Nightly Rate Board: A Free Comparables Set

What is a nightly rate board, and why read it as a comparables set?#

A nightly rate board is the public grid of listings, asking rates and review counts that a booking platform shows for a destination and a date range, and on Gili Air it is a comparables set an allocator can open at no cost. The page is built for travellers. It attracts a retail audience that has no interest in underwriting. Read as data, it is a continuously refreshed list of competing inventory with asking prices, bedroom counts, amenities, ratings and review histories, available at no cost. This research found no broker report with that cadence for a small island.

The instrument has a precise vocabulary, and the vocabulary decides what the board can and cannot say. Average daily rate (ADR) is room revenue divided by rooms sold. Occupancy is rooms sold divided by rooms available. RevPAR is ADR multiplied by occupancy, and unlike ADR it counts empty nights. A rate board shows none of the three. It shows an asking rate for a future night, which is a price offered and not a price achieved, and it shows reviews, which are a lagging trace of completed stays.

Everything built from the board is therefore a reconstruction, and each step carries a grade. The asking rate is directional. The review count is a proxy, converted through an assumed review rate that no public source has validated for this island. The platform's commission and fee layers are documented by the platform and can be treated as measured. This note sets out the conversion, the corrections, and the points where the method fails.

The scope is deliberately narrow. The geography is Gili Air, the small island off Lombok in Indonesia. The platform is Airbnb, because its review mechanics and fee schedule are documented in its own help centre and in its annual filing. The method is applied to comparables. Victaura develops on Gili Air, as its public project page shows, and no performance figure for that project appears in this article.

Can a booking-site rate board give a market ADR for Gili Air?#

Yes, as an estimate with a stated grade, and not as a measurement. A rate board gives asking rates, not achieved rates. A market ADR is reconstructed by sampling comparable listings across dated windows, converting the displayed total to room revenue, and then testing the result against any achieved-rate evidence the principal can obtain. Until that test is run, the output is directional.

The sample design decides most of the quality. The principal fixes the stay length, the lead time and the party size. The rate for each comparable is then recorded at 7, 30 and 90 days out, and again for a three-night and a seven-night search. Medians beat means, because a handful of outlier villas distort a thin island market. Listings are logged by bedroom count and by type, and never pooled into one figure.

The displayed total is not host revenue, and the platform's own filing shows the gap. Airbnb defines gross booking value as inclusive of host earnings, service fees, cleaning fees and taxes. The same filing reports gross booking value of $91,273 million on 533 million nights and seats booked for 2025. That is roughly $171 per night or seat (Victaura arithmetic, global, including experiences and taxes, so not a Gili Air figure). Airbnb also reports that its ADR rose 3% in 2025. The useful part is the definition. A platform-level rate is a gross figure, and a model that treats it as room revenue overstates income.

Conversion to a room-revenue ADR is a subtraction sequence, and it should be written down. Start from the displayed total for the stay. Remove taxes. Remove any guest-side platform fee. Remove the cleaning fee, or carry it as separate revenue with its own cost. Divide by nights. What remains is comparable to the hospitality convention, where the figure is room revenue over rooms sold. Skipping a step produces a number that looks like ADR and behaves like a menu price.

Which nightly rate on the board is the real rate?#

None of them is the real rate, and the board shows only the nights that have not sold. Airbnb's calendar does not differentiate a booked night from a blocked one, a limitation Inside Airbnb states in its own methodology. A night that sold leaves the grid. The rates still visible belong to inventory that remained unsold at the moment of observation, so the observed asking rates over-represent prices that did not clear. The direction of that bias is probably upward. That is an inference, not a measurement.

Length-of-stay pricing splits the board into several boards. Weekly and monthly discounts, minimum-night rules and cleaning fees spread across fewer or more nights all change the per-night figure for the same villa. A three-night search and a seven-night search on one listing can return different nightly averages. The principal should log both and state which one enters the model.

