A booking curve plots lead time (days before the stay date) on the horizontal axis and the booked ratio (OCC) on the vertical axis, drawing as a single line how reservations built up toward one stay date. Because one curve can be drawn for every stay date, peak days, weekdays and holiday periods each reveal their own "fill-up shape." Switch the vertical axis to the remaining-room ratio and the same curve can be read upside down. For hotel operators, revenue managers and investors, this article lays out how to read the fixed checkpoints (45, 30, 14 and 7 days out, and the day before), the measured October 2026 curve for 541 business hotels in Tokyo, how ramp-up differs by hotel type, pace management through month-on-month comparison, and the most common misreadings. Going forward, this article serves as the definitional reference for HotelBank's prefecture-level and event-level booking curve analyses.

Definition and How to Read It: Five Checkpoints, the Mirror Image of Remaining Rooms, and Pickup

A booking curve is a tool for tracking how reservations accumulate on a timeline whose finish line is the stay date. On the horizontal axis, the left end is the distant future (LT45) and the right end is just before the stay date (LT1); the curve rises as you move right. The vertical axis is the booked ratio; when drawn from OTA listed inventory, this is the occupancy rate on an OTA listed-inventory basis (estimated). Since each stay date has its own curve, averaging 31 days in a month shows "the typical fill-up pattern for that month," while isolating a specific Saturday shows "how that day filled up."

In practice, reading values at a few checkpoints is more useful than staring at the whole curve. This article uses five basic checkpoints: 45 days out (LT45), 30 days out (LT30), 14 days out (LT14), 7 days out (LT7) and the day before (LT1). LT45 is the starting point when sales allotments are fully loaded and booking begins in earnest; LT30 is the milestone for checking progress with one month to go; LT14 and LT7 are decision points for rate changes; and LT1 is the final level reached. However, the same checkpoint can mean different things depending on hotel type and day of week, so the differences by hotel type are examined later.

The relationship with the remaining-room ratio is simple: the booked ratio and the remaining-room ratio always add up to 100%. In other words, flipping a booking curve upside down gives the remaining-room curve, and the front desk's "how many more rooms can we sell" and the management report's "how full are we" are the same number seen from opposite directions. The number of points the booked ratio rises between two checkpoints is called "pickup." A market with large pickup between LT30 and LT14 can be read as one where the main build-up of bookings happens between one month and two weeks before the stay date.

Measured Example: 541 Tokyo Business Hotels Went from 61.6% at 45 Days Out to 95.8% the Day Before in October 2026

A definition alone makes the shape hard to picture, so here is a measured curve. The subject is business hotels in Tokyo: a fixed cohort (the same set of properties for every stay date) of 541 properties with 78,615 rooms that list at least 30% of their total rooms on OTAs and similar channels. For the 31 stay dates of October 2026, the booked ratio was averaged at each lead time. According to MetroEngines Research data, the booked ratio was 61.6% at 45 days out, 70.7% at 30 days out, 83.1% at 14 days out, 88.1% at 7 days out, and reached 95.8% the day before.

Source: Compiled by HotelBank Editorial Team from MetroEngines Research (Tokyo business hotel fixed cohort, N=541 properties / 78,615 rooms, average of 31 stay dates in October 2026, observation coverage LT45 to LT1)

Cutting the curve at the checkpoints shows where bookings concentrate. It rose 9.1pt over the 15 days from LT45 to LT30, 12.4pt over the 16 days from LT30 to LT14, 5.0pt over the 7 days from LT14 to LT7, and 7.7pt over the 6 days from LT7 to LT1. Per day, the LT30 to LT14 stretch works out to about 0.8pt/day and the LT7 to LT1 stretch to about 1.3pt/day, making the final week the fastest. At the same time, it should not be overlooked that just over 60% was already filled at 45 days out. That said, Tokyo's 61.6% is close to the 60.1% median for business hotels across Japan's 47 prefectures (ranked 19th), so the "starting point" of the curve is not exceptionally high. Its day-before level of 95.8% ranks 3rd; what really characterizes Tokyo is the size of the build-up right up to the last minute.

Source: Compiled by HotelBank Editorial Team from MetroEngines Research (N=541 properties / 78,615 rooms, average of 31 stay dates. Remaining rooms are cohort totals)

Read Price and Sold-Out Share on the Same Axis: Published Prices Up About 20%, Sold-Out Share from 6.0% to 34.5%

The horizontal axis of a booking curve only becomes a pricing tool once price and the sold-out property share are overlaid on it. The average published price (2 guests per room, tax included) for the same cohort rose about 20%, from about ¥26,900 at LT45 to about ¥32,400 at LT1. Two effects overlap here: cheaper plans sell out first as bookings build, raising the average of the plans still listed, and properties raise rates as they watch remaining rooms shrink. Note that this average published price is "the price of inventory still unsold at that point," not a settled rate.

