The Search Era Is Over. The Era of Being Remembered Has Begun.

A friend of mine who runs a brand recently took me out for a late-night dinner.

He sat down, barely touching his chopsticks, and let out a sigh. "Mr. Liu," he said, "I've got a budget I just can't make add up."

"Go on," I said.

"Most of next year's budget is riding on search," he said. "But the more I think about it, the less sense it makes. When people have a question these days, they just open ChatGPT or Perplexity and ask. Who still scrolls through pages of results? Is this budget even worth spending?"

I didn't answer right away. Instead, I walked him through how search got to where it is today.

1. What Is SEO?

For more than twenty years, there has been only one way to get seen online: get your page to rank ahead of the search results.

Want to buy a bag? You type in a brand name and the search engine throws a few dozen links at you. Land in the top two pages and you feast; fall behind and you don't even get the scraps.

That single word — "ranking" — grew an entire industry around the world. Some people obsess over keywords, some build backlinks, some optimize content structure… Agencies, engineers, media buyers — generation after generation piled in, until they'd fattened it into a market worth more than $80 billion.

That industry is SEO.

The essence of SEO is simple: get your website to the top of the results page.

That logic held steady for more than two decades.

2. But in 2025, Search Started to Change

In 2025, Apple made an announcement: it would build AI-native search engines like Perplexity and Claude straight into its browser.

What does "AI-native search" mean?

You no longer face a search box. You open your phone and just say something to a voice assistant; it lays out an entire page of answers for you — complete with sources.

Let that sink in for a second. What does this really mean?

For the past twenty years, most people's door to information opened inside Google's building. Now a new door has swung open right next to it, and the key is plain language — no training required, usable the moment you pick it up.

Google isn't going to collapse overnight. But the default-entry grip it has held for two decades has been pried open a crack.

The sharpest-nosed investors — Silicon Valley's a16z among them — started repeating one word over and over:

GEO.

3. What Is GEO?

GEO stands for Generative Engine Optimization.

Sounds like SEO with an "AI" prefix, doesn't it? If that's all you see, you're underselling what's new here.

Let me put it in plain language:

SEO gets you ranked high on a results page. GEO gets AI to name you directly when it answers a question.

One is standing in line; the other is being chosen. No matter how high you rank, you're still just a name in a list. Being named is how you actually become part of the answer.

That's the "being seen" every business is really after.

Why do I say that? Because three rules have quietly changed.

Rule One: How People Ask Has Changed

With the old search engines, people typed around four words and hit enter. Now you ask AI in full sentences — twenty-three words on average.

Four words. Twenty-three.

Back then, you'd click a link, glance at it for a few seconds, and move on. Now you go back and forth with AI for six minutes on average.

In the Google era, the contest was who made the most attractive page. In the GEO era, it's whose content gets woven into the model's answer.

So what kind of content does AI prefer? Content that is clearly structured, easy to parse, and dense with information. Ask it to "summarize this" or lay your key points out as a list, and the model can easily pull you into its answer. The old trick of stuffing keywords? It hardly works anymore.

Rule Two: Who Pays Has Changed

How did the old engines make money? Advertising.

Search "backpack," and the top result belongs to whoever paid the most in bid fees. Users contribute their attention, businesses contribute money, and the platform collects the toll. That three-way relationship fattened trillion-dollar companies.

AI doesn't play that game.

ChatGPT charged a monthly subscription from the start. It doesn't make money by cramming a screenful of links at you; it makes money by answering well enough that you're willing to keep paying.

So it has no incentive to force a link you never asked for into its answer.

Then why does it "cite" outside content at all? For one reason only: your content genuinely helps that answer — it fills out the experience and backs up what the model is saying. Want to buy your way in with bids? Sorry, that door doesn't exist.

Rule Three: The Metric Changed — From Click-Through Rate to Citation Rate

Back in the SEO days, everyone watched one number: click-through rate — how many users actually clicked into your page.

Now the scale that weighs your worth has changed: citation rate — how many times, across all its answers, AI names and cites you.

How much weight does that word carry? In the past, when you spent on advertising, you were buying "being seen." From now on, what you buy is "being remembered."

Let me ground that in two real stories.

In April 2025, the developer-tools company Vercel published a striking stat: 10% of its new signups were coming directly from ChatGPT. Sound absurd? Six months earlier, that figure was under 1%. Tenfold in half a year — without a single dollar of targeted ad spend.

