Evidence updated July 23, 2026
How Far in Advance Should You Book a Hotel?
There is no single cheapest day that works for every hotel. If your dates or hotel choice matter, booking 90 or more days ahead is a reasonable decision: early rates in this study were rarely clearly worse than rates seen 31–60 days before check-in. Waiting until 2–7 days out can uncover a lower price, but it also adds price and availability risk.
The evidence covers 1,937 hotels included in the study across 201 cities. A hotel was included when it was publicly listed, had recent comparable price observations in at least five of the eight booking windows, and passed the analysis confidence checks. Each hotel contributes one result, however often its prices were observed.
What the global data says
Most hotels did not reveal one isolated winning time to book. Instead, 83%—1,607 of the hotels analyzed—had a broadly flat pattern across booking windows. In practical terms, the differences were too small or too spread out to support a rule such as “always book exactly 21 days ahead.” This is the strongest reason to treat countdown-day claims cautiously.
The early-booking comparison is more specific. Among 1,923 hotels with both observations, 1.9% were at least 5% more expensive 90+ days out, 4.7% were at least 5% cheaper, and the remaining 93.4% had no meaningful difference under the study’s 5% threshold. These three outcomes cover the whole comparison; the 93.4% is context, not a promise of a particular rate.
Our interpretation is that four months was rarely a clear price mistake in this global sample. It is not a universal instruction to book four months ahead. A traveler who can switch dates and hotels may still prefer to wait, while someone attending a wedding, festival, or school-holiday trip may value securing a suitable room more than chasing a possible small reduction.
The shorter 2–7-day comparison shows why waiting is a trade-off, not a dependable bargain. Short-window rates were at least 5% higher than 90+ day rates for 13.6% of comparable hotels and at least 5% lower for 8.9%. The other 77.5% showed no meaningful difference. A lower rate is possible, but the more common result was little change, and a meaningful increase occurred more often than a meaningful decrease.
Explore the evidence
How often each window was near the lowest observed rate
Global, city, and individual-hotel views
In the global and city views, “near the lowest” means a booking window’s adjusted median was no more than 5% above that hotel’s lowest adjusted bucket median. A value of 80%, for example, means eight in ten hotels with data for that window met this test. It does not mean prices were 80% cheaper, and a high share does not guarantee that the window will be cheapest for your stay.
Select a featured city or search for another city or hotel. City results use the same share-of-hotels measure as the global view. An individual hotel uses a different scale: each bar shows how far that window sat above the hotel’s own lowest adjusted median, with 0% marking its lowest observed window. The chart compares grouped price observations, not exact identical rooms followed through time, so room type, meal plan, and cancellation terms may differ between observations.
Market view
All hotels
Near the lowest observed rate
How early should I book a hotel for the best price?
Start comparing as soon as your travel dates are firm, then book when you find an acceptable rate with terms that match your plans. The study does not identify one universally best day. It does show that booking 90+ days ahead was seldom meaningfully more expensive than booking 31–60 days ahead, so an early reservation is defensible when certainty matters.
If you are flexible on destination, dates, and property, you can watch prices longer because you have substitutes. If you need a particular hotel, family room, accessible room, or neighborhood, the set of acceptable substitutes is smaller. In that situation, availability risk may matter before any price difference does. The data measures observed prices; it does not tell us how many suitable rooms remained for sale.
How long before a trip is it best to book?
For an ordinary trip, use a monitoring period rather than a magic deadline. Compare the market one to three months before arrival, judge the total stay price and booking terms, and reserve once the choice is good enough. Recheck a cancellable booking at sensible intervals instead of refreshing daily; frequent checks do not make a random future discount more likely.
Move earlier for New Year, major conferences, concerts, public holidays, and other peak periods. That advice is based on managing scarce choices, not on a measured event effect in this study. With flexible off-peak dates, waiting may be tolerable, but the 2–7-day result still gives no basis for assuming that a late booking will cost less.
Is four months too early to book a hotel?