Source: Compiled by HotelBank Editorial Team from MetroEngines Research (N=541 properties, average of 31 stay dates. Prices are published per-room rates for 2 guests per room, tax included)

The sold-out property share (the share of properties with zero listed rooms remaining) rises from 6.0% at LT45 to 14.3% at LT14 and 18.3% at LT7, reaching 34.5% at LT1. Put differently, by the day before, about one-third of the cohort had used up its OTA sales allotment, and the 4.2% remaining-room ratio is made up of inventory at the other two-thirds of properties. Even when the market-average booked ratio is the same 90%, properties face different options in a market with a high sold-out share than in one with a low share. The stretch where the sold-out share climbs sharply (LT3 onward in this example) is when overall market supply is thinning for properties that still hold inventory, making it a decision point on whether to hold or raise rates.

Hotel-Type Differences: City and Resort at 76% at 45 Days Out, Business Catches Up Late

Even within the same Tokyo market and the same October 2026, the shape of the curve changes with hotel type. Placing three types on the same horizontal axis, business (541 properties / 78,615 rooms), city (52 properties / 15,995 rooms) and resort (12 properties / 742 rooms), shows the difference in how early they ramp up at a glance. At LT45, the booked ratio was 61.6% for business hotels against 76.1% for city hotels and 76.6% for resorts, so city and resort properties are about 15pt higher at the stage far from the stay date. The more a hotel type relies on long-lead bookings from leisure, special occasions and inbound travel, the higher its curve starts.

Source: Compiled by HotelBank Editorial Team from MetroEngines Research (Tokyo, average of 31 stay dates in October 2026. Business N=541 properties, city N=52 properties, resort N=12 properties / 742 rooms. The resort sample is small, so day-to-day swings show up easily in the average)

As the stay date approaches, however, the order flips. Business hotels build up 34.2pt from LT45 to LT1 to reach 95.8%, arriving at the stay date at almost the same level as city hotels at 95.0%. City hotel pickup is only 18.9pt and resort pickup 10.5pt, and resorts stayed nearly flat at around 83.5% for the two weeks from LT21 to LT7. The resort cohort is small at 12 properties / 742 rooms, so one property adding or pulling allotment shows up easily in the average and should be discounted; still, the character of the type itself, "fills early, limited late demand," has been observed repeatedly in analyses of other prefectures as well.

Source: Compiled by HotelBank Editorial Team from MetroEngines Research (Tokyo, October 2026, average of 31 stay dates. Booked ratio estimated on an OTA listed-inventory basis)

Price movements also split by hotel type. Published prices rose about 20% for business hotels and about 15% for city hotels from LT45 to LT1, while resorts fell about 8%. One plausible explanation for resorts is that high-priced plans sell first, leaving relatively affordable plans in the last-minute inventory. Precisely because curve shape and price direction differ by hotel type, benchmarking against a cohort of the same type is a prerequisite when reading your own property's curve.

Use 1: Pace Management, Comparing Checkpoints with the Same Points Last Month or Last Year

The most basic use of a booking curve is to compare checkpoint values with the same checkpoints in the past. The proper comparison is the same checkpoint in the same month last year, but OTA listed-inventory curves have only been accumulated since March 2026 and last year's data is not yet available. Here, the same checkpoints in the previous month (September 2026, 550 properties / 79,533 rooms / 30 stay dates) are used as the benchmark to illustrate how to read them. September comes before the autumn high season and its demand level differs from October's, so rather than reading the gap in absolute levels as "strong or weak," treat this as an example of comparing curve shapes (which stretch grew).

Source: Compiled by HotelBank Editorial Team from MetroEngines Research (Tokyo business hotels. September 2026 N=550 properties / 30 stay dates, October 2026 N=541 properties / 31 stay dates. Cohorts are built month by month)

Source: Compiled by HotelBank Editorial Team from MetroEngines Research

The October curve sits above September's at every checkpoint, with gaps of 9.7pt at LT45 and 5.5pt at LT1. What deserves attention is how the gap narrows: it peaks at 10.5pt at LT30 and shrinks as the stay date approaches. Because October books earlier, its last-minute pickup (LT7 to LT1) is 7.7pt, smaller than September's 11.0pt. This is the information to read in pace management. For example, if your property's LT30 is above the same-type cohort's checkpoint, there is room to limit the inventory held back for the last minute and maintain rates; if it is below, that is grounds for bringing forward additional channels or plans. Once same-month data from the previous year has accumulated, simply use the same procedure with last year's same-month checkpoints as the comparison.

Use 2: Timing Rate Changes and Judging Whether Event-Day Demand Spills Over into Adjacent Days

When using a booking curve to decide on rate changes, what matters is the combination of "the booked ratio at a checkpoint" and "the sold-out property share at that point." In the example of Tokyo business hotels in October, LT14 showed a booked ratio of 83.1% with a sold-out share of 14.3%, and LT7 showed 88.1% and 18.3%. If your property's booked ratio is above these checkpoints and the market's sold-out share is starting to rise, it can be read as a phase where raising rates on remaining inventory is unlikely to drive demand away. Conversely, if the market is at a high checkpoint but only your property is low, the first fix is exposure (plan mix and channels), not price. Whether a rate change is right is not determined by the curve alone but by your position relative to the same-type cohort.

