Insights · AI search in Malaysia

Clarentino Aduk, Founder & Chief Executive Officer | October 2026 | 9 min read

Ask most Malaysian marketing teams about AI search and you will hear a reasonable answer: it is interesting, but it is not where our customers are yet. For a business serving local customers, the numbers support that view. For an organisation whose audience lives overseas, they do not.

This difference matters more than it first appears. Many of the organisations that shape how Malaysia is seen abroad, from tourism boards and hotels to universities, hospitals and investment agencies, are making decisions about AI search based on national figures that describe a different audience from the one they serve. The result is a quiet, growing gap between where their audiences are going and where their attention is.

This article sets out what the numbers show, why the gap exists, which organisations are most exposed, and what leadership can sensibly do about it now, without overreacting to a technology that is still changing quickly.

Across Malaysia, AI assistants such as ChatGPT still send well under 1% of the traffic that search engines send to websites. Google remains overwhelmingly dominant. Most Malaysian consumers still begin with a search box, and most Malaysian websites still receive almost all of their search visitors from Google.

Google itself is changing. Its AI Overviews summarise answers at the top of the results page, and AI Mode, which became available in Malaysia in October 2025 with support for Bahasa Melayu, offers a conversational search experience much closer to an AI assistant than to a list of links. Even within Google, the experience of searching is moving towards a single assembled answer. But measured by where website visitors come from, the national picture is still one of a search engine, not an assistant.

Our own websites tell a different story. They serve travellers planning trips to Sabah, most of them from Europe and Australia. On our three largest sites, one in four visitors now arrives from an AI assistant. On the largest, ChatGPT is the single biggest source of enquiries, ahead of paid advertising. On another, at a point when Google search had not yet produced a single enquiry, ChatGPT was producing about one in five.

The national average describes local audiences. It says very little about the overseas traveller, investor or student deciding whether to choose Malaysia.

These are not marginal differences that a few unusual weeks could explain. A share of under 1% and a share of 25% describe two different worlds. One is a world in which AI search is an emerging curiosity. The other is a world in which it is already one of the main ways people arrive.

The difference is not mysterious. Three forces explain most of it.

1. Adoption is fastest in the markets Malaysia wants to reach

AI assistants have been adopted fastest in many of the markets Malaysian organisations most want to reach: Western Europe, Australia, North America and parts of East Asia. In those markets, it has become ordinary to ask an assistant to plan a trip, compare universities, explain a visa process or summarise the investment incentives of a region.

The enquiries on our own websites reflect this. In ten weeks, travellers wrote from at least 29 countries, led by the United Kingdom, Germany, France, the Netherlands, Australia, Italy and Spain. These are audiences for whom AI assistants are already a normal part of planning.

2. Overseas audiences face an information gap

The second reason is more fundamental. Overseas audiences face an information gap. A traveller in Germany cannot easily judge which rainforest lodge is reputable or which operator is licensed. An investor in Japan cannot easily compare incentives across Malaysian states. A parent in Indonesia cannot easily tell which university programme is well regarded by employers.

When you cannot judge for yourself, you look for someone you trust to judge for you. For decades that meant a travel agent, a consultant, a guidebook or a friend who had been. Increasingly it means an AI assistant, which is available at any hour, speaks the person’s language and appears to have read everything.

The more unfamiliar the destination or the decision, the more valuable that recommendation becomes. That is precisely the situation of an overseas audience considering Malaysia.

3. Local audiences have other options

Local audiences are in a different position. They know the places. They have friends and relatives who have been. They can compare prices themselves, often in person, and they can usually find a cheaper option through word of mouth. They have less need for a trusted guide, so they have less reason to change how they search.

This is why the national average is misleading for some organisations. It is dominated by local behaviour, and local behaviour is shaped by local knowledge. An organisation serving local customers can reasonably follow it. An organisation serving international customers should not.

It would be easy to conclude that AI search is simply another channel, one that matters a little more for some organisations than for others. That would understate it, for two reasons.

A shortlist of one

A search page offers ten links, several advertisements and often a map. A person can scroll, compare and come back later. An AI assistant usually offers one recommendation, sometimes two, written as a confident answer. Organisations outside that answer are rarely considered at all.

This changes the economics of being found. In traditional search, ranking fifth still produced some visitors. In AI search, the difference between being named and not being named is close to the difference between everything and nothing.

Visitors who arrive ready to act

People who arrive from an AI assistant have usually already asked their questions and received a recommendation. They arrive later in their decision and with more intent. On our largest site, visitors who arrived from ChatGPT were about three times as likely to send an enquiry as visitors from Google search: 8.6% against 2.6%.

So even where AI assistants send fewer visitors than search engines, those visitors can be worth considerably more. Measured by enquiries rather than by visits, the shift is larger than the traffic numbers suggest.

For organisations whose success depends on international audiences, waiting for the national figures to rise is the wrong signal to watch. The most exposed share two features: their audiences are overseas, and those audiences find it hard to judge the options for themselves.

