Insights · Case notes

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

Between 1 August and 14 October 2026, our destination websites received more than 280 unique enquiries from travellers planning trips to Sabah. They came from at least 29 countries, led by the United Kingdom, Germany, France, the Netherlands, Australia, Italy and Spain.

We built those websites to test a simple idea: that overseas travellers, increasingly guided by AI assistants, would reward clear and complete information about places that are hard to research from abroad. The enquiries confirmed it. They also taught us a great deal that we did not expect, much of it about what happens after a person decides to get in touch.

This article sets out what we learned. Some of the lessons are specific to tourism. Most apply to any organisation that receives enquiries from people overseas, whether those people are travellers, students, patients or investors.

It is worth describing briefly what we built, because the lessons depend on it.

Our websites are destination guides. Each is devoted to a single place in Sabah and written to answer, in depth, the questions travellers actually ask about it: how to get there, how many days to allow, what wildlife they might see and when, what accommodation is available, what it costs and what to expect on arrival. The facts are kept consistent and the pages are structured so that search engines and AI assistants can read them reliably.

Travellers who want to go further can send an enquiry from the page. Every enquiry is handled through our Enquiry Desk and passed to licensed Sabah tour operators, who prepare quotations and manage bookings and payment. We do not sell tours ourselves.

So the websites sit at a useful vantage point. We see where travellers come from, what they ask and how they behave between their first enquiry and their final decision.

August brought 29 enquiries. September brought 179. In the first two weeks of October there were 76 more. An increase of roughly six times from one month to the next is welcome, but it exposes every weakness in how enquiries are handled.

In August, a handful of enquiries a week could be managed informally. Someone could read each one, reply personally and remember who had been answered. By September, with several enquiries arriving most days, that approach stopped working. Replies were slower, details were easier to miss and it became harder to know, at any moment, which travellers were still waiting.

The lesson for organisations planning for AI visibility is to plan for success. AI assistants do not increase their recommendations gradually and predictably the way a marketing campaign does. If an assistant begins naming you in its answers, the extra demand can arrive quickly and without warning. The process for handling it needs to be in place before it is needed, not built in a hurry afterwards.

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%. Over the period we measured, ChatGPT produced about a third of all enquiries on that site, more than paid advertising.

On a second site, the contrast was sharper still. At a point when Google search had not yet produced a single enquiry, ChatGPT was already producing about one in five.

“Someone who has already asked an assistant for a “recommendation arrives with a shortlist of one.

The explanation, we believe, lies in where each visitor is in their decision. A person who searches on Google is often still exploring. They open several results, compare them and leave to think. A person who arrives from an AI assistant has usually already asked their questions, read a summary and received a recommendation. When they click through, they are checking that the recommendation is credible and looking for a way to act.

The implication for organisations is significant. AI assistants may send fewer visitors than search engines, but those visitors are often much further along. Judging AI search by visitor numbers alone underestimates it. Judging it by enquiries, applications or bookings gives a truer picture.

The spread of countries surprised us more than the volume. We expected interest from the United Kingdom and Australia, which have long-standing connections with Sabah. We did not expect at least 29 countries within ten weeks, with continental Europe so strongly represented.

Germany, France, the Netherlands, Italy and Spain together accounted for a large share of enquiries. Many of these travellers wrote in English, but not all, and those who did often wrote in a second language. That has practical consequences. Replies need to be clear and simple. Quotations need to state plainly what is included. Assumptions that a local or British traveller would share cannot be taken for granted.

For organisations in other sectors, the lesson is that AI assistants can open markets you have not deliberately targeted. An assistant answering a question in German or Dutch may draw on English-language information about Malaysia and recommend it to someone you would never have reached through advertising. The organisation needs to be ready to respond to that person well.

Our forms recorded more submissions than unique enquiries: 338 submissions against 284 unique travellers over the period. In other words, about one submission in six came from someone who had already been in touch.

Travellers wrote again for many reasons. Some found a second page that interested them and enquired there too. Some had an additional question or a change of dates. Some were planning a longer trip that combined several destinations.

Without a check for repeats, the same traveller can receive two different answers from two different people, sometimes with two different prices. Nothing undermines trust faster, and it happens at exactly the moment the traveller is deciding whether to commit.