Lead time matters as much as length. A rate quoted 90 days out is a plan. A rate quoted 7 days out is a reaction to unsold inventory. Repeating the observation at fixed intervals and keeping the series is the only way to see which listings discount late, and by how much. A single screenshot is an anecdote, and it should never sit in an investment paper as a data point.

Rate tiers should be separated from rate levels. On a small island, a handful of villas set the top of the board and a large number of simple rooms set the bottom. A blended median across the two describes neither. The comparables set for a villa product is the villa tier, filtered by bedroom count and by whether the listing rents the entire home, and nothing else.

What do review counts say about occupancy on Gili Air?#

Review counts say how many stays ended in a published review, and occupancy is derived from that count by two assumptions, neither of which has a verified public value for Gili Air. The first is the review rate, the share of stays that produce a review. The second is the average length of stay. Estimated stays equal reviews divided by the review rate. Estimated nights equal stays multiplied by length of stay. Occupancy is estimated nights over available nights.

The mechanics of Airbnb reviews make the proxy usable but lagged. According to the platform's help centre, guest and host each have 14 days after checkout to submit a review, and the review is published when both have submitted or when the 14 days expire. A published review therefore dates a stay to within roughly two weeks. A stay that ended last week may not appear yet, so the most recent month is always understated.

The simultaneous reveal is part of why review rates are not trivially low. A field experiment on Airbnb by Fradkin, Grewal and Holtz, published in Marketing Science in 2021, found that hiding feedback until both parties had reviewed stimulated more reviewing overall. That is evidence that the platform's design pushes the review rate up. It is not an estimate of the rate itself.

The public default parameters come from Inside Airbnb's published methodology. Where a city's average stay is unknown, it applies 3 nights per booking, and it uses the listing's minimum nights when that exceeds the average stay. It caps the occupancy rate at 70%, to control for hosts raising minimum nights in high season before reviews catch up, and to stay conservative. An academic paper on short-term rental regulation in Ireland restates the same three-night fallback, although its own improved model uses 4.6 nights taken from public Airbnb data, which shows the convention travels. It does not show that the parameters are right.

How far can the review rate move the occupancy estimate?#

On one hypothetical listing with 30 reviews, it moves implied occupancy from 34% to 81% before the 70% cap, a spread of 47 percentage points. Inside Airbnb documents three candidate review rates. It rejects 72% as unverifiable. It reports that a comparison of public review data against the New York Attorney General's October 2014 report on Airbnb pointed to 30.5%, which it describes as more fact-based but probably not conservative enough. It then chose 50%, which sits almost exactly between the two.

The table applies those rates to one hypothetical listing with 30 reviews in the last twelve months. The listing is illustrative. It is not a Gili Air listing and not a Victaura project. With a three-night stay, the same 30 reviews imply 34% occupancy at a 72% review rate, 49% at 50%, and 81% at 30.5%, which the 70% cap then truncates. The spread is 47 percentage points on identical input.

Stay length compounds the problem. At a 50% review rate, moving from 3 to 5.5 nights per stay, the figure Inside Airbnb quotes for San Francisco from Airbnb, lifts the same 30 reviews from 49% to 90%, which the cap again truncates to 70%. C9 Hotelworks puts the average stay of visitors to the three Gili islands at 2 to 3 nights (directional, no stated method, 11 August 2025). The head of the North Lombok Tourism Office told TTG Asia on 21 September 2026 that Gili Air guests often stay for weeks (press-reported, directional). Neither is an average for Gili Air alone, and the two sit far apart. The principal should treat length of stay as the second unknown and not as a constant.

The practical response is a range, not a point. Run the model at all three review rates. Report occupancy as a band. Use the band for downside testing, and use the cross-section of listings, ranked by review count, for relative performance. Ranking is robust to the review rate in a way that levels are not, provided review rates and stay lengths are similar across listings, because a constant review rate scales every listing equally.

