Lessons from Bali: the occupancy gap

Bali nearly tripled its foreign arrivals between 2009 and 2025, to a record 6.95 million. It built rooms, restaurants and agencies faster than the visitors came, so its hotels are barely fuller than they were in 2000.

For Sri Lanka, which reads Bali as a case study more than any other destination, that gap between arrivals and occupancy is the lesson.

By Dilanke Panagoda · June 2026

Bali has spent the past two years as the world’s shorthand for too much tourism: the gridlocked road out of Canggu, the queue for the temple photograph, the moratorium on new hotels and villas that Jakarta proposed in September 2024 and has argued about ever since.

Underneath it sits a longer record. BPS-Statistics Bali Province, the island’s statistics office, has been counting tourism for more than half a century: every foreign arrival since 1969, hotel occupancy by star class every month since 2000, and censuses of hotels, rooms, beds, restaurants and travel agencies by regency.

The numbers carry a finding the congestion footage doesn’t show. Growth on this scale never made the island’s hotels any fuller.

Foreign arrivals nearly tripled between 2009 and 2025. The occupancy line barely moved. The sections that follow are how those two fit together; the last one is about why that should interest Sri Lanka.

01

The boom everyone knows

The arrivals series begins at 11,278 foreign visitors in 1969 and crossed one million in 1994. Across fifty-six years only three setbacks are legible at the chart’s scale: the year after the 2002 Kuta bombing, the year after the 2005 one, and the pandemic.

The pandemic was the deepest. Arrivals fell 83.0% in 2020, and then to fifty-one people for the whole of 2021, the year Indonesia kept its borders shut.

The recovery was as steep as the collapse: 5.27 million in 2023, 6.33 million in 2024, then 6.95 million in 2025, the most Bali has ever recorded and 10.7% above the pre-pandemic peak of 2019. Nearly all of it (99.4% in 2025) enters through a single door, Ngurah Rai airport.

02

Room occupancy barely moved

When 1.41 million foreign visitors came in 2000, Bali’s hotels were a little over half full. When 6.95 million came in 2025, they were no fuller in any way that mattered.¹ Across the twenty years before COVID, annual occupancy never left a narrow band.

The only times it dipped to the bottom of that band, the cause was a bomb rather than a soft market: the years after the two Kuta bombings. COVID broke the floor completely. Then the recovery climbed back into the same band and stopped.

A flat line through a quintupling of demand is not a sign that nothing happened. It is a sign that something very large happened on the supply side of the ledger.

One caveat belongs in the body rather than a footnote. The occupancy survey covers classified, star-rated hotels only, and part of why the line never rose is that demand kept leaking into everything the classification misses: villas, guesthouses, and lately unregistered rentals. That points the same way: the non-classified places BPS does survey ran emptier still, in the thirties through every pre-pandemic year on record.

03

What absorbed the growth

Supply did. Between 2000 and 2024 the count of classified hotels quintupled, from 113 to 593,² and the growth ran upmarket: thirty 5-star properties became 129, the 3-star tier grew six times over, the 4-star tier nearly ten.

Rooms at classified hotels went from 17,027 to 61,094,² beds from 29,892 to 86,082. The non-classified estate, the guesthouses and small inns beneath the star system, nearly tripled, to 3,582 establishments and 45,395 rooms.

The commercial layer on top of the rooms grew faster still. Bali’s registries listed 1,339 restaurants in 2012 and 10,448 in 2025; travel bureaus went from 334 to 3,347, ten times over, across thirteen years in which arrivals grew 2.4 times.³

The registries grew straight through the pandemic. In 2021, the year fifty-one foreign visitors entered Bali, the restaurant register added more than six hundred entries.

Length of stay completes the arithmetic. The average guest at a classified hotel stayed 4.1 nights in 2000, 2.84 in 2019 and 2.81 in 2025, close to a third fewer than at the start.

Hold the pieces side by side and the flat occupancy line stops being a puzzle. Visitors multiplied, each one bought fewer nights, and the stock of rooms grew faster than the nights did.

04

Nine per cent of the island

Badung, the regency holding Kuta, Canggu, Seminyak, Nusa Dua and Uluwatu, is most of what the world pictures as Bali. With the city of Denpasar beside it, the two cover 9.5% of the island and hold the shares the chart shows: nine in ten classified hotel rooms, nine in ten travel bureaus, three in four restaurants.

That concentration is old, not new. Badung’s share of the island’s classified rooms was 79% in 2024 and 76% in 2000; almost all the new building went up where the building already stood.

The strip is what the moratorium argument is really about: not whether Bali is full, but whether a tenth of it is.

05

2026: the trend holds

The first four months of 2026 averaged 55.6% occupancy.¹ The same months of 2000 averaged 52.5%. Twenty-six years and nearly five times the visitors separate those two numbers, and the average hotel’s spring is about three points fuller than it was, no more.

Arrivals are not running away either. January to April brought just over two million foreign visitors, 1.1% behind the same four months of 2025, with February and March ahead of last year and January and April behind. Month for month, occupancy tracked 2024 almost exactly.

06

Eleven times the land

Bali would fit inside Sri Lanka eleven times over, yet in 2025 it drew 6.95 million foreign visitors onto that one eleventh of the area: 1,202 for every square kilometre, against Sri Lanka’s 36.

Sri Lanka’s tourism debate is still conducted mostly in arrivals: how fast the country could climb back above its 2018 peak, which it did in 2025 at 2.36 million. Bali is the comparison that debate reaches for most often, used both as a case study and as a leading indicator.

It started earlier and grew larger, with one airport, a global name, and fifty years of compounding demand. And it kept the records to show what that growth did: more than fifty years of arrivals, twenty-five years of occupancy figures, and repeated censuses of the hotels, rooms and restaurants built to hold the visitors.

That record is one Sri Lanka can read: how fast arrivals grew, how fast the supply behind them expanded, and what occupancy did in response. This piece reads a single column of it, the occupancy gap.

Hotels are barely fuller than in 2000, at 60.6% against 56.4%. The average guest stays a third less time, 2.81 nights against 4.1. The rooms that share the demand have tripled, the restaurants have grown nearly eightfold, the licensed operators tenfold, and nine tenths of the trade is pressed into a tenth of the island, where the government is now debating whether to stop the building.

Occupancy is one column. Bali’s record holds the others too, the cultural, environmental, social and infrastructure impact of carrying that many visitors on that little land, each its own lesson.

Sri Lanka, with eleven times the land and a thirtieth of the density, is early enough to read all of them, and to ask whether the visitor count is the number worth chasing or only the easiest one to read.