Running it: the cost stack between gross and net

A brochure quotes you two numbers: a nightly rate and an occupancy percentage. Multiply them and you get gross revenue. Everything in this guide is what happens between that figure and the money that arrives in your account, and there is more of it than most buyers expect – on the assumptions set out below, rather less than half of what a guest pays for the room survives to the owner before a single fixed cost is paid.

Stage 08 covers the licence, the operator and the tax at one level. This is the layer underneath: the individual lines, where each percentage comes from, which of them you can move, and where our data runs out.

Start at the booking value, not at the nightly rate

The first mistake is arithmetical. The rate a guest sees is not the rate you receive, and the platform takes its cut before anyone else does.

Airbnb publishes two fee structures. Under the split fee, the host pays 3 per cent and the guest pays 14.1 to 16.5 per cent of the booking subtotal on top. Under the single fee, the host pays the lot: most hosts 15.5 per cent, the rest typically 14 to 16 per cent. The single fee is not optional for everyone – Airbnb requires it of hotels, serviced apartments, and any host connected through property management software. That last clause catches almost every professionally managed Bali villa, because that is exactly how a manager runs a portfolio. If your villa is on a channel manager, assume the 15.5 per cent, not the 3 per cent.

We have not verified a Booking.com or Agoda commission rate for Indonesia. Booking.com's partner commission pages refuse automated access, and we are not going to publish a number we have not read. Ask your operator for the actual rate on each channel and the actual channel mix, and note that a mix of 60 per cent Airbnb and 40 per cent Booking.com is a different cost line from 100 per cent of either.

So the first line of your model is not USD 214 a night. It is USD 214 less the platform's share.

The four lines that are percentages

Four costs scale with revenue. Work in that order, on a notional USD 100 of booking value, and the structure becomes visible.

• Platform commission at 15.5 per cent takes USD 15.50. USD 84.50 reaches the operator's account.
• Management commission at 20 per cent of that takes USD 16.90.
• Running costs – cleaning turnovers, linen, consumables, guest-driven electricity and water, restocking – at 12 per cent take USD 10.14.
• Final tax on gross rent at 10 per cent takes USD 8.45.

USD 49.01 is left. That is 49 per cent of the booking value, before you have paid for anything that does not move with occupancy.

Two of those four percentages are ours rather than a published source's. Bali operators charge 15 to 25 per cent of gross and we use 20 as a midpoint; running costs at 12 per cent is an assumption we would defend but cannot cite. The platform fee and the tax rate are both published, and both are linked below.

The word that costs three points

Ask one question of the management agreement: is the fee charged on gross booking value, or on what reaches the account after platform commission?

On the same USD 100, a 20 per cent fee charged on gross booking value takes USD 20 rather than USD 16.90, and USD 45.91 is left instead of USD 49.01. Three percentage points of net, on one clause, on a contract most buyers sign without reading. It is the single highest-return question in this guide.

If your manager charges on gross booking value and you want to model it in our yield calculator – which takes a nightly rate already net of platform commission – gross the fee up. Twenty per cent of booking value is 23.7 per cent of what survives the platform. Type 23.7, not 20.

Tax: check which regime you are actually in

Indonesian income tax on the rental of land and buildings is a final tax at 10 per cent of the gross amount, under Government Regulation 34 of 2017, in force since 2 January 2018. Final means it discharges the liability on that income. Gross means it does not shrink when your costs rise. Our calculator charges it on gross for exactly that reason.

The part almost nobody mentions: that same regulation excludes lodging and accommodation services from the final-tax regime. A villa run as short-stay accommodation, particularly through a PT PMA, is plausibly an accommodation business rather than a land-and-building rental, and is then taxed on profit under the ordinary corporate regime instead. That is a different shape of liability, not a smaller one, and which side of the line you fall on depends on your contracts rather than on your intentions.

We are not going to tell you which regime applies to you. We will tell you that the two are not interchangeable, that our calculator can only model tax as a percentage of gross, and that if you are taxed on profit you should set the tax rate to zero in the tool and apply your own rate to the net figure it produces.

Separately, there is a regional consumption tax. Under Pasal 58 of Law No. 1 of 2022 on financial relations between central and regional government, PBJT on hotel and accommodation services is capped at 10 per cent of the amount paid by the consumer, with the rate set by each regency. It is borne by the guest, not by you – but you are the party who has to collect and remit it, and an operator who is not doing so is creating a liability in your name. We have not been able to open the Badung regulation to confirm the regency's actual rate, so we are not publishing one.