How should seasonality be handled on Gili Air?#

Seasonality is the first correction, because a twelve-month review count hides the shape of the year. Reviews carry a date, and each falls within about two weeks of a checkout. Binning them by month gives the stay profile of each listing. An annual occupancy built from a flat count assumes that nights sold evenly across the year, and an island that depends on a tourist season does not behave that way.

Regional demand data frames the base but does not substitute for an island series. BPS Bali recorded 6,948,754 direct foreign arrivals to Bali in 2025, up 9.72% on 2024, announced on 2 February 2026, with Australia the largest source market at 1.63 million (measured, official statistics). That is a Bali figure. Gili Air lies off Lombok, outside Bali's statistical territory, so the number is regional context only. No verified island-level monthly series was found for this article.

The island-level evidence that does exist comes from a consultancy and should be graded accordingly. C9 Hotelworks, in a note published on 11 August 2025, reported 773,148 visitors to the three Gili islands in 2024, 26% above 2019, with Gili Trawangan at 541,199 or 70% of the total and Gili Air at about 25%. Applying that share gives roughly 193,000 visitors to Gili Air (Victaura arithmetic on a directional figure). The note does not name an underlying dataset or state a method.

A seasonal index can be built from the principal's own repeated observations and the review calendar. Take the share of annual reviews falling in each month, shift it by the average length of stay, and compare the shape with the asking-rate calendar. If rates peak in the months where reviews peak, the board is pricing demand. If rates stay flat while reviews swing, hosts are under-pricing the peak, or the review pattern reflects something else. Either reading is informative, and neither needs a broker.

How are villas separated from rooms in a Gili Air comparables set?#

Size is the second correction, and villas are separated from rooms by filtering on bedroom count, capacity, entire-home listing type and a pool. A one-bedroom bungalow and a three-bedroom villa with a pool can sit side by side in search results and be sorted by the same rating. They are different products, with different guest parties, different nightly rates and different cost bases. A per-listing ADR across both is not a market ADR.

Normalise by bedroom count and capacity, and report both per unit and per bedroom. The per-unit rate is what the guest pays. The per-bedroom rate is what allows a two-bedroom to be compared with a four-bedroom. The relationship between size and rate need not be linear, and the board lets the principal measure the curve for Gili Air instead of assuming it.

Quality tiers need an explicit rule. Photographs, review scores and amenity lists are the available signals. A pool, air conditioning, a kitchen and a private garden are binary tags and can be counted. Review scores on rental platforms are assumed to cluster near the top of the scale, which would limit their power to separate good from excellent (Victaura Research assumption, not measured here). Review count per year is the better signal of throughput.

Product type matters because the revenue model differs. Entire-home listings, private rooms and hotel-style rooms share a search page and not a revenue logic. The tier relevant to a private-residence product is entire-home listings with a pool. Hotel-room inventory belongs in a separate set, benchmarked on room-revenue terms with its own occupancy. C9 Hotelworks counted a hotel market of 364 establishments and 4,150 keys, or rentable rooms and units, across the three islands, with Gili Air holding about 31% of inventory, in a note of 11 August 2025 that does not date the count (directional, no stated method). A board that is mostly small rooms will describe that mix, not the villa tier.

What do platform commissions do to a Gili Air ADR?#

Commissions are the third correction, and they are the only layer that the platform documents itself. Airbnb's help centre describes two fee structures. Under the split-fee structure most hosts pay 3% and the guest pays a separate service fee. Under the single-fee structure the entire fee comes out of the host's payout, 15.5% for most hosts, 14% to 16% for others, and 16% in Brazil and Mexico. The platform states that the split structure is being phased out and that hosts who move to the single fee cannot switch back.

A board that mixes the two structures is not internally consistent. Under the single fee, the displayed rate has the platform's cost embedded, because no separate guest fee is added. Under the split, the guest pays the fee on top. Grossing a 15.5% deduction back up requires dividing by 0.845, an increase of about 18.3% (Victaura arithmetic), and not adding 15.5%. A listing that has migrated can show a higher headline rate with no change in host income.

