Customers Are Now Asking AI Instead of Google — Where Does a Business with Only FB/IG Pages Stand? (AEO for SMEs 2026)

Every friend group has that one person who knows everything — which AC repair shop is reliable, which sign maker in the area does clean work. They rattle off two or three names on the spot, and everyone just calls those without searching further.

These days, a large number of people are switching to asking AI instead of that friend.

And this new friend has one habit worth knowing: they only know businesses that have written about themselves somewhere readable. No matter how skilled a shop is, if they've never written anything in a place this friend can read, the name simply won't come up.

Try a Quick Three-Minute Test Before Reading On

Open ChatGPT or Gemini and type a question the way a stranger who doesn't know you would ask — for example, "recommend a vinyl sign shop near Ratchaburi" or "which beauty clinic in Sukhumvit is good?" Plug in your business category and location, then see if your name shows up.

If it doesn't appear at all, let's be clear: that doesn't mean your shop isn't good.

Shops that customers deeply love, where regulars have been coming back for ten years, with a full queue all month — they also don't show up. Because this has nothing to do with skill. It's entirely about "where" — where your expertise is stored.

Why Doesn't This New Friend Know Your Business Yet?

When AI recommends a business, it doesn't make things up. It pulls text from web pages it can access and read, then rearranges that into an answer.

Now think about where your best information lives on your page.

Prices are in beautiful posters the graphic designer made. Terms are in comments you replied to three months ago. The deepest details are in private chat — where you answer best.

All three are readable by people, but not a single one is readable by machines.

Posters are images, not text. Comments flow down and disappear with posts. Chats are private spaces no one else can enter. So all the knowledge of your business — prices, service areas, actual hours — is stored in formats humans can see but machines cannot.

The "Message Us" Problem

The approach of not posting prices publicly works — I won't argue that. It draws people into chat very effectively, and once you're talking, conversion rates are much higher.

But there's one invisible side effect: when there are no numbers written anywhere and someone asks AI "roughly how much does this kind of work cost?", the answer it gives is pulled from whoever wrote their answer as text.

It's not that we're perceived as more expensive. It's that we're not in that conversation at all.

One Page for Your Business, But Multiple Answers Needed

Think about a shop that handles both new installations and repairs. These two customer groups ask entirely different questions. One asks about price per square meter, the other asks how quickly you can come assess the job.

A website can separate these into different pages, so AI picks the page most relevant to each question. A Facebook page has one main feed with old posts flowing down. Even if accessible, no single post completely answers any one question on its own.

The Most Commonly Overlooked Point

Many people tell me "search your shop name on Google and the page comes up" — and that's true, it does.

But the people who can type your name already know you exist.

The questions that bring new customers are questions with no trace of your name in them — "which one is good," "roughly how much," "who can do rush jobs?" This is precisely where websites can go where pages can't — and it's the exact space where AI is now stepping in to answer instead of Google.

The Numbers That Led Me to Bring This Up with Every Client

I don't want you to believe me just because I say so, so let me share verifiable numbers — and I'll tell you where each comes from, because context matters for interpretation.

On the user scale: OpenAI disclosed in February 2026 that ChatGPT has over 900 million users per week. Google stated that its AI Overviews answer boxes are seen by over 2 billion people per month across 200+ countries and 40+ languages. The Gemini app crossed 1 billion monthly users in August.

On behavior — this is US market data: SparkToro analyzed real-click data from Similarweb in early 2026 and found that around 68% of Google searches end with no click at all, up from 60% in 2024. Ahrefs, analyzing 300,000 keywords, found that when an AI answer box appears, the top-ranked result gets roughly 58% fewer clicks.

The finding I find most interesting is BrightLocal's consumer survey from March 2026, which found that people using AI to find local businesses grew from 6% to 45% within a single year.

For Thailand: the best available reference is a SCB EIC survey from July 2026 finding that 72% of Thais have used Generative AI, with around 30% using it regularly. Statcounter shows ChatGPT holds about 74% of chatbot usage in Thailand, followed by Gemini at 16%. DataReportal reports Thailand has 67.8 million internet users, representing 94.7% of the population.

There's no zero-click research for Thailand equivalent to the US data, so I won't claim our rate is the same 68%. But having followed these trends for years, this kind of behavioral shift typically arrives here one to two years behind other markets.

The Other Side of the Data That Deserves to Be Told

There's research analyzing over 100,000 queries that found global brands are mentioned in 73% of relevant AI answers, while small or niche brands are mentioned in only about 11%.

Skimming past it sounds like bad news. But I actually love this number because it clearly shows how to play this game.

If we compete on broad national queries like "where to build a website" — we lose before we start. But when the question narrows to "rush sign making near Photharam" or "machine shop handling urgent jobs near Samut Sakhon" — big brands don't come down to play there. And in most Thai-language niche categories, no one has claimed that territory yet.

"Business Is Still Going Well Right Now"

This is the phrase I hear most often. And every word of it is true. I never argue with anyone about this.

The only thing I'd invite you to consider is: this month's sales tell yesterday's story, not tomorrow's.

The typical sequence is: search behavior changes first. People who don't know us yet gradually can't find us. But existing customers keep coming back, so numbers stay steady. By the time you see the numbers shift, it's when the existing customer base has thinned — usually one to two years too late.

So what I suggest watching instead is the proportion of new customers who reach out because they "found you on their own" versus those from referrals and paid ads. If the first number keeps shrinking while totals look fine, that's a signal that arrives before the numbers do.

