priced for a US traveler.
Second snapshot, September 25, 2026
The headline held, and this time we matched room names
We repeated the study in the same 25 cities for stays 35 days out instead of 21 (434 of 436 hotels priced). A listing 5% or more below the going rate existed at 79% of hotels (78% on September 22), and the cheapest listing named a room at 9.7% (10.6%).
This snapshot also gives the study's first figure for cheaper rooms whose names matched. At 25 of 434 hotels (5.8%), a listing 5% or more below the going rate was for a room whose name matched the room at the going rate, using the room-name matching the Underbook extension now runs. Most came from Agoda (9), Booking.com (6), Priceline (4). Matching names is strong evidence, not proof, so check the room before you book.
It also recorded cancellation terms. At 183 of 434 hotels (42%), the lowest listed rate on at least one of Booking.com, Expedia, Hotels.com or Agoda had no free cancellation while another site listed free cancellation for the same price or less (within 2%). A cheaper price you can cancel is also the safe way to check a room you are not sure about: book it, read the room on the confirmation, and cancel before the deadline if it is wrong. More: is free cancellation worth it?
One snapshot each, US market, USD. Per-hotel data for this snapshot: CSV download (CC BY 4.0, cite Underbook). The CSV lists before-tax nightly rates, and sites_listed counts the listing entries kept in the table, the 12 lowest-priced at most (a site listed twice counts twice); the going rate and the cheapest price use every listing. The hotel pages show the same stays all-in (taxes and fees as the sites reported them) and every site found, so their dollar figures are higher and their site counts larger. The discount percentages usually agree within a few points, but for some hotels they differ by more, and a hotel near the 5% line can land on different sides of it, so compare like with like.
Third snapshot, September 28, 2026
The same pattern in 25 new cities
We ran the same method in 25 cities the first two snapshots did not cover (Athens, Atlanta, Boston, Budapest, Cancun, Charleston and 19 more), for stays 35 days out (427 of 428 hotels priced). A listing 5% or more below the going rate existed at 345 of 427 hotels (80.8%), against 78% and 79% in the first two snapshots.
A cheaper listing whose room name matched the room at the going rate turned up at 38 of 427 hotels (8.9%), most often on Agoda (9), Priceline (5), Hotels.com (5). Matching names is strong evidence, not proof, so check the room before you book.
Free cancellation: at 171 of 427 hotels (40%), the lowest listed rate on at least one of Booking.com, Expedia, Hotels.com or Agoda had no free cancellation while another site listed free cancellation for the same price or less (within 2%).
Every hotel from these runs has its own page with its measured prices: hotel prices by city.
One snapshot, US market, USD. Per-hotel data for this snapshot: CSV download (CC BY 4.0, cite Underbook). The CSV lists before-tax nightly rates, and sites_listed counts the listing entries kept in the table, the 12 lowest-priced at most (a site listed twice counts twice); the going rate and the cheapest price use every listing. The hotel pages show the same stays all-in (taxes and fees as the sites reported them) and every site found, so their dollar figures are higher and their site counts larger. The discount percentages usually agree within a few points, but for some hotels they differ by more, and a hotel near the 5% line can land on different sides of it, so compare like with like.
Fourth and fifth snapshots, September 29, 2026
50 more cities, the same pattern
We ran the same method in 50 more cities none of the earlier snapshots covered (Boise, Charlotte, Cincinnati, Cleveland, Columbus, Dallas and 44 more), in two runs of 25 and 25 cities, for stays 35 days out. A listing 5% or more below the going rate existed at 307 of 403 hotels (76.2%) in the first run and 305 of 420 (72.6%) in the second, against 78%, 79% and 81% in the first three snapshots.
A cheaper listing whose room name matched the room at the going rate turned up at 24 of 403 (6.0%) and 33 of 420 (7.9%) hotels. Matching names is strong evidence, not proof, so check the room before you book.
Free cancellation: at 139 of 403 (34%) and 133 of 420 (32%) hotels, the lowest listed rate on at least one of Booking.com, Expedia, Hotels.com or Agoda had no free cancellation while another site listed free cancellation for the same price or less (within 2%).
Every hotel from these runs has its own page with its measured prices: hotel prices by city.
