You Rank #1 on Google — But Users Have Stopped Opening Google

A while ago, I had a pretty specific question I needed to look up.

Out of habit I opened the browser, my fingers already typing into the search box — and halfway through, I stopped.

I thought for a second, closed that tab, and went to ask ChatGPT instead.

It gave me a paragraph with a few citations, and after reading it I had my answer. I never scrolled down a single page. I never clicked a single one of those blue links.

In that moment, something hit me:

The way people look for answers has changed.

Yet plenty of companies are still out there fighting tooth and nail for "page one of the search results."

You Can't Chart New Territory With an Old Map

Think about it: for the past decade-plus, how have we gone about "getting seen"?

SEO.

Make your website lightning-fast, pack in keywords until they're dense, plaster backlinks all over the internet — all to grab one of those ten blue links on Google.

But now, when users ask ChatGPT, ask Perplexity, ask Gemini, what do they get?

A synthesized answer, with a few cited sources attached.

Nobody scrolls anymore.

What does that mean? It means the SEO craft you've polished for over a decade for Google might land every punch on empty air in a world that doesn't even look at links.

A 2025 study laid this out perfectly. The researchers put ChatGPT, Perplexity, and Gemini — the three AI search engines — side by side with Google and ran a large-scale comparative experiment: across industries, across languages, and across multiple ways of rephrasing the query. They wanted to understand one thing clearly:

How do AI search engines actually pick their sources?

The results were pretty startling.

AI Search Plays Favorites

What do I mean by "plays favorites"?

It means who it chooses, and who it doesn't, shows a very clear bias — and the bias isn't as "reasonable" as Google's.

The single most important finding of the study, in one sentence:

AI search engines systematically favor "earned media" (independent third-party coverage about you, as opposed to content you publish yourself) — that is, when authoritative third parties write about you, rather than you writing about yourself.

This is completely different from Google. The traffic Google sends you comes from a relatively balanced mix: some to your official site, some to third parties, some to social platforms, all blended together.

But AI search? It trusts third-party authorities far more than the content your brand publishes itself — let alone social content.

To put it bluntly: if you want ChatGPT to mention you in its answer, just making your own website look beautiful won't get you very far.

It would rather trust the version of you in other people's mouths than the version in your own.

Sit with that for a second and it's actually pretty interesting. In some sense, AI search cares even more about "word of mouth" than Google does. It's like a finicky diner — it won't even glance at the menu the restaurant printed itself; it only trusts the reviews written by food critics.

The Three AI Search Engines Are More Different Than You Think

So are ChatGPT, Perplexity, and Gemini basically doing the same thing?

No. They differ a lot more than you'd think.

The study compared them head-to-head along several dimensions:

Domain diversity: how wide is the range of sources they cite?

Freshness: do they pull recent content, or do they dig up years-old stories?

Cross-language stability: if you ask the same question in Chinese and in English, do you get consistent answers?

Sensitivity to phrasing: change the wording, change the question — does the answer shift dramatically?

Each of these engines has its own personality.

Which means one thing: you can't win on every AI search engine with a single playbook.

What works on ChatGPT may fall flat when you take it to Perplexity. The strategy you build for the Chinese market may need to be rebuilt from scratch for the English one.

And Then There's the "Big-Brand Bias" Hurdle

What's even trickier is something else the study uncovered — I'll call it the "big-brand bias."

When AI search picks its sources, it has a natural tendency to favor brands that are already famous. That's a pretty frustrating fact.

So what about the niche players, the new companies, the small brands in vertical fields?

The researchers didn't treat this as a dead end. Their take: it's a problem that calls for a deliberate, targeted strategy — not something you solve by banging your head against the wall.

A "New SEO" Checklist for the AI Era

The study closes with four action recommendations. After reading them I found them remarkably clear, so let me put them into plain English for you:

First, make your content "machine-readable."

The articles you write need to let AI scan them and grab the key points at a glance — and lift a passage as a citation. That means clear structure, solid facts, and arguments that a machine can "slice off" and use directly. Gimmicks, empty claims, and marketing fluff — AI skips right past them.

Second, push hard on "earned media."

Get other people to write about you. Get authoritative third parties to write about you. Get people with real weight in your industry to mention you. Because AI search trusts them, not you.

Honestly, this has always been PR's job. It's just that before, PR was about getting seen; now it carries an extra layer of meaning — it's the source of your authority in the eyes of AI.

Third, differentiate by engine and by language.

Don't take the lazy route of one-size-fits-all. ChatGPT, Perplexity, and Gemini each need their own playbook. The Chinese market and the English market each need their own strategy. Spend a little extra effort, cover another class of engine.

Fourth, smaller players should actively crack the "big-brand bias."

Big brands have a natural advantage — don't go toe-to-toe with them. But in verticals, in the long tail, in those niche scenarios where the big brands can't be bothered to do the work carefully, smaller players actually have a shot at being picked by AI. The precondition: in that niche, you have to be specialized enough, deep enough, and recognized enough by third-party authorities.

This Playbook Has a New Name

The researchers gave this brand-new playbook a name:

GEO — Generative Engine Optimization.

I think it's a good name.

It's not an upgrade to SEO — it's a different species. SEO serves an algorithm's ranking; GEO serves an AI that reads, chooses, and cites.

What it serves is no longer "page one of the search results" — it's "that one sentence inside the AI's answer."

That might sound like a small difference, but it could well decide whether anyone sees your business in the next decade.

Writing this, I can't help but ask you one question:

Next time you open ChatGPT and ask a question about your industry — will its answer mention you?

If it won't, then right now is probably the moment to rethink how you "get seen."