Lombok · public data · read 11 October 2026
392 Lombok properties.
Who you can book direct.
On one day in October we read the public presence of 392 hotels, villas and resorts across Lombok and the Gili Islands: the website, the way to book on it, its speed on a phone, and whether AI assistants name the property. These are the totals. No property is named.
What we read, and how
Public pages only.
Nothing behind a login.
Every figure comes from what a guest could see on 11 October 2026. We had no access to anyone’s systems, accounts or bookings, and where we could not see something from outside we marked it unknown rather than guess.
392 properties, from the map.
Hotels, villas and resorts listed on Google Maps across thirteen areas of Lombok and the Gili Islands: 168 resorts, 95 villas, 81 boutique hotels, 21 retreats, 18 glamping sites and 9 surf camps. The rating and review count come from the same public listing.
Can a guest book on it?
We opened the website on each listing and looked for a booking engine: a page on the property’s own site where a guest picks dates and books. A social page or an agent’s listing in place of a website counts as a stand-in, not a website.
How long the page takes.
Each website went once through Lighthouse, the open tool Google publishes, on a simulated phone. We report the time until the main image appears, which Google calls the largest contentful paint.
Who gets recommended.
We asked ChatGPT, Gemini and Claude 14 questions a traveller asks, such as the best boutique hotels in Kuta Lombok or the best places to stay on Gili Meno. Each question went to each assistant twice, in English, and each property was checked against the questions for its own area. Two of the 84 answers failed and were not repeated: ChatGPT’s second run for Sekotong and for Gili Meno.
Six findings.
One number each.
What the totals mean for the owner of a property in Lombok.
80 of 392
have no website at all. Another 30 rely on a social page or an agent’s listing instead.
A guest who searches the property’s name is likely to land on an agent’s page, so the first booking conversation happens on the agent’s terms.
169 of 392
have no page of their own that takes a booking: the 80 with no website, the 30 stand-ins, and 59 websites where we found no way to book. Of the 282 with a website of their own, 150 show a booking engine and on 73 we could not tell.
Nearly every booking those properties take goes to an agent at a commission or into a message thread that waits for someone to reply.
10.6 seconds
median time for the main image to appear on a phone, across 251 websites we could measure. 225 of the 251 took longer than 4 seconds, the point Google treats as poor.
Guests compare places on a phone, often with several tabs open, and the page that has not loaded is the one they close.
87 of 392
were named by at least one AI assistant for their own area: ChatGPT named 62, Gemini 52 and Claude 42. All three agreed on 21.
A traveller who asks an assistant gets a short list of names, and a property outside it is competing for a guest who never saw it.
40% against 5%
of properties with 1,000 or more reviews were named by an assistant (18 of 45), against those with fewer than 50 reviews (4 of 83).
The reviews a property earns on its listings appear to travel into the assistants’ answers, so they shape who gets recommended as well as who gets booked.
21 of the 141
best rated, 4.7 stars or higher on 50 or more reviews, have no website at all, and 13 more have a website where we found no way to book.
These are the places guests rate most highly, and the commission on their bookings is paid on a guest the reviews had already won.
The figures
The totals,
in one place.
The same figures as a file: lombok-2026-summary.csv
What this
cannot tell you.
It is one day’s reading
Everything was read on 11 October 2026. Websites, booking links, ratings and assistant answers change, so a property counted one way here may count another way now.
The list is not every property in Lombok
The 392 are what Google Maps returned for our searches by area and type. Homestays and guesthouses were mostly outside those searches, and a property with no map listing is not here.
A booking engine was judged from outside
We looked for a way to book on each website as a guest would. On 73 websites we could not tell, and we left them out of the 169 rather than guess. We counted by the website on each map listing, so the 169 holds a few that can be booked direct: two stand-ins link a booking engine from a social page, and one property whose listing shows no website has a site of its own with one. Five whose website is a bare booking-engine page count as websites of their own. A property may take bookings well by message or email; this reading only shows whether its own page takes one.
Speed is a single run
Each site went through Lighthouse once, on a simulated phone. A score moves with the hour and the connection, and 31 of the 282 websites could not be measured at all.
Assistant answers vary
The answers change with the wording, the language, the day and the version of each assistant. We asked through each company’s own command-line tool rather than the consumer app, which can answer differently. We matched names by reading the answers, so a property named in an unusual form may have been missed. Finding 05 shows that review volume and being named go together; it does not show that one causes the other.
No private data, and no verdict on any owner
We saw no one’s accounts or bookings. An owner who prefers to sell through agents may be right for their property; these figures show where a booking can go, not what it costs any one owner.
If one of the 392
is yours.
We can show you what we saw on your own pages. The free AI Visibility Check runs the assistant questions for one property. To look at the whole booking path, request a review.
Public information only. No passwords or system access.