One snapshot per run, US market, USD. Per-hotel data: first run CSV and second run CSV (CC BY 4.0, cite Underbook). The CSVs list before-tax nightly rates, and sites_listed counts the listing entries kept in the table, the 12 lowest-priced at most (a site listed twice counts twice); the going rate and the cheapest price use every listing. The hotel pages show the same stays all-in (taxes and fees as the sites reported them) and every site found, so their dollar figures are higher and their site counts larger. The discount percentages usually agree within a few points, but for some hotels they differ by more, and a hotel near the 5% line can land on different sides of it, so compare like with like.
Sixth and seventh snapshots, September 29, 2026
Leisure destinations: a lower share in US resort towns
We ran the same method in 25 US leisure destinations (Aspen, Atlantic City, Bar Harbor, Branson, Destin and 20 more) and 24 international ones (Amalfi, Aruba, Brussels, Cabo San Lucas, Cape Town and 19 more), for stays 35 days out. A listing 5% or more below the going rate existed at 254 of 409 US hotels (62.1%) and 286 of 394 international hotels (72.6%). The five earlier snapshots ranged from 73% to 81%, so the US resort towns came in lowest of the seven snapshots up to then (a later run of smaller US metros came in lower).
A cheaper listing whose room name matched the room at the going rate turned up at 12 of 409 (2.9%) US and 34 of 394 (8.6%) international hotels. Matching names is strong evidence, not proof, so check the room before you book.
Free cancellation: at 130 of 409 (32%) US and 140 of 394 (36%) international hotels, the lowest listed rate on at least one of Booking.com, Expedia, Hotels.com or Agoda had no free cancellation while another site listed free cancellation for the same price or less (within 2%).
Correction, October 1, 2026: the international run also searched "hotels in Naples", meant for Naples, Italy, and Google Hotels returned hotels in Naples, Florida. They are now labelled Naples, Florida (region US) in the CSV and on the city and hotel pages, and the international figures above leave out its 14 priced hotels (with them: 293 of 408, 71.8%; name-matched 34; free cancellation 143 of 408).
Every hotel from these runs has its own page with its measured prices: hotel prices by city.
One snapshot per run, US market, USD. Per-hotel data: US leisure CSV and international leisure CSV (CC BY 4.0, cite Underbook). The CSVs list before-tax nightly rates, and sites_listed counts the listing entries kept in the table, the 12 lowest-priced at most (a site listed twice counts twice); the going rate and the cheapest price use every listing. The hotel pages show the same stays all-in (taxes and fees as the sites reported them) and every site found, so their dollar figures are higher and their site counts larger. The discount percentages usually agree within a few points, but for some hotels they differ by more, and a hotel near the 5% line can land on different sides of it, so compare like with like.
More cities, September 29, 2026
Smaller US metros look like the resort towns; big cities abroad look like big cities at home
We ran the same method in 24 more US metros (Albuquerque, Birmingham, Buffalo, Burlington Vermont, Chattanooga and 19 more) and 25 more international cities (Abu Dhabi, Auckland, Bogota, Bologna, Brisbane and 20 more), for stays 35 days out. A listing 5% or more below the going rate existed at 243 of 397 US hotels (61.2%), close to the US leisure destinations above, and 339 of 431 international hotels (78.7%), in line with the big-city snapshots.
A cheaper listing whose room name matched the room at the going rate turned up at 24 of 397 (6.0%) US and 47 of 431 (10.9%) international hotels. Matching names is strong evidence, not proof, so check the room before you book.
Free cancellation: at 111 of 397 (28%) US and 141 of 431 (33%) international hotels, the lowest listed rate on at least one of Booking.com, Expedia, Hotels.com or Agoda had no free cancellation while another site listed free cancellation for the same price or less (within 2%).
Every hotel from these runs has its own page with its measured prices: hotel prices by city.
One snapshot per run, US market, USD. Per-hotel data: US metros CSV and international cities CSV (CC BY 4.0, cite Underbook). The CSVs list before-tax nightly rates, and sites_listed counts the listing entries kept in the table, the 12 lowest-priced at most (a site listed twice counts twice); the going rate and the cheapest price use every listing. The hotel pages show the same stays all-in (taxes and fees as the sites reported them) and every site found, so their dollar figures are higher and their site counts larger. The discount percentages usually agree within a few points, but for some hotels they differ by more, and a hotel near the 5% line can land on different sides of it, so compare like with like.
