Is Your Name in AI's Answers?

A while back, a friend of mine who runs a home renovation business invited me to dinner.

Somewhere into the meal, the sighs started.

He said: it makes no sense. His website's keywords were still on page one, his rankings hadn't slipped an inch — but the inbound calls were visibly falling off.

I asked him: the last time you needed something installed at home, did you open a search engine and dig through links, or did you just ask an AI?

He said: obviously I ask an AI. It asks follow-up questions, compares prices, and talks like an actual person. Who has the patience to flip through ten web pages anymore?

I said: well, that explains it.

Your ranking is still there. It's search itself that has moved.

People used to search because they wanted to pick one link out of a pile. Now people ask because they want an answer, straight up.

And those answers are written by AI.

When AI writes an answer, it picks a few sources it can trust from across the web, lifts passages from them, and cites them.

So here's the new question: is your name in AI's answers?

You used to compete for rankings. Now you compete for citations.

The marketing world overseas has given this thing a name: GEO, Generative Engine Optimization.

The name is fancy; the substance is plain. Let me walk you through it.

What Is GEO?

First, look at how search has changed.

The old search engine was like a farmers' market. You paid money and put in effort to grab the stall with the most foot traffic, stacked your display high, hung your banners, and waited for shoppers to wander past.

Today's ChatGPT, Perplexity, and Gemini — plus the AI-generated answer Google now serves at the top of its results — are more like a personal advisor your users have hired.

Users bring a question, the advisor gives the answer straight, and casually drops a few names he trusts.

Did you notice? In this game, the concept of a "stall" no longer exists.

Users don't browse anymore. They only listen to what the advisor says.

So what is GEO?

GEO is everything you do to make that advisor mention you, cite you, and treat you as a reliable source when he writes his answers.

Put simply, who AI cites is decided not by who shouts loudest, but by whose words it dares to copy.

Why Does This Worry You More Than a Ranking Drop?

Picture a scenario.

A user wants to buy a dishwasher. He asks an AI: "Family of three, small kitchen, budget of 5,000 yuan — which model should I get?"

The AI produces an answer: the key points to consider when choosing, two or three recommended brands, all well-reasoned.

The user reads it, is satisfied, and places an order.

So where's the problem?

From start to finish, the user never opened a single web page.

Insiders have a term for this: zero-click search. He read the answer, got the job done, and never clicked a link once.

More importantly, trust now forms earlier than ever.

In the past, users comparison-shopped first and formed their judgment after comparing. Now the judgment is made for them by AI. However AI's answer describes you is exactly what you are in his mind.

If your ranking drops to page two, users can at least still scroll down a bit.

If you're not in AI's answer, the user doesn't even know you exist.

How Do GEO and SEO Relate?

Many people hear GEO and immediately assume SEO is doomed.

Let me stop you right there.

SEO isn't dead. SEO is the foundation — it gets search engines to find your pages, index your pages, and push your pages up the rankings.

GEO is the floor you build on top of that foundation — it gets AI to understand your content, trust your content, and carry your content into its answers.

One gets you found. The other gets you cited.

SEO decides whether you can be found. GEO decides whether you can be cited.

Both of these run on the same set of content assets. A clearly structured, factually solid article is one that search engines are willing to rank and AI is willing to copy.

So don't set the new against the old. If the foundation is weak, the floors built on it are wasted no matter how pretty; if you only pour the foundation and never build the floors, everyone else has moved into their new homes while you're still mixing concrete.

What Makes AI Cite You?

So what does AI actually look at when picking sources?

Here's an analogy. When AI writes an answer, it behaves very much like a rigorous straight-A student writing a literature review.

He's holding a mountain of material from across the web, but he writes with great caution. There are three kinds of material he doesn't dare touch.

The first kind: incomprehensible. Stuffed with keywords from beginning to end, circling around without ever reaching the point — he can't excerpt it.

The second kind: unsourced. "Noticeably improved results" — improved by how much? Who measured it? Without evidence to show, he doesn't dare put it in his paper.

The third kind: self-contradictory. Your website says founded in 2015, the encyclopedia says 2018, the media says something else again. He fails the background check on you and simply routes around you.

Flip it around: what does he like?

What he can understand, what's said plainly, and what stays consistent across the whole web.

And there's one more layer: even the act of asking has changed.

Users used to type four or five words into search and be done. Now, when people talk to an AI, an average sentence runs 23 words.

What does that mean? Users will lay out the background, the constraints, the budget — all in one breath. What AI has to catch is a complete question, not a keyword.