We now identify repeat enquiries automatically before anyone responds, so that each traveller is handled as one conversation, however many times they write. This is a small piece of process, but it is the kind that only becomes visible once volume grows.

A traveller arriving next week needs an answer today. One arriving next year can wait a little. Treating enquiries in the order they arrive misses the point.

In the early weeks, enquiries were answered roughly in sequence. As volume grew, that meant an urgent request sometimes waited behind several that could easily have waited. For a traveller with a fixed itinerary and limited time, a slow reply is not an inconvenience. It is a reason to book elsewhere.

Our Enquiry Desk now flags urgent enquiries by travel date and sends a daily alert listing the travellers who need attention first. The rule is simple, but it changed the order of work in a way that matters to the people waiting.

The same principle applies in other sectors. A student facing an application deadline, a patient with a scheduled procedure or an investor with a board meeting next week each has a clock running. The organisation that recognises it responds better.

Most enquiries do not become bookings after a single reply. Over the period, the largest group of enquiries were those that had received a quotation and were still deciding. Travellers compare options, consult partners and family, check flights and come back later, sometimes weeks later.

In that interval, it is easy for an enquiry to go quiet. The traveller assumes the operator will follow up. The operator assumes the traveller will decide in their own time. Neither acts, and the opportunity fades.

A structured follow-up, sent to every traveller who has received a quotation, keeps the conversation open without relying on anyone’s memory. It need not be pushy. A short, helpful message offering to answer further questions is usually enough. What matters is that it happens every time, not only when someone remembers.

We also send a thank-you message to every traveller once they have been in touch, confirming that their enquiry has been received and explaining what will happen next. It costs nothing and removes a common source of anxiety for someone dealing with an unfamiliar organisation in another country.

This lesson was the most uncomfortable, and perhaps the most useful.

Analytics that accurately record where enquiries come from are easy to neglect when a website is new. On one of our sites, our analytics recorded no enquiries at all over the period, while our own enquiry records showed more than a hundred. The enquiries were real. The tracking was not working, and for a time we had no way of knowing which sources were producing them.

Without reliable measurement, you cannot tell whether AI assistants, search engines or advertising are producing results. You cannot decide where to invest. And you cannot defend the investment to anyone else. We learned to check that every enquiry is being recorded correctly before we look at any other number.

For organisations beginning to take AI search seriously, the same applies. Before setting targets or commissioning work, make sure you can see where visitors and enquiries actually come from, including AI assistants, which many analytics setups do not separate clearly by default.

Overseas travellers are taking a risk when they enquire with an organisation they have never dealt with, in a country they may never have visited. Many of the questions they asked were not really about the trip. They were about trust: is this operator licensed, is the price fair, what happens if plans change, who will meet me at the airport.

The websites that answered these questions clearly, before they were asked, produced more confident enquiries. The replies that answered them plainly produced more bookings. Small signals, such as a named contact, a clear explanation of who handles payment and a prompt acknowledgement, did more to build confidence than any amount of promotional language.

This is also, we suspect, part of why AI assistants recommended the sites. Information that answers a traveller’s real concerns clearly is exactly the kind of information an assistant can rely on and quote.

Looking back over the ten weeks, three changes would have helped from the start. First, we would have built the full Enquiry Desk before launching the websites, rather than adding its parts as volume grew. Second, we would have tested the tracking on every site, with real test enquiries, on the day each one went live. Third, we would have agreed with our operator partners, in writing and in advance, how quickly each kind of enquiry should be answered and who would follow up if a traveller went quiet.

None of these is complicated. Each would have saved time, protected a few enquiries that waited longer than they should have and given us a clearer picture sooner. We now treat them as standard for any new site or client engagement.

The specific numbers belong to tourism, but the pattern is general. Any organisation that serves overseas audiences and wants AI assistants to recommend it should expect three things.

1. The demand, if it comes, will come quickly. Prepare the process for handling enquiries before investing in being found.
2. The people who arrive will be further along in their decision. They deserve fast, clear and accurate replies, and they will judge the organisation on them.
3. Most of the value is won after the first reply. Repeat checks, prioritisation by urgency and consistent follow-up turn interest into outcomes.

These are not technical problems. They are questions of process and discipline, and they are well within the reach of any organisation that decides they matter.

<- All Insights

How we help organisations ->

hello@mudahtemplates.com.my

+60 16-840 7481

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