When Customers Stop Googling You and Start Asking ChatGPT: What Should You Do?

A few days ago, a friend of mine who runs an enterprise-services business was venting to me.

He said they'd been doing SEO for years, and their keyword rankings had always been steady. But lately he'd noticed something — the very first question new clients asked on the phone had changed.

It used to be, "I found you on Baidu" (China's dominant search engine).

And now?

"I checked with ChatGPT, and it recommended you."

He was completely baffled. Why would ChatGPT recommend me? It's not like I'm paying them for ads.

I told him, hold on. This is actually kind of interesting.

The Way Customers Find You Has Changed

Think about it — how do you do your homework before buying something?

Five years ago, you'd open a search engine, type in a few keywords, scroll through the first couple of pages, and click through them one by one. The search engine ranked them for you, and you picked in order.

And now?

You just open an AI assistant and throw the entire question at it. "I run a cross-border e-commerce company. I need a system that can handle inventory, do reconciliation (matching the books between two parties), and ideally connect to the Amazon API (the interface that lets two pieces of software talk to each other). Budget under 200,000 RMB. Recommend a few options."

Then the AI takes ten seconds and hands you back a paragraph. It names three or four brands, each with a one-line reason for the recommendation.

Tell me — are you still flipping to the second page of search results?

Nope.

The entry point for customer decisions is migrating from the "search box" to the "chat box."

This shift sounds small, but for businesses the impact is brutal. Because the ranking logic has changed.

The old search-engine logic was: whoever's content best matches the keywords, whoever has the most backlinks, whoever has the highest authority — ranks first. Do those things well, and your traffic was assured.

The new AI logic is: whoever's content is easiest for me to read, whoever most directly answers the user's question, whoever is most persuasive — gets cited.

One is about ranking; the other is about citation. The two mechanisms sit a single word apart in Chinese, yet the game couldn't be more different.

Someone even gave this new game a name — GEO (Generative Engine Optimisation). I won't get hung up on terminology; let me just tell you three real stories first.

A Web Agency Gets 10% of Its Traffic from ChatGPT

The first story.

There's a company that specialises in building websites for businesses using Webflow. Webflow is a site-building tool — you can think of it as the more flexible, more developer-leaning cousin of WordPress.

They noticed something: more and more of the clients finding them weren't coming from search engines. They were coming from AI tools like ChatGPT and Perplexity.

A client asks inside the AI, "Any recommended Webflow development agencies?" — and the AI recommends them.

So what did they actually do?

You might not believe it, but what they did was remarkably simple.

They rewrote the content on every single page of their website. On every page they honestly spelled out three things: what we do, whose problems we solve, and what makes us good at it.

No flashy marketing spin. Just plain talk. Whether you're an AI or a human, you understand it at a glance.

Then on the technical side, they added some structured markup (schema code that helps AI systems "read" how the content relates).

And the result?

10% of their total traffic now comes from citations by ChatGPT and Perplexity. And of that 10%, 27% turned into qualified sales leads.

10% doesn't sound like much? Think about it — this is a traffic source that didn't exist at all before. And it's traffic arriving with a specific need, so the conversion rate is far higher than generic traffic.

A Building-Materials Supplier Sees Traffic Jump 67%

The second story is even more striking.

A building-materials company that supplies construction contractors. Their product information was rock-solid, but it was all written the old-fashioned way: product spec sheets, product catalogues, industry news — all organised by product category.

How does a contractor ask an AI when buying building materials?

He doesn't ask, "Please give me the technical parameters for your C30 concrete." He asks: "For the floor of an underground parking garage, what material gives the best value for money? What should I watch out for during construction?"

See — he's asking a complete question, carrying a real scenario with it.

This company realised that. So they flipped their entire content strategy on its head.

Instead of organising content by product category, they reorganised it around "the customer's questions." How to choose materials, how to do the construction, cost comparisons between different approaches — all written up as detailed guides. Every page is written "answer first, then expand" — exactly mirroring how a user poses questions inside an AI tool.

They also put work in outside their own website. Expert articles in industry media, joining discussions on Reddit, posting how-to videos on YouTube.

The result?

Organic traffic up 67%, traffic value up 400%, and appearances in Google AI Overviews up 540%.

400% and 540%. Those two numbers made me pause.

Building materials — an industry that seems worlds away from AI. Just by switching to a different way of organising content, they pulled off growth this dramatic.

A PropTech Company Sees Results in Six Weeks

The third story.

A PropTech (property technology) SaaS (software-as-a-service) company whose clients are mostly large property operators.

They also noticed that the way clients look for systems had changed. Instead of typing in two or three keywords to search, clients were now describing complete usage scenarios and requirements inside AI tools.

What this company did: they completely redid their help-centre pages and system-integration guides. For every usage scenario they wrote detailed documentation — not marketing language, but real, hands-on operational explanations. The goal: make it so the AI tool understands it at a glance — and gets the recommendation right.

Six weeks. Just six weeks. Of their new qualified client leads, 32% came from AI search tools.

Six weeks.

What does that mean? It means this isn't something you have to lay the groundwork for over three or five years. Get your content right, and the AI tools will "see" you very quickly.

It's Not Just Case Studies — Academia Has Proven It Too

The three stories above are the results companies are seeing in practice. Academia has started studying this too.

In 2024, a paper analysed a large volume of interaction data between AI search engines and web content. The conclusion?

Pages optimised for AI reading habits are 58% more likely to be cited by AI than unoptimised pages.

The researchers found that AI models prefer content that is semantically rich, logically coherent, reads naturally, and conveys expertise.

In plain terms — AI loves a "good article" too.

So What Should You Actually Do?

Three stories told. You'll notice that the core logic behind what these three companies did is the same.

Let me lay out a few key moves for you.

First, speak human. Make clear who you are.

Every page of your content should let a reader who knows absolutely nothing about you understand, within thirty seconds: what you do, who you help, and on what grounds. It sounds simple, but a huge number of corporate websites — you go in and read for five minutes and still have no idea what they actually sell.

Second, organise content around questions, not products.

Customers ask questions — so organise by question. Someone buying building materials asks "what to use for an underground garage," so you write a "guide to choosing flooring materials for an underground garage." You don't just pile every product into a catalogue and make people fend for themselves.

Third, give the answer first, then expand.

When an AI tool lifts a snippet of your content, what it wants is that single most direct answer. Put the answer up front, supplementary information behind. Humans find it comfortable to read, and it's easy for the AI to lift your answer too.

Fourth, build a presence outside your own website too.

AI tools don't only read your official site. They read industry forums, read YouTube, read Reddit, read all kinds of public content. Your participation and contributions in these places will all be "seen" by the AI.

And there's one more thing that's especially important.

How do you measure results?

Don't just fixate on "how many times the AI mentioned us." That number looks impressive, but it isn't necessarily useful.

What you should really be tracking is: are more people coming in to enquire? Are more people downloading your materials? More people filling out forms? More repeat customers?

The end goal of AI visibility isn't being cited — it's being trusted, being chosen.

A Final Word

That friend who called me later told me he went back and read through his company's website, and found the very first sentence on the homepage was: "Empower enterprise digital transformation, build a new smart-business ecosystem."

He said, "I don't even know what that means myself."

I said, "Then what makes you think ChatGPT knows?"

He laughed.

You see, a lot of people feel that the AI era has arrived and they need to learn some new technology, adopt some new tool.

But the truth is, you might not even have the most basic things right yet.

Explain what you do, in plain human language. So that a complete stranger — whether a human or an AI — can understand it.

That, no matter the era, will never go out of style.