One hundred more cities, September 30, 2026
Abroad, same-name cheaper rooms were nearly twice as common as in the US
We ran the same method in 50 more US cities and towns (Anaheim, Bend, Big Bear Lake, Bozeman, Breckenridge and 45 more) and 50 more international cities (Banff, Barbados, Bath, Bordeaux, Bruges and 45 more), four runs on the same day for stays 35 days out. A listing 5% or more below the going rate existed at 537 of 819 US hotels (65.6%) and 637 of 848 international hotels (75.1%).
A cheaper listing whose room name matched the room at the going rate turned up at 66 of 819 (8.1%) US and 130 of 848 (15.3%) international hotels. The two international runs (15.0% and 15.7%) are the highest shares in any snapshot so far. Matching names is strong evidence, not proof, so check the room before you book.
Free cancellation: at 272 of 819 (33%) US and 229 of 848 (27%) international hotels, the lowest listed rate on at least one of Booking.com, Expedia, Hotels.com or Agoda had no free cancellation while another site listed free cancellation for the same price or less (within 2%).
Every hotel from these runs has its own page with its measured prices: hotel prices by city.
One snapshot per run, US market, USD. Per-hotel data: US cities CSV, US towns CSV, international cities CSV and more international cities CSV (CC BY 4.0, cite Underbook). The CSVs list before-tax nightly rates, and sites_listed counts the listing entries kept in the table, the 12 lowest-priced at most (a site listed twice counts twice); the going rate and the cheapest price use every listing. The hotel pages show the same stays all-in (taxes and fees as the sites reported them) and every site found, so their dollar figures are higher and their site counts larger. The discount percentages usually agree within a few points, but for some hotels they differ by more, and a hotel near the 5% line can land on different sides of it, so compare like with like.
Fifty more cities, October 1, 2026
Two more runs, both inside the range of the earlier snapshots
We ran the same method in 25 more US cities and towns (Alexandria Virginia, Amarillo, Annapolis, Beaufort South Carolina, Carmel-by-the-Sea and 20 more) and 24 more international cities (Antigua Guatemala, Antwerp, Bilbao, Cairns, Cannes and 19 more), two runs on the same day for stays 35 days out. A listing 5% or more below the going rate existed at 266 of 402 US hotels (66.2%) and 300 of 417 international hotels (71.9%), both inside the range of the 13 earlier snapshots (61% to 81%).
A cheaper listing whose room name matched the room at the going rate turned up at 34 of 402 (8.5%) US and 48 of 417 (11.5%) international hotels, also inside the earlier range (2.9% to 15.7%). Matching names is strong evidence, not proof, so check the room before you book.
Super.com was the cheapest site most often in both runs (48 US and 93 international hotels), with KAYAK.com second in the US run (44).
Free cancellation: at 104 of 402 (26%) US and 110 of 417 (26%) international hotels, the lowest listed rate on at least one of Booking.com, Expedia, Hotels.com or Agoda had no free cancellation while another site listed free cancellation for the same price or less (within 2%).
Correction, October 1, 2026: the international run's search "hotels in Hoi An", meant for Hoi An, Vietnam, returned US East Coast beach hotels (Myrtle Beach and North Myrtle Beach, South Carolina; Virginia Beach, Virginia; the Outer Banks, North Carolina), so no hotel in Hoi An was priced. Its 15 priced hotels are real hotels with real prices and stay in the study's totals; the CSV keeps them under Hoi An with region US. The international figures above leave them out (with them: 312 of 432, 72.2%; name-matched 48, 11.1%; Super.com cheapest at 96; free cancellation 113 of 432), and the study no longer counts Hoi An as one of its cities. Their prices, hotel by hotel: Hotels returned for a Hoi An search (Myrtle Beach area, USA).
Every hotel from these runs has its own page with its measured prices: hotel prices by city.
One snapshot per run, US market, USD. Per-hotel data: US cities CSV and international cities CSV (CC BY 4.0, cite Underbook). The CSVs list before-tax nightly rates, and sites_listed counts the listing entries kept in the table, the 12 lowest-priced at most (a site listed twice counts twice); the going rate and the cheapest price use every listing. The hotel pages show the same stays all-in (taxes and fees as the sites reported them) and every site found, so their dollar figures are higher and their site counts larger. The discount percentages usually agree within a few points, but for some hotels they differ by more, and a hotel near the 5% line can land on different sides of it, so compare like with like.