Let that sink in. That is the power of being cited — users didn't go looking for you in a search box; the model remembered you on their behalf, mid-conversation.

The second story is about Canada Goose.

Starting in 2025, the down-jacket brand did something unusual: it hired people to pack tens of thousands of simulated questions and feed them into all kinds of large language models, to see whether — when the models answered related questions — the words "Canada Goose" would come out of their own accord.

It didn't want to know which channels users discovered it through. It wanted to know: at the critical moment, would a model untouched by any advertising spontaneously spit out the name?

In the branding world, this is called "unprompted recall."

In the past, brands bought their way into your memory with ads. That path still exists, but there's a new prerequisite: you have to get AI to remember you first.

New tools are already popping up to meet this need: Profound, Goodie, Daydream — built to measure "how your brand looks inside AI answers." The old SEO giants aren't sitting still either: Ahrefs, Semrush, one after another, have launched their own AI-visibility tools.

How you appear inside AI is becoming a new kind of moat.

4. The Flip Side: Why SEO Never Produced a Dominant Player in Twenty Years

Let me tell you a counter-example.

SEO is a massive industry. In more than twenty years, how many giants have you heard of?

Semrush, Ahrefs, Moz, Similarweb.

They're all solidly built, right? But look at the ground each one claims: the backlink tool sticks to backlinks, the traffic tracker sticks to traffic, the keyword researcher and the technical auditor each play their own lane.

Why has no one taken it all? One phrase: too fragmented.

First, the industry is a wholesale marketplace of services. The work is spread across thousands of companies — outsourcing layered on outsourcing, freelancers picking up one gig after another.

Second, SEO's "truth" has always been inferred, never verified. The ranking algorithm sits inside the search engine, while the real data — which links users actually clicked — is locked away with carriers, SDKs, browser extensions, and all manner of data brokers.

Put plainly, the layer furthest from the commercial truth is SEO itself. It's nearly impossible for a monopoly platform to grow out of an industry built on inference.

5. So What About GEO — Why Does It Have a Chance?

Because this time, the data comes to you.

Every answer a model gives and every citation it makes stays in the API logs. What users asked, what the model answered, what users ended up believing — all of it can be traced.

That is the most fundamental difference from the SEO era.

So in my view, the GEO platforms that go big won't stop at dashboards. They'll do three things:

First, build their own models. Using billions of real questions, they train a model to understand what's worth citing. That capability is built out of money and data — something no one can copy.

Second, connect the data. Stitch first-party data (what you already own) and third-party data (clickstreams, search behavior) into one complete map.

Third, turn observation into steering. After an assessment, it doesn't just hand you a report and call it done — it hands you an action pack: how to adjust your content next week, which brand personas to reinforce, which direction deserves another article. So that tomorrow, the model remembers you again.

Once a company gets that loop running smoothly — seeing you and changing you — it stops being just a tool. It becomes the operating system between brands and AI.

Let me make the claim bigger: it no longer manages "search" — it manages the entire relationship between companies and the AI layer. Your memory map, your brand context, your room for future rewriting — all of it lives inside that relationship. Whoever takes charge of the relationship layer takes charge of the budgets and priorities that follow.

The a16z crowd puts it even more bluntly: this was never a search business — it's a new form of marketing. Search is just the first tentacle it has grown.

6. The Last Layer: Timing

Finally, let's talk timing.

Ad money is never slow. It only flows to places that can convert precisely and be measured.

In the 2000s, money poured into Google's AdWords — because search engines could, for the first time, track conversions.

In the 2010s, money shifted to Facebook's targeted ads, because that's where the user profiles were.

And now? A new kind of measurability is taking shape inside the large language models. When the question itself becomes the strongest targeting signal — AI knows you better than you do — ad budgets are destined to move here.

Do you want to just watch that day arrive, or claim your spot before it does?

Epilogue

That night, my friend's message stayed on my phone.

He never actually asked me whether he should spend the money.

What happens next comes down to a single sentence we all need to understand:

"Will AI think of you?" has become the most valuable channel of this era.

Getting AI to remember you is the prerequisite. Being remembered is the answer.

That's what I gave my friend that night — and now I'm giving it to you too.

May your brand claim a good spot in the minds of these models.