No—not when the stay is important and the terms are acceptable. Only 1.9% of hotels in the direct comparison were at least 5% more expensive at 90+ days than at 31–60 days. More were at least 5% cheaper early (4.7%), while most sat inside the neutral band. That makes four months a reasonable time to secure a room, not proof that it is the cheapest time.
Check what “refundable” really permits, note the cancellation deadline, and compare like with like. A breakfast-inclusive room with free cancellation is not equivalent to a room-only, non-refundable offer. If the early offer is non-refundable, the decision also depends on the chance that your own trip changes—a personal risk this price study cannot calculate.
What is the right time to book hotels?
The right time is when the price, room, location, and terms meet your needs at a level of risk you accept. Fixed dates and a preferred hotel favor booking earlier. Flexible dates and many acceptable properties make continued monitoring more practical. This distinction is more useful than choosing a single countdown number for every trip.
Separate price risk from availability risk. Price risk is the possibility that the rate rises after you wait; availability risk is the possibility that your suitable choices disappear. The last-day comparison illustrates the first: rates 0–1 day out were meaningfully higher at 26.7% of comparable hotels and lower at 13.5%. The study cannot measure the second, so travelers with strict requirements should not read a possible late discount as a reason to ignore sell-out risk.
City examples
The global pattern still varies by destination
2–7 days vs. 90+ days
These examples use the median of each hotel’s 2–3-day and 4–7-day adjusted rates, and require both short windows plus the 90+ day window. A city median describes the middle hotel, while the higher and lower shares show how widely individual hotels departed from that middle. Use them to set expectations for monitoring, not to forecast the rate of a specific property.
Paris
101 hotels analyzed; comparison covers 101 hotels with both booking periods
The median 2–7-day rate was 0.1% below the 90+ day rate. Short-window prices were at least 5% higher at 7.9% of hotels and at least 5% lower at 14.9%. Meaningful short-window decreases were more common than increases, so flexible travelers had some reason to monitor—but most hotels still did not deliver a large change.
Dubai
96 hotels analyzed; comparison covers 95 hotels with both booking periods
The median 2–7-day rate matched the 90+ day rate. Short-window prices were at least 5% higher at 10.5% of hotels and at least 5% lower at 12.6%. Meaningful short-window decreases were more common than increases, so flexible travelers had some reason to monitor—but most hotels still did not deliver a large change.
London
86 hotels analyzed; comparison covers 86 hotels with both booking periods
The median 2–7-day rate was 0.9% above the 90+ day rate. Short-window prices were at least 5% higher at 15.1% of hotels and at least 5% lower at 2.3%. Meaningful short-window increases were more common than decreases, so waiting carried more downside than bargain evidence in this sample.
New York
57 hotels analyzed; comparison covers 57 hotels with both booking periods
The median 2–7-day rate was 1.1% below the 90+ day rate. Short-window prices were at least 5% higher at 1.8% of hotels and at least 5% lower at 21.1%. Meaningful short-window decreases were more common than increases, so flexible travelers had some reason to monitor—but most hotels still did not deliver a large change.
Bangkok
48 hotels analyzed; comparison covers 48 hotels with both booking periods
The median 2–7-day rate matched the 90+ day rate. Short-window prices were at least 5% higher at 12.5% of hotels and at least 5% lower at 0%. Meaningful short-window increases were more common than decreases, so waiting carried more downside than bargain evidence in this sample.
Tokyo
42 hotels analyzed; comparison covers 41 hotels with both booking periods
The median 2–7-day rate was 2.5% above the 90+ day rate. Short-window prices were at least 5% higher at 31.7% of hotels and at least 5% lower at 0%. Meaningful short-window increases were more common than decreases, so waiting carried more downside than bargain evidence in this sample.
Practical strategy, not a measured outcome
When a refundable booking can help
One practical response to uncertain prices is to reserve a cancellable rate and keep checking before its deadline. This can protect a suitable room while preserving the option to switch, but it is not automatically cheaper. Refundable rates are often more expensive than non-refundable rates, so a later saving must first recover that premium.