Weekly snapshots, newest October 3, 2026
Every week: new cities, and the oldest cities priced again
From October 3, 2026 the study adds weekly snapshots (1 so far), with the same method and the same scoring code as the snapshots above. The newest, on October 3, 2026, priced 10 new cities (Colorado Springs, Naples Italy, Jaipur, Panama City Panama, Ann Arbor and 5 more) and again the hotels of 14 cities whose newest snapshot was the oldest (New York, Las Vegas, Orlando, Los Angeles, Chicago and 9 more), for a 2-night stay 35 days out (November 7 to November 9). A listing 5% or more below the going rate existed at 310 of 409 hotels (75.8%), inside the range of the 15 earlier snapshots (61.2% to 80.8%). Super.com and Vio.com tied as the cheapest site most often (49 hotels each). Free cancellation: at 171 of 409 hotels (42%), the lowest listed rate on at least one of Booking.com, Expedia, Hotels.com or Agoda had no free cancellation while another site listed it for the same price or less (within 2%).
The hotels priced again, against their own previous snapshot: 233 of the 239 had a going rate both times. For a stay 8 days later at the same lead time, their going rate was a median 0.8% higher (it moved more than 2% up at 113 and down at 104). Of the 193 that had a listing 5% or more below the going rate before, 154 still had one, and at 48 of those the cheapest site was the same. Two different stays, so this is how each hotel's price for the same lead time moved, not the same stay priced twice.
| Snapshot | New cities | Priced again | Hotels with a price | 5%+ below the going rate | Room name matched | Data |
|---|---|---|---|---|---|---|
| October 3, 2026 | 10 | 14 | 409 | 310 (75.8%) | 23 (5.6%) | CSV |
The newest snapshot in a few findings, updated each week: this week in hotel prices. Every hotel has its own page with its newest measured prices: hotel prices by city.
One snapshot per week, US market, USD. New cities are sampled like every city above (the first 20 hotel-type results Google Hotels returns for "hotels in city"); a city priced again keeps the hotels of its previous snapshot, each priced by its Google property token. "Room name matched" is the original room check (the CSV column cheaper_room_name_matched). Per-hotel data in each CSV (CC BY 4.0, cite Underbook), same columns as the other snapshots.
A week later, September 29, 2026
The same stay, priced again: the cheaper listing stayed, the cheapest site changed
On September 29 we priced the first snapshot's stay again (Oct 13 to 15, 2 adults, the same 25 cities, the same method). Google Hotels returned a different mix of hotels, so this compares the 181 hotels priced both times.
The going rate moved at most of them, in both directions: up more than 2% at 76 (42%), down more than 2% at 72 (40%). The typical hotel moved 10.6% either way, while the median change was 0.0%, so there was no overall drift.
A listing 5% or more below the going rate existed at 142 of 181 hotels (78%) on September 22 and 137 (76%) a week later. Of the 142 that had one on the 22nd, 121 (85%) still had one, but the same site was cheapest at only 53 (37%). Which site is cheapest for a stay changes from week to week, so a check before booking is worth repeating closer to the date.
The going rate is the price most sites agree on. When the rooms on offer change, it can move without any site repricing, so part of this movement may be different rooms rather than new prices. Before-tax nightly rates. Per-hotel data: CSV download (181 rows, CC BY 4.0, cite Underbook).
A week later, again: October 2, 2026
The second snapshot's hotels, priced again 7 days later
On October 2 we priced the second snapshot's stay again (Oct 30 to Nov 1, 2 adults) at the same hotels, looked up this time by Google's own ID for each hotel rather than by a fresh city search, so 427 of its 434 priced hotels were priced both times. Of these, 196 were still among the first 20 hotels a fresh city search returned.
The going rate moved up more than 2% at 197 (46%) and down more than 2% at 129 (30%). The typical hotel moved 9.0% either way; the median change was 0.0%.
A listing 5% or more below the going rate existed at 341 of 427 hotels (80%) on September 25 and 341 (80%) on October 2. Of the 341 that had one at first, 293 (86%) still had one, and the same site was cheapest at 108 (32%). In the first repeat (the first snapshot's hotels, September 22 and 29) the figures were 85% still cheaper and 37% the same site.
