A shortlist you never see
Picture a procurement manager in Penang who needs a freight forwarder. Five years ago she would have searched, opened six websites and compared them. Today she may type the question into an AI system and read one answer with three names in it. If your company is not one of the three, you were never considered, and you will never know.
Gartner predicted in February 2024 that traditional search engine volume would drop 25% by 2026, with search marketing losing share to AI chatbots and other virtual agents. Whether the figure lands exactly matters less than the direction. Generative AI tools are becoming answer engines for questions that used to go to search.
For most of my career, being found was a discipline with its own budget line. This is the same problem in a new place, and most of the old measurement does not reach it.
Why one look tells you little
The obvious response is to ask an AI system about your brand and see what it says. That tells you very little.
Answers vary. The same question asked on a different day, in a different country or to a different system can return a different list. One answer is an anecdote. You need the same questions asked the same way across several systems, with the date recorded.
The question matters more than the brand name. Buyers rarely ask about you by name. They ask which firms handle cold-chain shipments to East Malaysia, or who can audit a small manufacturer. The test is whether you appear in answers to the questions your buyers ask.
A failed request is also different from an absence. AI systems time out and decline to answer. A blank result should be recorded as a failure and never read as proof that the brand is unknown.
What the answers are built from
An AI system assembles its answer from what it has read and what it can retrieve. In practice that means your own website, directories, trade press, review sites and what other people have written about your category.
Three weaknesses come up often. The company's own site describes what it does in language no buyer uses. The facts about the company differ between its site, its directory listings and its social profiles. And nobody independent has written about it, so a system has nothing to cite.
None of these is exotic. They are the basics of being clearly described in public, and they now matter to a new kind of reader.
Treat it as measurement
The useful approach is the one a good researcher would take. Write down the questions a buyer would ask, in the buyer's words. Ask them of several AI systems. Record every answer with its date. Note where you appear, where competitors appear and which sources are cited.
Then fix what you control. Start with how your own site describes your services and whether the facts about you agree everywhere. Wait a few weeks and ask the same questions again.
Be careful with conclusions. A result is an observation on a date. It is not a ranking, and it cannot prove that a change you made caused a later mention. What it can do is show you a gap you did not know about.
Resist the temptation to game the answers. Publishing thin pages stuffed with the phrases you want repeated tends to age badly. Clear, consistent and specific descriptions of what you do are the durable fix.
What this means in Southeast Asia
The regional picture adds complications. Buyers here ask in English, Malay, Mandarin, Bahasa Indonesia and other languages, and the same system can give different answers to the same question in each. A brand that is well described in English may be much harder to find in Malay.
Markets differ as well. A company known in Kuala Lumpur can be absent from answers about Singapore or Jakarta, because most of what has been written about it is local. Measure one brand in one market at a time, in the language your buyers use, before drawing wider conclusions.
Smaller firms have an opening here. An AI answer has room for only a few names, and a specialist with a clear and specific description can appear beside much larger competitors.
Start with ten questions
You can begin this afternoon. Write the ten questions your best customers asked before they found you. Put them to three AI systems and save the answers with today's date. Read them as a buyer would.
For teams that want this done consistently, SkillPilot offers an automated product. The AI Visibility Review costs US$99 for one brand in one market and includes dated findings, ranked fixes and a recheck after 30 days. AI Visibility Tracking repeats the same questions every month for US$49. Both are software, with no consulting attached.
The brands that check now will have a year of dated evidence by the time their competitors start asking the question.