For July 15–17, 2026, each price below is built from two consecutive nightly medians for that two-night window. These are observed medians by booking window, not quotes for an identical offer.
Fairmont Quasar Istanbul Hotel
- 90+ days 6 observation days
- $417
- 31-60 days 8 observation days
- $582.50
- 15-30 days 15 observation days
- $656
- 8-14 days 7 observation days
- $703
- 2–7 days 6 observation days
- $752
+$335 / +80% from 90+ to 2–7 days
The Artisan Istanbul MGallery
- 90+ days 5 observation days
- $315
- 31-60 days 8 observation days
- $276
- 15-30 days 15 observation days
- $302
- 8-14 days 7 observation days
- $266
- 2–7 days 6 observation days
- $258
−$57 / −18% from 90+ to 2–7 days
Conrad Istanbul Bosphorus
- 90+ days 10 observation days
- $482
- 31-60 days 8 observation days
- $444
- 15-30 days 15 observation days
- $484
- 8-14 days 7 observation days
- $486
- 2–7 days 6 observation days
- $484
+$2 / +0.4% from 90+ to 2–7 days
The archive does not identify which offers were refundable and does not match room type, meal plan, taxes, or cancellation terms. Rebooking requires a like-for-like comparison.
Consider a simplified choice: a non-refundable room costs $100 now, a refundable version costs $110 now, and you hope a $95 rate appears later. The refundable premium is $10. The possible saving from replacing the $110 booking with the $95 offer is $15. The price-only break-even calculation is:
Break-even probability = ($110 − $100) / ($110 − $95)
= $10 / $15
= 66.7%Under this two-outcome example, the $95 rate must appear more than 66.7% of the time for “book refundable and wait” to beat paying $100 now on expected price alone. If it appears less often, the $10 premium costs more on average than the chance to save $15.
The study cannot estimate the chance that the later target rate will appear. It does not follow identical room types, cancellation terms, and availability from one observation to the next. The expected-price calculation also excludes the traveler’s separate value of cancellation flexibility. Paying more for the ability to change plans may still be worthwhile, but that is a personal benefit, not a saving demonstrated by this dataset.
How the study works
We examined hotel price observations grouped by how many days remained before check-in: 90+ days, 61–90 days, 31–60 days, 15–30 days, 8–14 days, 4–7 days, 2–3 days, and 0–1 day. For each hotel and window, the analysis used a median, which reduces the influence of an unusually high or low observation. The source analysis also adjusts for seasonality before comparing windows, so an expensive holiday month is less likely to be mistaken for an early-booking effect.
The included population consists of publicly listed hotels with recent, structurally complete observations for at least five of those eight windows and a high or medium confidence result. The evidence dates run from July 19, 2026 through July 19, 2026. We count each hotel once in global results, regardless of how many underlying observations it has, so a frequently checked property cannot outweigh another property by volume alone.
We call a difference meaningful at 5% or more. Smaller differences go into the neutral group because modest changes may not be useful after taxes, exchange-rate movement, or differences between offers. This is an analytical threshold, not a claim that every traveler considers 4.9% unimportant. Pair comparisons include only hotels with every window needed for that comparison; that is why their hotel counts can be slightly below the full 1,937.
What this evidence cannot tell you
The observations do not consistently match an identical room, occupancy, meal plan, tax treatment, or cancellation policy at every point in time. The analysis also cannot observe rooms after they sell out, measure how quickly a preferred category disappears, or prove why a rate changed. Medians summarize the offers that were observed; they are not complete inventories of everything a hotel sold.
Finally, these findings describe past observations rather than forecast a future stay. A citywide event, a hotel renovation, a change in demand, or a new pricing policy can produce a different result. Use the global pattern as a baseline, then check the actual total price and terms for your dates. The sound conclusion is not “waiting never works” or “early is always cheapest.” It is that the evidence does not support one universal cheapest booking day, while the observed penalty for booking early was usually low.