Super.com was the cheapest listing at 60 of the 342 hotels with a listing 5% or more below the going rate (18%), against 118 of 343 (34%) on September 25, and it was again the site cheapest most often. On our original room check, a cheaper listing whose room name matched turned up at 30 of 429 hotels (7.0%), against 25 of 434 (5.8%). The stricter check, which also looks past the cheapest listing, found a matching room name at 91 of 342 hotels with a cheaper listing (27%).
Prices for a traveler in the US only, a second measure added before this run (why): 271 of the 429 answers this time were priced for a visitor in the US, 153 for a visitor in another country, and 5 named no country. At those 271 hotels, a listing 5% or more below the going rate existed at 228 of 271 (84%). Of the 212 of them that had one on September 25, 192 (91%) still had one and the same site was cheapest at 82 (39%). Super.com was the cheapest listing at 39 of 228 (17%).
Same method and code as the second snapshot for every comparison with it; the stricter check is the October 1 version. Before-tax nightly rates. Per-hotel data: CSV download (427 rows, CC BY 4.0, cite Underbook).
The headline, and the catch
For 331 of 424 hotels that had prices, some booking site listed the stay at least 5% under the rate most sites agree on. The typical gap was 19%, about $28 a night. That part of "the same hotel costs less somewhere else" is simply true.
The catch is what the cheaper listing is for. At 286 of those 331 hotels the cheapest listing named no room at all: the site publishes a price, not which room it buys. At only 45 of 424 (10.6%) did the cheapest listing name a room, and naming a room is not the same as being the same room: in a spot-check, some were a bed in a shared dorm priced against private rooms. They can still be real bargains. You just cannot know from the price alone, which is why our extension says "check the room" unless the room names match.
One more honesty number: for 31% of hotels, no site reported a before-tax rate at all, so the "all-in" price on screen may be missing resort or property fees. We flag those too.
Who is cheapest
Of the 331 hotels where a site undercut the going rate by 5% or more, this is who had the lowest price, how much lower it typically was, and how often that lowest price named a room.
| Site | Cheapest at | Typical discount | Named a room |
|---|---|---|---|
| Super.com | 116 | 20% | 0 |
| klook | 21 | 17% | 0 |
| Agoda | 18 | 19% | 9 |
| Traveloka.com | 17 | 18% | 0 |
| Vio.com | 16 | 19% | 3 |
| Booking.com | 11 | 37% | 9 |
| Amimir.com | 8 | 13% | 0 |
| KAYAK.com | 8 | 22% | 0 |
| Traveluro | 8 | 28% | 0 |
| Hotels.com | 7 | 11% | 7 |
| Evendo | 6 | 24% | 0 |
| Billabook.com | 5 | 17% | 0 |
Super.com was cheapest at 116 hotels and named a room at 0 of them: the wholesaler pattern, a rate with no room attached. When Booking.com or Agoda were cheapest, about half the time they named the room. That difference is the reason a "20% off" rate deserves a second look before you enter a card number.
Chains vs independents, and by region
We expected chains to be where the cheaper listing names its room, because chains publish room names everywhere. The data says the opposite. Chain hotels had a cheaper listing 82% of the time, but the cheapest listing named a room only 2.7% of the time; independents 77% and 13.5%. One possible reason, which this data cannot confirm: chains tend to keep their own site and the big booking sites at the same rate, so most cheaper listings come from wholesalers that publish a rate and no room; independents lean on Booking.com, Agoda and hostel sites, which carry room names (for hostels, often a dorm bed).
By region, a cheapest listing that named its room was most common in Europe (15.2% of hotels), then the US (9.0%), then Asia-Pacific (7.5%). A cheaper listing of some kind existed in 84% of US hotels, 82% in Asia-Pacific and 67% in Europe.