The lines that are not percentages

Everything so far falls when occupancy falls. These do not, and they are the reason a break-even occupancy exists at all.

• Staff. Badung's minimum wage for 2026 is IDR 3,791,002.57 a month, set by Governor of Bali Decision No. 1021/03-M/HK/2025 of 23 December 2025 and in force from 1 January 2026. At the Bank Indonesia JISDOR rate of IDR 17,727 to the dollar on 1 September 2026 that is about USD 214 a month, or USD 2,566 a year in wages alone for one full-time person, before the statutory holiday allowance and social insurance contributions on top. Denpasar is IDR 3,499,878.78 and Gianyar IDR 3,316,798.48, so where the villa sits changes this line. A villa with permanent staff has converted a variable cost into a fixed one, which is a service decision with a balance-sheet consequence.
• Utility standing load. Electricity for households above 3,500 VA is IDR 1,699.53 per kWh, roughly USD 0.096. A pool pump alone, at 750 W for eight hours a day, is 2,190 kWh and about USD 210 a year whether or not anyone stays. We have no dataset for total villa consumption in Bali and will not invent one; ask the developer or the operator for twelve months of actual meter readings on a comparable unit.
• Land and building tax, insurance, licence renewals, Indonesian accounting and, if you incorporated, company compliance filings.
• Replacement capital, which gets its own section because it is the line that is almost always missing.

Replacement capex is a cost, not a surprise

Short-stay use destroys interiors. Mattresses, sofas, outdoor furniture, air-conditioning units, pool equipment, linen stock and the pool surface itself all have a service life measured in single-digit years under weekly guest turnover, in a climate that is hard on everything.

A brochure that shows you a ten-year net income projection with no capex line is showing you a villa that is never refurnished. Reserve for it annually rather than meeting it as a shock in year five, and treat the reserve as a fixed cost in your model, because it is a number you set rather than a number occupancy sets.

We do not publish a benchmark reserve rate. We have not seen a dataset of actual refurbishment spend on Bali villas and we are not going to derive a percentage from nothing. What we do is ask the developer or operator for the actual replacement spend on their oldest comparable unit. If a developer has been operating for five years and cannot answer that, the answer is that they have not been reserving either.

What the market actually does

This is where a claimed occupancy meets a measured one.

AirROI aggregates public Airbnb listing data. For Canggu, over the twelve months to July 2026, it records 3,920 active listings, an average daily rate of USD 214, occupancy of 36.2 per cent, RevPAR of USD 79 and average annual revenue of USD 21,244. Supply fell 8.0 per cent year on year. Across the 30 Bali markets it tracks, the trailing twelve months to 8 August 2026 give average occupancy of 36.0 per cent and an average daily rate of USD 159.

The distribution matters more than the average. In Canggu, AirROI puts median occupancy at 36 per cent, the top quartile above 62 per cent and the top decile above 80 per cent, while the bottom quartile sits at 18 per cent. On revenue, the median listing takes USD 1,552 a month and the top decile takes USD 6,486 or more. A brochure quoting 80 per cent occupancy is quoting the top ten per cent of the market as though it were the expected case.

Read that data with four limitations in mind, because they are real.

• It is listing-level, not villa-level. A dormant listing, a part-year listing and a professionally run five-bedroom villa all count as one line.
• We do not know whether its average daily rate is before or after platform commission. AirROI does not say. We have assumed it is the gross listing price in the worked example below, because that is what a scraped listing page shows, but that assumption is worth about 15 per cent of the figure and it is ours rather than theirs.
• AirROI's public methodology page does not define occupancy, average daily rate or revenue, does not say which listings are included, and states no limitations. We asked; the page does not answer.
• Its own figures do not obviously reconcile. AirROI reports Canggu peak-season revenue of USD 3,332 a month and low-season revenue of USD 2,442 a month, which cannot average to its own headline of USD 21,244 a year – about USD 1,770 a month. The two must be computed over different denominators. Without a published methodology we cannot say which, and we are not going to guess in order to make the page tidier.

That is the honest position: the best public data available says the middle of the Canggu market runs at about 36 per cent occupancy at a USD 214 rate, and the same source contains an internal inconsistency it does not explain.

