Watercolour illustration: EasyFound's blue mascot, Arrow, stands on the five-foot-way of an old shophouse row in a Borneo town, holding a small map and pointing at a kopitiam and an aircon repair shop.
AIResearchMalaysia

Does ChatGPT know businesses in Sabah and Sarawak?

EasyFound · · 7 min read

TL;DR

Mostly, yes. In our test, ChatGPT named local businesses for most questions asked from Kota Kinabalu, Kuching, Miri and Sandakan, roughly as often as in our earlier KL, Penang and JB test. What changed by city was where it read about them: own websites in Kuching, TripAdvisor and directories in Sandakan. A few picks were in the wrong city. Give ChatGPT a clear page that says which town you serve.

Mostly, yes: in our test, ChatGPT named local businesses in Kota Kinabalu, Kuching, Miri and Sandakan for most everyday questions we asked, roughly as often as in our earlier KL, Penang and Johor Bahru test. What changed from city to city was where it read about them, and how often it slipped.

Sabah and Sarawak are home to nearly one in five Malaysians, according to the Department of Statistics Malaysia. So a customer in Kuching asking ChatGPT for an aircon man is not a rare case. In our test, 18 of 20 answers named at least one local business.

Does ChatGPT recommend local businesses in Sabah and Sarawak cities?

ChatGPT recommended local businesses in all four Borneo cities we tested. We asked five chatty questions with no town named, such as "My son has a toothache and he's quite scared of dentists. Can you recommend a good dentist near me?", with the location set to each city.

Every search it wrote included the set city, and none of the 20 answers asked which city the customer was in. Of the 70 businesses it named, 63 were in the right city, going by the answer or the page it cited.

How ChatGPT answered five everyday questions in four Borneo cities (EasyFound test via OpenAI's API, 9 October 2026, one answer per question per city; CSV)
Customer's city Answers that named a local business Businesses named
Kota Kinabalu 5 of 5 18
Kuching 5 of 5 19
Miri 4 of 5 20
Sandakan 4 of 5 13

Miri's total is lifted by one tuition answer that listed nine centres. Each city got one answer per question, so treat each row as an example, not a city score.

Is ChatGPT's knowledge of Borneo businesses as good as KL, Penang and JB?

On three questions we can compare, ChatGPT named businesses in Borneo roughly as often as in our earlier peninsula test. We reused three questions from our KL, Penang and JB test: aircon servicing, kitchen renovation and SPM tuition. There it named a business in 26 of 27 answers. In the four Borneo cities, it was 11 of 12.

That earlier test asked each question three times and didn't require a web search, though every answer searched anyway, so treat this as a rough comparison.

The gaps were in the smaller cities. From Sandakan, the SPM tuition question got advice and no names. From Miri, the dentist question got this: "I wasn't able to reliably identify a specific local Miri dentist from the search results I found just now."

Our reading: there is less for ChatGPT to find online in some towns and trades, so a business that is easy to read about stands out more there.

Where does ChatGPT read about businesses in Kota Kinabalu, Kuching, Miri and Sandakan?

ChatGPT's sources for Borneo businesses changed a lot by city and trade. About half of the 71 links ChatGPT cited went to businesses' own websites. The rest went mostly to directories, TripAdvisor and Google Maps.

  • Kuching: 15 of 20 links were businesses' own sites. All five dentists it named were cited from their own pages.
  • Sandakan: 2 of 15 links were own sites. Aircon firms came from directories and brand service-partner lists, and the dentist from a Google Maps address.
  • Cafés: for breakfast in Miri and Sandakan, every link was TripAdvisor. In Kuching, two café sites were cited directly.

No answer cited a Facebook page directly. If you rely on one, read whether a Facebook page helps ChatGPT.

Does ChatGPT recommend businesses from the wrong city in Sabah and Sarawak?

ChatGPT sometimes offered a business from the wrong city, and tuition was where it happened. We checked a sample of cited pages by hand.

  1. Kuching tuition: two of the three centres it named give addresses only in Selangor, around Klang, on their own pages.
  2. Miri tuition: one pick was a page titled for Miri on a tuition platform whose centres are in the Klang Valley, with classes run online.
  3. Pages that list every town: a renovation firm whose website lists almost every town in Malaysia was offered in both Sandakan and Miri. We couldn't confirm a branch in either.

A national aircon firm with a branch page for Kota Kinabalu was named there, as it was in KL, Penang and JB. Its general page was offered "near Miri" too, though it lists no Miri branch. Our reading: town pages work for ChatGPT whether or not you have premises there, so say plainly where you really are.

What can't this test tell me about ChatGPT in Sabah and Sarawak?

How do I get ChatGPT to recommend my business in Sabah or Sarawak?