| Segment | Hotels | Cheaper listing | Named a room |
|---|---|---|---|
| chain | 112 | 82% | 2.7% |
| independent | 312 | 77% | 13.5% |
| Region | Hotels | Cheaper listing | Named a room |
|---|---|---|---|
| APAC | 120 | 82% | 7.5% |
| EU | 138 | 67% | 15.2% |
| US | 166 | 84% | 9.0% |
All 25 cities
| City | Hotels | Cheaper listing | Named a room |
|---|---|---|---|
| Honolulu | 18 | 100% | 5.6% |
| Miami | 18 | 94% | 11.1% |
| Los Angeles | 17 | 94% | 11.8% |
| Las Vegas | 15 | 93% | 6.7% |
| London | 18 | 89% | 11.1% |
| Orlando | 18 | 89% | 5.6% |
| Seoul | 17 | 88% | 11.8% |
| Singapore | 17 | 88% | 5.9% |
| Bali | 16 | 88% | 0.0% |
| Chicago | 15 | 87% | 6.7% |
| Barcelona | 18 | 83% | 22.2% |
| Dubai | 18 | 83% | 11.1% |
| Sydney | 18 | 83% | 22.2% |
| Bangkok | 17 | 82% | 0.0% |
| Nashville | 17 | 76% | 5.9% |
| Vienna | 17 | 76% | 11.8% |
| Austin | 15 | 73% | 6.7% |
| Lisbon | 15 | 67% | 6.7% |
| New York | 18 | 67% | 11.1% |
| Paris | 18 | 67% | 22.2% |
| San Francisco | 15 | 67% | 20.0% |
| Tokyo | 17 | 65% | 0.0% |
| Berlin | 18 | 61% | 11.1% |
| Rome | 16 | 56% | 25.0% |
| Amsterdam | 18 | 33% | 11.1% |
What to do with this when you book
- Assume a cheaper listing exists. It did for 78% of these hotels. The going rate is the price most sites agree on, not the floor.
- Before you pay less, confirm the room. Open the cheaper listing and check the room name, bed type and cancellation terms against the one you were about to book. If the site does not name the room, that is the answer.
- Check the all-in number. If no site shows a before-tax rate, resort or property fees may be missing from every price you see.
- If you are not booking today, watch it. Prices for the same stay move; a watch on one listing compares like with like and never needs a room match.
Underbook does steps 1 to 4 on the hotel page you are already on.
A free browser extension for Chrome, Edge and Firefox. On a hotel page with your dates, it checks the other booking sites Google Hotels lists and shows the going rate, the cheapest listing, whether that listing names the same room (it says "check the room" when the names do not match), whether the prices include taxes and fees, and lets you watch the price by email. It sends the hotel name, the town, your dates and party size to our price service (privacy), and you book wherever you choose.
Method, limits and data
- 426 hotels: the first 20 hotel-type properties Google Hotels returns for "hotels in city" in each of 25 cities (Amsterdam, Austin, Bali, Bangkok, Barcelona, Berlin, Chicago, Dubai, Honolulu, Las Vegas, Lisbon, London, Los Angeles, Miami, Nashville, New York, Orlando, Paris, Rome, San Francisco, Seoul, Singapore, Sydney, Tokyo, Vienna).
- One snapshot per hotel on September 22, 2026, for a 2-night stay checking in 2026-10-13 and out 2026-10-15, 2 adults, US market, USD. A snapshot is not a guarantee: prices move and this is one day.
- "Going rate" is the rate the largest number of credible listings agree on, with ties broken low. "Cheaper listing" is any site 5% or more below it. "Named a room" means the cheapest site published a room name for its rate. These are the rules the Underbook extension applied live on September 22; the study reused its code rather than a copy.
- Prices come from Google Hotels' listing of each booking site. Sites not listed there are not in the data. A "cheapest" site is cheapest among what was listed that day.
- "Chain" means the hotel's name matched a list of chain brands fixed before the run (Marriott, Hilton, Hyatt, IHG, Accor, Wyndham, Choice and others); every other hotel counts as independent. It is a name rule, so a few chain hotels without the brand in their name count as independent.
- Full per-hotel table: CSV download (CC BY 4.0, cite Underbook). The CSV lists before-tax nightly rates, and
sites_listedcounts the listing entries kept in the table, the 12 lowest-priced at most (a site listed twice counts twice); the going rate and the cheapest price use every listing. The hotel pages show the same stays all-in (taxes and fees as the sites reported them) and every site found, so their dollar figures are higher and their site counts larger. The discount percentages usually agree within a few points, but for some hotels they differ by more, and a hotel near the 5% line can land on different sides of it, so compare like with like. Questions or corrections: hi@underbook.co.
Citing this study
Underbook, "Hotel price study: 6,663 prices, 358 cities, 16 snapshots" (hotel price study, updated October 3, 2026), https://underbook.co/hotel-price-dispersion-study
The charts and the data are free to reuse under CC BY 4.0 with a link back. Chart images: the headline split, who is cheapest, where the cheapest listing named its room. Journalists who want a per-city or per-site cut: hi@underbook.co.