Seasonality and the void you have to fund

AirROI puts Canggu's peak in August, September and October at 47.9 per cent occupancy, and its low months as February, March and July at 36.7 per cent. Note what does not move: the daily rate sits between USD 197 and USD 203 across all three bands. In this market the season shows up in nights sold, not in price achieved, which means discounting into a low month buys volume at the cost of the rate you will find it hard to raise again.

The consequence for the model is a cash-flow one. Percentage costs disappear in a quiet month. Fixed costs do not. Staff are paid in February at the same rate as in September, and the reserve still accrues. Break-even occupancy – the number our calculator reports and almost nothing else does – is the figure that tells you how long a bad run you can absorb.

Putting it together

An example, with every input stated. The purchase price is a round number chosen for the arithmetic, not a market figure; we do not publish a Canggu price index and will not pretend to.

• Purchase price: USD 450,000. Assumption.
• Nightly rate: USD 214 gross, from AirROI's Canggu average for the twelve months to July 2026, less Airbnb's 15.5 per cent single fee. USD 181 net of platform commission.
• Management commission: 20 per cent of that. Midpoint of the 15 to 25 per cent range Bali operators charge.
• Running costs: 12 per cent. Assumption.
• Tax: 10 per cent of gross, being the final rental tax rate.
• Fixed annual costs: USD 9,500. Built as USD 2,570 for one full-time staff member at the Badung minimum wage, USD 1,300 for part-time pool and garden, USD 1,400 for standing utilities, USD 1,730 for land and building tax, insurance, licence renewal and accounting, and USD 2,500 as a replacement capital reserve. Only the first of those six has a published source.

At AirROI's Canggu market occupancy of 36.2 per cent, that villa lets 132 nights, takes USD 23,916 net of platform commission, and returns USD 4,371 – a net yield of 0.97 per cent, with 81.7 per cent of revenue consumed before it reaches the owner.

At top-quartile occupancy of 62 per cent it lets 226 nights, takes USD 40,960 and returns USD 14,257: 3.17 per cent net, 65.2 per cent leakage.

At top-decile occupancy of 80 per cent – the figure a brochure will quote – it lets 292 nights, takes USD 52,852 and returns USD 21,154: 4.70 per cent net, 60.0 per cent leakage, and a payback period of 21.3 years.

Break-even occupancy on all three is 24.8 per cent. Below that the villa loses money at any rate of letting.

The gross yield a brochure would print for the 80 per cent case, using the full USD 214 rate before the platform takes its share, is 13.89 per cent. The net is 4.70 per cent. That gap is not a marketing exaggeration; it is the cost stack, and every line of it is above.

Turn it round: the price is the variable you control

Occupancy, rate and cost percentages are mostly set by the market and the operator. The purchase price is set by you, and it is the denominator of every yield on this page.

Run the same villa the other way. At 62 per cent occupancy it earns USD 14,257 net, which supports a purchase price of about USD 238,000 at a 6 per cent net yield. At 80 per cent it earns USD 21,154, which supports about USD 353,000. Those are the prices at which the operating economics work at those occupancies, on these assumptions.

If the asking price is materially above that and the pitch rests on capital growth rather than income, that is a legitimate position – but it is a different investment, and it should be argued as one rather than dressed up as a yield.

What you actually control

• The price. The largest lever, and the only one entirely yours.
• The management fee and, more importantly, its base. Negotiate the base before the percentage.
• The channel mix, which sets your blended platform commission.
• The staffing model, which decides how much of your cost base is fixed and therefore what your break-even occupancy is.
• The reserve. Setting one does not change the cost of replacement; not setting one changes when you find out about it.
• Not the occupancy. Not the seasonality. Not the tax rate.

Put your own figures through the yield calculator at /guides/yield-calculator. It reports the gross-to-net waterfall, the leakage percentage, the payback period and the break-even occupancy, and it takes the nightly rate net of platform commission – so take the OTA's share off before you type it in.

Where our numbers stop

We have no dataset for villa-level utility consumption, staffing levels, refurbishment spend, insurance premiums or Indonesian accounting fees in Bali. Every figure of that kind on this page is a stated assumption and is labelled as one. We have not verified Booking.com or Agoda commission rates for Indonesia. We have not opened Badung's own regulation to confirm its PBJT rate.

Those sentences are the point. A page that gives you a confident number for every line is not better informed than this one; it has simply decided not to tell you which of its numbers it made up.