So your content has to switch from "stacking keywords" to "properly answering a complete question."

How Do You Actually Do It? Five Things

That's the theory. Now the practice. The overseas crowd has a whole pile of frameworks; I've boiled them down to five things you can explain in plain words.

The first: write your articles like answers.

Make the title the exact sentence users actually ask — for example, "Does a small apartment work with central air conditioning?" Then, in the very first sentence, give the conclusion.

No setup, no cliffhangers. When AI extracts content, it grabs passages that are clearly defined and conclusion-first. The more your writing reads like a ready-to-lift answer, the higher your odds of being cited.

The second: show your evidence.

Behind every claim, follow up with numbers, cases, studies, and named experts.

State the number exactly as it is, write down whoever ran the study, and spell out what qualifies the author to speak on this industry.

AI quotes a claim that can show its source without a second thought — and gives anything vague a wide berth.

The third: cut your content into blocks a machine can lift easily.

Break sections up with subheadings, keep paragraphs short, use lists where lists fit, use tables where tables fit. On top of that, you can add a layer of Schema structured data — a kind of label written for machines, effectively tagging in advance "this is a product, this is the company, this is an FAQ" so AI reads it without guessing.

And don't scoff at the basics: pages should be fast, open fine on a phone, run on HTTPS — and don't block crawlers at the door. AI fetches content through the same old pipeline search engines use; if it can't get through the door, citations never even come up.

Same for images and video. For every image, honestly write out what is in the picture. AI "sees" an image through the one line of caption you write for it.

The fourth: one story across the whole web.

Your website, encyclopedias, media, directory pages, review sections — whenever your company is mentioned, the name, address, phone number, and main business must fit together seamlessly.

Because before AI decides to trust you, it runs a background check. Only when the story matches everywhere does it confirm you are you. If the stories clash, it splits you into several unclear entities — and gives up on all of them together.

If you run a local business, this one is make-or-break. Fill out your Google Business Profile honestly and completely — business hours, service categories, photos, don't leave out a single one; reply to customer reviews with care. Plenty of small shops lose on exactly these unglamorous details.

The fifth: go out and show your face.

Don't let good content just sit on your own website. Industry media, vertical communities, professional platforms — go wherever you can.

The more third parties mention you, the more AI feels the whole industry recognizes you, and the more at ease it is citing you.

Oh, and one thing that's easy to miss: content expires. Evergreen articles should be revisited and updated every six months to a year; pages with prices and features should be changed the moment things change. A business card printed with an outdated phone number is more embarrassing than having no card at all.

How Long Does This Take, and How Much Does It Cost?

Time first, so you don't go checking results tomorrow morning.

Generally, in two to four weeks you'll see the first batch of AI citations — a sign your content has entered the pool of sources AI draws from.

Three to six months in, AI starts reusing you across different questions.

Six to twelve months before things are truly stable.

Why so slow?

Because the way AI recognizes authority is "seeing you repeatedly" plus "multiple sources agreeing." One appearance is a coincidence; appearing every time is real strength. There is no crash course for this thing.

Now the money.

Rough market rates overseas: a freelancer taking a small project, $150 to $1,000; a serious small project, $2,000 to $5,000 a month; an enterprise-scale operation, starting at $10,000.

Expensive or not, take it and compare it against what you'd pay a content team or spend on feed ads. My view: this is money to hold the gate, not a nice-to-have.

So how do you know whether it works?

Watch three numbers: on any given topic, how often you get named in AI answers; the wording AI uses when describing you — does it add points or subtract them; and for the same question, whether you get mentioned more or less than competitors.

A wave of monitoring tools built specifically to track AI citations has already hit the market. All of these numbers are now directly readable.

Finally, My Own Take

AI is getting pickier about its sources — and increasingly, what it's picky about is the "who" behind them.

It's no longer satisfied with knowing what you sell; it wants to know who you are: what year you were founded, what you're good at, who in the industry speaks for you, what the images and videos actually show. The clearer and more consistent your identity is across the web, the more likely you are to get named.

Sit with what this really means.

For the past decade, competing on content meant competing on whether people thought you were good.

From now on, you also compete on whether machines think you are credible.

Users no longer comparison-shop, because AI has already finished the comparing for them.

At the end of that dinner, my advice to my friend was simple: go home, open an AI, take the ten questions your customers ask most, and ask them one by one.

Then look at the answers and see whether you're in them.

If not, you'll know what to do.

Here's to staying half a step ahead of your competitors — and settling into AI's answers before they do.