It's like a well-located shop where a new road is about to be built. Sales are still good today. But the map people use is about to change.

On the risks of building a business on someone else's platform — pages getting suspended, ad costs rising every year, not owning your own customer data — I've written about that separately at Facebook Page Alone: Is It Enough? 5 Business Risks in 2026. This article focuses only on being discoverable.

Why Those Who Start Earlier Have a Real Advantage

I don't like the phrase "you need to hurry" without a reason behind it. So let me explain why this game is different from the traditional SEO we're familiar with.

First, available spots are shrinking fast. The old Google page had ten results and people would scroll to positions five through eight. An AI answer is one paragraph that mentions three or four names. Being third versus eighth in the old world just meant a traffic difference. In the new world, it means being mentioned versus not existing at all.

Second, most Thai-language categories are genuinely unclaimed. Try asking specific local questions in Thai and answers are often vague, mentioning large national companies or pulling from directory sites with decade-old data. No one in your category is yet a reliable enough source for AI to reference.

Third — and most important — this kind of credibility compounds, and you cannot buy it back in time.

What makes AI confident enough to mention a specific business is having the same information about that business consistently appearing across multiple sources over time — on their own website, on Google Business Profile, in directories, in news, in reviews. You can't rush this with money. Someone who starts today will have a 12-month track record in a year. Someone who starts then will be starting from zero. This is the only kind of advantage competitors genuinely cannot buy back.

What Kind of Website Does This New Friend Remember?

Good news: you don't need a 20-page corporate website. What AI needs to pick up and answer from is clarity far more than impressiveness.

Separate pages by service — don't stuff everything on the homepage. Different customer groups ask different questions, and a good answer must come from the most relevant page.

Write prices as text — even a range is better than nothing. If you can't give exact numbers, write a starting price or what most jobs typically fall within, and explain what drives prices up or down. We do this ourselves: TumWebSME's website states directly that we start from 79,000 THB, because we want people researching to get an answer before needing to message us.

Create a FAQ page from your actual chat history. This is the easiest and most effective method. Go through the last 50 chats and find the most frequently repeated questions. That's your FAQ. Your customers have already written it for you.

Keep your identity consistent everywhere. Business name, address, phone number — must be spelled identically across your website, Google Business Profile, page, and LINE. Even small inconsistencies make the system uncertain whether we're the same entity, and it will choose silence over risking a wrong answer.

Never let important information live only in images. Beautiful posters are fine — just make sure the same information also exists as text alongside it. Prices in images are prices search systems will never see.

As for Schema Markup — add it, it helps with clarity and search features. But set the right expectations, because Google writes in their own AI Search guidelines that no markup guarantees being picked up by AI. What actually works remains text-based content that's accessible and genuinely answers questions.

Another thing few people know: many new websites are configured to block AI bots from day one without the owner realizing. If your goal is to be recommended, make this clear with your web team from the start — that you want AI crawlers to be able to read your content.

Pages and websites aren't competing with each other. Pages are still great at attracting people and closing sales. Websites serve as the place where you've told your full story — ready for both Google and AI to use. They have different jobs, and you should have both.

If you'd like to go deeper on the technical side, read more at What Is AEO and How Is It Different from SEO and GEO? and AI Search-Ready Website Checklist 2026.

If You're Starting This Month, Where to Begin

I like starting with what can be done right now, without any budget.

• Week 1 — Measure your starting position. Write 10 questions the way a customer would ask, then run them through ChatGPT, Gemini, and Google. Note who gets mentioned. Screenshot it as your before-baseline.

• Week 2 — Gather real questions. Go through past chats and pick the 15 most repeated questions, along with the answers you've actually typed out before. This is nearly all the raw material for your website.

• Week 3 — Launch your first website version. No need to wait for perfection. A homepage, separate service pages, a pricing page or price range, a FAQ page, and a contact page with info matching all your channels — that's enough to start.

• Week 4 — Connect all the signals. Update Google Business Profile to match the website, add your website link on your page, Instagram, and LINE, then submit a sitemap to Google Search Console.

How Will You Know It's Working?

Don't measure by feel. Here are three things I recommend.

First: bring the same 10 questions back every start of the month and count how many times your name is mentioned. This is the most direct method and costs nothing.

Second: ask your web team to check server logs for bots like GPTBot, PerplexityBot, or ClaudeBot visiting. If none are showing up, something is still blocking them.

Third: ask every new customer how they found you. The day someone answers "I found you on Google" or "I asked AI and it recommended you" — that's the day you know the machine is running.

Closing Thoughts

Facebook and Instagram pages are still good tools. I've never once recommended anyone stop running their page.

There's just one thing I want to leave with you: pages were designed for people to read — not for machines. And the entity doing the job of answering "which one is good?" for customers today is changing faster than most of us realize.

The businesses that will have an advantage in the next two years don't necessarily need to be the best in their category. They're usually the ones that have told their story in places machines can read — before everyone else. And right now, in most Thai-language categories, that space is still unclaimed.

If you want to know whether your category is genuinely open, reach out. Tell us a bit about your business and we'll help you ask AI together — see what it says now, who it mentions, and if you want to be in that answer, where to start.

Follow TumWebSME

Stay updated on website building and online marketing knowledge at:

• Facebook: TumWebSME รับทำเว็บไซต์ธุรกิจ

• Instagram: @tumwebsme

• TikTok: @tumwebsme

• YouTube: TumWebSME

Contact Us

• 088-983-9386 (Khun Ploy)

• 099-856-3198 (Khun Saennan)