Other channels add their own layers, and the principal should use contract numbers and not blog averages. Booking.com and other intermediaries charge commissions that depend on market, property type and programme participation, and the figures that circulate publicly come from third-party guides that disagree with each other. This article therefore cites no rate for them. A model takes the commission from the operator's actual contract, channel by channel, weighted by the channel mix.

Below ADR sit the costs the board cannot show at all. Management fees, cleaning labour, linen, utilities, local taxes, payment processing and the marketing cost of direct bookings all fall between the headline rate and net operating income. The correction is to carry ADR into the model net of taxes and gross of channel cost, and to apply a cost stack taken from operator quotes. The rate board supplies the top line. It does not supply the margin.

What does the Gili Air rate board not tell the principal?#

The weak points are structural, and they are disclosed here because they bound every number above. The review rate is the largest. The values discussed in Inside Airbnb's methodology range from 30.5% to 72%, the San Francisco Chronicle's 2016 methodology used 75% on Airbnb's statement that 3 out of 4 guests leave reviews, and none has been validated against booking records for Gili Air. Every occupancy level derived from reviews is directional, and the 70% cap means the method cannot tell apart listings that perform above that level.

Reviews can disappear and can be influenced. The platform allows an author to request removal of a published review within 30 days of publication, so a review count is not a monotone ledger. A field experiment by Fradkin and Holtz, released as a 2021 working paper, found that offering guests a coupon to review listings with no prior reviews produced more reviews, which shows that review volume is a variable the platform can move. Neither effect is quantified for Gili Air.

The board is a snapshot and covers one platform. Inside Airbnb states that its data is a snapshot of listings available at a particular time, that listings can be deleted, and that hosts may not keep calendars updated. A villa sold only through direct channels or another platform does not appear. The comparables set is the visible part of the market, not the whole of it.

The independent occupancy figures are also estimates. The C9 Hotelworks hotel occupancy numbers, 41.6% for Gili Air and 42% across the three islands for 2024, are directional figures from a private consultancy, published on 11 August 2025, with no named underlying dataset and no stated method. The author has not reconciled them with an official series, and no island-level government series was found that could be verified. The principal who needs a measured figure should obtain operator-level data, such as monthly booking records and payout statements, under a confidentiality agreement.

What does a defensible Gili Air comparables set look like?#

A defensible set is small, dated and written down before anyone looks at the result. It holds twenty to thirty entire-home comparables, filtered by bedroom count and pool, observed at fixed lead times and stay lengths, repeated at least monthly, with each assumption recorded with its date.

It reports bands, and it cross-checks against an independent figure. C9 Hotelworks estimated Gili Air hotel occupancy at 41.6% for 2024, in a note dated 11 August 2025. Under the 50% review rate and a three-night stay, a listing needs about 25 reviews a year to imply 41% (Victaura arithmetic, 25 divided by 0.5, times 3, over 365 nights). If the median comparable on the board sits far from that, either the review rate, the stay length or the C9 estimate is off, or the villa tier differs from a hotel-market average, and the principal should find out which before the model is used.

It states what it does not claim. It claims an asking-rate distribution and a review-implied occupancy band for comparable listings, on one platform, at stated dates. It does not claim an achievable rate for any specific project, including Victaura's own development on Gili Air, whose public project page is the only source this article uses for it.

The value of the rate board to an allocator is the discipline it forces, and the number it yields comes second. A free, daily comparables set turns an assumption into something that can be falsified next month. That is what underwriting asks of an input. Where the set is thin, the honest output is a wider band and an explicit request for operator data.

Skin in the game disclosure. Victaura, through its parent Greystone B.V. (Netherlands), develops projects on Lake Como (Italy), in Nungwi, Zanzibar (Tanzania), on Gili Air and in Uluwatu, Bali (Indonesia), and holds an off-plan capital position on Al Marjan Island, Ras Al Khaimah (UAE). Readers should assume commentary may be influenced by, or benefit, these positions. This document is classified as marketing material under MiFID II Article 24(3). It is not investment advice.