• Tourism boards, hotels, parks and operators. Travellers increasingly ask assistants where to go, how long to stay, how to get there and whom to book with. A destination described thinly or inaccurately loses visitors to places that are described well.
• Investment promotion agencies and economic corridors. Investors and their advisers use assistants to build early shortlists of locations and to understand incentives, approvals and costs. Being absent from that shortlist means being absent from the conversation that follows.
• Universities and colleges recruiting international students. Students and parents ask assistants to compare programmes, fees, rankings and graduate outcomes across countries.
• Hospitals and clinics serving medical travellers. Patients research specialists, procedures, costs and safety before they travel, and they want reassurance from a source they consider neutral.
• Government bodies. Public information about visas, regulations, heritage and safety shapes how a state or country is perceived abroad. When assistants repeat outdated information, the reputational cost falls on the government, not on the assistant.

For each of them, the first impression is increasingly formed by an AI assistant before the person ever visits a website. If that answer is incomplete, outdated or points elsewhere, the organisation may never know it lost the opportunity. Lost enquiries do not appear in any report.

There is a specific risk for Malaysian organisations that deserves attention. When AI assistants cannot find clear, complete and trustworthy information from an organisation itself, they often draw on intermediaries instead: international booking platforms, aggregators, directories and review sites.

Those intermediaries are frequently better at publishing structured, detailed information than the organisations they list. As a result, assistants may describe a Malaysian destination accurately but direct the traveller to book through a platform based elsewhere, or describe a Malaysian institution using a directory entry that is several years out of date.

When we began building our destination websites, this was one of the clearest patterns we saw. Much of the information available to overseas travellers about Sabah was scattered across platforms outside the state. The opportunity, and the commission, flowed to whoever published most clearly, not to whoever was closest to the experience.

For a state or country trying to capture more of the value of international demand, this is not a marketing detail. It is a question of who benefits from the interest that already exists.

A new technology tends to provoke two opposite mistakes, and AI search is no exception.

The first is to dismiss it because the national numbers are small. For organisations serving local audiences, patience is reasonable. For organisations serving international audiences, it means ignoring a shift that is already happening among the people who matter most to them.

The second mistake is to overreact: to commission a new website, buy software promising guaranteed AI rankings or hire an agency to produce large volumes of content written for machines rather than people. None of these address the underlying question, which is whether the organisation’s information is accurate, complete and trusted. Some actively make things worse, by adding thin or inconsistent material that assistants learn to ignore.

No one can honestly guarantee that an AI assistant will recommend a particular organisation. What can be done is to measure how assistants describe it today, fix what is wrong or missing, and track the change over time.

Leaders do not need to become experts in AI to respond sensibly. They need answers to three questions.

1. What do AI assistants say about us today?

Ask the questions your audiences actually ask, in the languages they use, across the assistants they are most likely to use. A tourism board might ask what to do in its state for a week, which areas are best for wildlife and which operators are reputable. An investment agency might ask where in Malaysia to locate a particular kind of facility and what incentives apply. A university might ask which Malaysian institutions are strongest in a field.

Read the answers carefully. Note who is named, who is not and where the information appears to come from.

2. Is it accurate?

Look for outdated facts, missing services, wrong contact details and recommendations that go to competitors or intermediaries instead. In our experience, most organisations find at least one significant error the first time they look, often one that has been repeated for years without anyone noticing.

3. Who owns this?

What AI says about an organisation is now part of its reputation. In most organisations, no one is responsible for it. Marketing teams own the website, communications teams own the press, and digital teams own advertising, but the assembled answer an AI assistant gives about the organisation falls between all of them.

Someone needs to own it, and that ownership needs support from leadership, because fixing it usually requires cooperation across several teams.

For an organisation serving international audiences, a measured approach over the first year might look like this:

1. Establish a baseline. Record what the main assistants say about the organisation and its sector today, in its key markets and languages.
2. Fix the most damaging errors. Correct inaccurate information at its source, starting with anything that affects trust, safety or the ability to make contact.
3. Fill the most important gaps. Publish clear, complete answers to the questions overseas audiences ask most, on the organisation’s own website first.
4. Train the people who maintain information. Give communications and customer-facing teams a working understanding of how assistants find and use information.
5. Measure monthly. Track how answers change and how many visitors and enquiries arrive from AI assistants, alongside existing channels.
6. Make sure enquiries are answered. Being recommended is only valuable if the enquiries that follow are captured, prioritised and followed up.

None of this requires a large budget or a new website. It requires attention, ownership and a willingness to look honestly at how the organisation appears to the people it most wants to reach.

<- All Insights

hello@mudahtemplates.com.my

+60 16-840 7481

Lot 25-3(R), 3rd Floor, Beverly Hills Plaza, Jalan Bundusan, 88300 Kota Kinabalu, Sabah, Malaysia