Another practical use is judging whether demand on an event day "spills over" into the surrounding days. On days with a marathon, a concert or a national tournament, the curve for that stay date alone runs high from an early stage. Lining up the curves for the night before, the event night and two nights before shows whether demand is concentrated on one day or spreads to adjacent days as multi-night stays. HotelBank has used this method to compare the race eve and race night of the Okayama Marathon 2026, the race eve, race night and two nights before the Shimonoseki Kaikyo Marathon, and day 1 and day 2 of Sanfair SAGA 2026. If demand concentrates on the eve and the event night is weak, that points to building the event-night pricing and plan messaging earlier.

The same tool works for comparing days of the week and holiday periods. Five prefectures where demand reverses between the Oct 24 and Oct 31 Saturday stays, an analysis comparing Citizens' Day Friday stays across four November Fridays from 45 days out, and the Toyota City curve toward the Korankei Momiji Festival in the autumn foliage season all read checkpoints based on the definitions in this article. An analysis splitting Fukuoka Prefecture's Oct 24 stay by property distribution shows how widely inventory absorption at individual properties is scattered behind the market-average curve, and is a useful reference for reading the average curve and the distribution together.

Common Misreadings: Points Beyond 45 Days and the Definition of OTA Listed-Inventory OCC

The first misreading is treating observation points earlier than 45 days out as "no bookings yet." With OTA listed inventory, properties commonly release sales allotments in stages, so remaining rooms far from the stay date include "allotments not yet listed" rather than reflecting "bookings that came in." For this reason the curves in this article are limited to LT45 through LT1. The data provider's note states: "Points with lead time beyond 45 days include observations made before allotments were loaded and may not represent booking progress. When using points beyond 45 days, always explain to users the number of properties covered, the observation coverage, and that the figures are estimates based on listed inventory. Use of lead times within 45 days only is recommended." HotelBank's analysis articles follow this policy and use only points within 45 days.

The second misreading is treating the booked ratio in this article as the same thing as the room occupancy rate shown by the Japan Tourism Agency or in a property's financial results. The vertical axis in this article is defined as follows: "Occupancy rate on an OTA listed-inventory basis = 100 − 100 × OTA listed remaining rooms ÷ total rooms. Covers a fixed cohort of properties that list at least 30% of total rooms on OTAs (the same set of properties for every stay date; the number of properties covered is shown as the count for each hotel type). It is an estimate based on how inventory sold on OTAs is being absorbed, and its definition differs from actual room occupancy (it counts unlisted rooms as absorbed, so it runs higher)." In other words, rooms not listed on OTAs (direct sales, corporate contracts, blocks held for groups) do not appear as listed remaining rooms and are therefore counted as absorbed. As a result, the figure runs higher than actual room occupancy for the same month. The room occupancy rate in the Japan Tourism Agency "Overnight Travel Statistics Survey" is based on actual results collected from properties and is defined differently. HotelBank has explained why the Japan Tourism Agency occupancy rate and the OTA inventory-based estimate for Okinawa were 27.5pt apart in the same month, so please refer to it as well when placing the two side by side.

Third, do not confuse the average published price with the settled rate. The price on the curve is "the average of listed inventory still unsold at that point," not the average of rooms that sold. The estimated settled ADR is a metric calculated by a separate procedure, shown in the definitions block of this article. Fourth, do not apply the market-average curve directly as a forecast for your own property. Curves change shape with hotel type, day of week and location, so comparisons should be limited to cohorts of the same type in the same area. Combining the curve with demand-side macro indicators such as JNTO "Visitor Arrivals Statistics" makes it easier to explain, from the composition of demand, why a curve's "starting point" is high.

Frequently Asked Questions

Summary

A booking curve is a booking-progress curve that plots lead time to the stay date on the horizontal axis and the booked ratio on the vertical axis, and it is the mirror image of the remaining-room ratio. Measured data for 541 Tokyo business hotels with 78,615 rooms in October 2026 show bookings building to 61.6% at 45 days out, 70.7% at 30 days, 83.1% at 14 days, 88.1% at 7 days and 95.8% the day before, while published prices rose about 20% and the sold-out property share climbed from 6.0% to 34.5%. In the same month, city and resort hotels started in the 76% range at 45 days out with little late growth, so the shape differs by hotel type. In practice, the curve supports pace management by comparing five checkpoints against the same checkpoints of a same-type cohort, rate-change decisions by combining checkpoints with the sold-out share, and judging whether event-day demand spills over into adjacent days. Keep two points in mind, that points beyond 45 days are not progress and that ratios on an OTA listed-inventory basis run higher than actual room occupancy, and HotelBank's prefecture-level and event-level analysis articles can be read on the basis of these definitions.

Related Articles