In our test, ChatGPT recommended what it could read about online in each town. In some Borneo towns and trades that is a short list, which leaves room for a business that is easy to read about.

  1. Request EasyFound's free check. See what ChatGPT says about your business when customers in your town ask, then go through it with us on a free call.
  2. Put your town and address on your own website. Name the town and the areas you cover, such as Penampang or Petra Jaya. In Kuching, most of what ChatGPT cited were businesses' own pages.
  3. Fix the listings it reads. In Sandakan and Miri it leaned on TripAdvisor, directories and Google Maps, so check your entry on each is current. Here's how a Google Business Profile feeds ChatGPT.
  4. Don't claim towns you don't serve. Town pages can get a business named where it has no premises. That can win you a mention, then lose you the customer.

EasyFound is an AI SEO service for Malaysian businesses, from the Klang Valley to Kota Kinabalu. We check how you're named across six AI tools in English, Malay and Chinese, then work on the pages they read. Start with the free check below.

Does ChatGPT know your business in Sabah or Sarawak?

Customers in Kota Kinabalu, Kuching, Miri and Sandakan ask ChatGPT too. See what it says about your business, then go through it with us on a free call.

Frequently asked questions

Can ChatGPT recommend businesses in Kota Kinabalu or Kuching?

Yes, in EasyFound's test it did. We asked five everyday questions, such as finding an aircon servicing company or a dentist, with the location set to Kota Kinabalu and then Kuching, through OpenAI's API with web search. Every one of those ten answers named at least one local business. All but two of the businesses named were in the right city, going by the answer or the page it cited. The two exceptions were Kuching tuition picks whose own pages give addresses only in Selangor, around Klang.

Does ChatGPT know fewer businesses in Sandakan and Miri than in KL?

A little fewer, in EasyFound's test. ChatGPT named a local business in 4 of 5 answers each for Miri and Sandakan. For Kota Kinabalu and Kuching, it was all five. Asked about SPM tuition from Sandakan, it gave advice and no names. Asked about a dentist from Miri, it said it couldn't reliably identify one. On the three questions shared with our earlier KL, Penang and Johor Bahru test, it named a business in 26 of 27 answers there. That test repeated each question and didn't require a search, and one answer per city on one day can't rank the cities.

Where does ChatGPT find businesses in Sabah and Sarawak?

It depended on the city and the trade in EasyFound's test. About half of the 71 links it cited went to businesses' own websites. Most of the rest went to directories and listing sites, TripAdvisor and Google Maps. In Kuching, 15 of 20 links were businesses' own sites. In Sandakan, 2 of 15 were. For breakfast cafés in Miri and Sandakan, every link was TripAdvisor.

Does ChatGPT ask which town I'm in before recommending a business?

Not when it already has a location. In EasyFound's test, with the location set to a Sabah or Sarawak city, none of the 20 answers asked which city the customer was in. Every search ChatGPT wrote included the set city, and 13 answers offered to narrow the list to the customer's neighbourhood.

Can ChatGPT recommend a business from the wrong city?

Sometimes. In EasyFound's spot-check, three tuition picks were elsewhere: two centres offered to a Kuching parent give addresses only in Selangor, around Klang, on their own pages, and one offered in Miri was a town page for online classes run from the Klang Valley. A renovation firm whose website lists almost every town in Malaysia was offered in both Sandakan and Miri, and we couldn't confirm a branch in either.

Sources

Our data: EasyFound, Does ChatGPT know businesses in Sabah and Sarawak? Five questions with no place named (aircon servicing, kitchen renovation and SPM tuition, worded exactly as in our 1 October KL, Penang and JB test, plus a dentist for a nervous child and a breakfast café), asked once each through OpenAI's API with web search required, model gpt-5.4-mini, default settings, no chat history, on 9 October 2026. The API's approximate user location was set to Kota Kinabalu or Sandakan (Sabah) and Kuching or Miri (Sarawak): 20 answers in all. We read every answer and took each business's location from the answer or its cited page, and checked a sample of cited pages by hand. Directories, platforms, schools and government pages aren't counted as businesses. The KL, Penang and JB figures come from the 1 October test, not a new run, which asked each question three times without requiring a search. Download the per-answer results (CSV). We don't name the businesses, because one test isn't a ranking.

Other sources:

  1. Department of Statistics Malaysia, Current Population Estimates, Malaysia, 2024: Sabah 3.7 million (11.0%) and Sarawak 2.5 million (7.4%) of Malaysia's population, read 9 October 2026 (link)
  2. OpenAI, Searching the web with ChatGPT, Help Center (link)

The results are shared under CC BY 4.0: you're free to reuse them, with credit to EasyFound.

Research by Janice Pang, EasyFound.