The demand side is real
Bali recorded 6,948,754 direct foreign arrivals in 2025, up 9.72 percent on 2024, with Australia alone accounting for roughly a quarter of all visitors. Cumulative arrivals through July 2025 were already running 12.46 percent ahead of the prior year. Early 2026 has softened, not collapsed: Bali's governor reported January to April 2026 arrivals down only 0.23 percent, with a sharper dip in May tied partly to global disruption, alongside hotel and restaurant tax receipts through May 2026 up roughly Rp300 billion year on year. Whatever the exact monthly wobble, the demand base underneath Bali tourism has not gone anywhere.
The wider Indonesia picture is stronger still, and matters because a share of Bali-bound travelers first land at another Indonesian gateway before connecting onward, arrivals a Bali-only count does not capture. Indonesia welcomed 13.98 million foreign tourists in the first eleven months of 2025 alone, up 10.44 percent year on year, and growth has continued into 2026: BPS reported 6.07 million foreign arrivals for January to May 2026, up 7.68 percent on the same period in 2025. Domestic tourism adds a further, largely separate demand layer: Indonesians made 1.09 billion domestic trips in 2025, up close to 19 percent on 2024, a real source of Bali visitation that foreign-arrival counts do not touch at all.
The supply side is where the story changes
AirDNA's public Bali market data, refreshed in July 2026, counts roughly 86,000 active short-term rental listings across Airbnb, Vrbo and Booking.com, at 47 percent average occupancy, a $146 average daily rate, and $12.8K average annual revenue per active listing. Revenue per listing was down 12.7 percent and ADR down 10.3 percent year on year, even though active supply grew only 0.8 percent. Read together, that is not a demand collapse. It is discounting pressure from a market that got more crowded faster than it got more visited.
That headline listing count also needs a footnote most articles skip: the raw number blends very different products, studios, single rooms, family villas, long stays, short stays, actively managed assets and passive listings left on autopilot. The islandwide number is a real saturation signal, but it is not the right comparison for any single project. The right response to "how many listings are there" is not the raw count. It is the count of genuinely comparable assets in your specific bedroom range, submarket and stay pattern.
What a realistic return actually looks like
Marketing material across the island still advertises 15 to 20 percent gross yields, and some resort-community products push projected figures even higher. Islandwide public data does not support that as an average outcome. AirDNA's own figure, $12.8K average annual revenue per active listing before host expenses, is the plainest antidote to brochure inflation available.
| Total project cost | Gross yield | Net, 25% opex | Net, 40% opex |
|---|---|---|---|
| $200,000 | 6.4% | 4.8% | 3.8% |
| $300,000 | 4.3% | 3.2% | 2.6% |
| $500,000 | 2.6% | 1.9% | 1.5% |
These are not Bali averages for a finished villa. They are what happens when you apply the islandwide average revenue figure to a real cost basis, an honest underwriting exercise, not a market forecast. A brochure quoting 18 percent is not lying about the number existing somewhere on the island. It is usually just not showing you the denominator.
If your return only works before tax, before maintenance, before staff, before licensing and before enforcement, it is not a return. It is a brochure.
The honest range: strong, compliant, well-operated assets in the right submarket can still reach high-single-digit net yields. Average product, bought at today's prices and run passively, more often underwrites to low- or mid-single-digit net returns once costs are counted honestly. Neither figure is a government statistic. Both are defensible reads of public data. None of this includes land value appreciation, a separate driver worth underwriting on its own terms rather than folding into an annual yield percentage.
That is also why we still build here. Harmonie's three villas average more than 80 percent occupancy against the 47 percent islandwide average cited above, one metric among several. Harmonie as a whole had already secured Rp 3.45 billion in 2026 revenue by 6 July, and is forecast to close the year 24 to 40 percent above 2025, double-digit growth in a market where average revenue per listing is falling. That gap is not luck. It is design, shared infrastructure, and service, and we explain exactly how in a separate note on our operating model.
Compliance is not a footnote anymore
Two regulatory notes on this site cover the mechanics in detail: what PP 28/2025 actually requires, and what PP 20/2026 changed about company tax eligibility. The pattern underneath both is the same. Indonesia has moved toward risk-based licensing with real supervision and real sanctions, and enforcement in 2025 and 2026 has stopped being theoretical. The clearest example: Bali authorities demolished dozens of businesses at Bingin Beach in mid-2025 after they were assessed as illegal, one of several visible actions tied to the island's broader push against unlicensed accommodation and overdevelopment, a pattern we trace in more detail in a separate note on apparent versus real profitability.
Land structure carries the same weight as licensing. Indonesian agrarian law does not allow a foreigner to hold Hak Milik, the strongest freehold title, directly. Serious structures use leasehold arrangements or a PT PMA company; anything resting on informal nominee logic is a legal-risk trade wearing a real estate costume, not a yield play, and Bali courts have already annulled nominee agreements outright in more than one case.
What actually separates a good outcome from a bad one
- Zoning and land-use match, checked against the parcel's actual spatial plan, not the color an agent uses in conversation, covered in full in our note on how Bali zoning actually works.
- Correct licensing from the start, sequenced into the build timeline rather than rescued after opening.
- A conservative occupancy assumption. Islandwide average is 47 percent. Underwriting 70 percent as a baseline is optimism, not analysis.
- Real operations, not a listing left on autopilot. The gap between the top and bottom of Bali's market is now mostly a gap in who is actually running the asset.
- An honest cost stack, with management, maintenance, tax and vacancy treated as real numbers before anyone talks about yield.
The reframe
The right question in 2026 is no longer "is Bali good." Demand answers that on its own. The right question is which Bali, under what structure, bought at what basis, run by whom. Islandwide averages are too blunt a tool for a real capital decision. The market has shifted from a place where almost anything worked to a place that quietly rewards operators and quietly punishes everyone else.
Related: What is a realistic net yield for a Bali villa? · How to legally rent a villa short-term in Bali · The 80,000+ listing problem · Can you still build in Bali?
Bring the land, the number, or just the ambition. We will tell you honestly whether the project clears a real bar, using our own operating data, not a projection.
Talk to FrançoisSources: Badan Pusat Statistik (BPS) Bali, 2025 and early 2026 arrivals; Indonesian Ministry of Tourism and BPS national foreign-arrival and domestic-trip statistics, 2025 to Q1 2026; Bali provincial government public statements, May 2026; AirDNA Bali market data, demand/supply/seasonality pages, checked July 2026; Colliers Q1 2026 Bali Hotel report; OSS Indonesia business classifications; Peraturan Pemerintah 28/2025 and Permenpar 6/2025; Undang-Undang Pokok Agraria on foreign land rights; Reuters and The Guardian reporting on Bali enforcement actions, 2024 to 2025. Figures move; this note reflects data checked as of the last reviewed date below and should be reconfirmed for any investment decision.
last reviewed: july 2026