Someone Asks ChatGPT a Question — Is Your Company in the Answer?

A while back, a friend of mine in enterprise services was chatting with me.

He said, "I spent three years getting my website onto Google's front page. Traffic kept climbing, and I was happy about it. Then last month, a big client tells me they don't even check Google anymore for purchasing decisions. They ask ChatGPT, they ask Perplexity, and they get back a paragraph with a recommendation. Those three years I spent on front-page rankings — they never even clicked through."

My heart sank a little.

Not because he lost that one deal, but because I suddenly realized something: the way users look for information has quietly shifted to a different track. It used to be flipping through links on a search results page. Now it's getting back a ready-made "answer." You're still competing for who ranks first, while users have stopped looking at rankings altogether.

That's what I want to talk about today.

The Underlying Logic of Search Has Changed

Let's start with the most basic point.

What do I mean by "changed"?

Back then, you'd search "best project management tool," and Google would give you a page of blue links. You'd click through them one by one, spend half an hour, and make a decision.

Now you ask ChatGPT the same question, and it doesn't hand you a pile of links. It just tells you the answer. Which tools it recommends, the pros and cons of each, which kind of team they suit — all in one paragraph, with cited sources hanging at the bottom.

See, a fundamental shift has happened in between: users went from "finding the information themselves" to "having AI find it for them, read it for them, and summarize it for them."

What does that mean?

It means if you're not inside the AI's answer, you effectively don't exist. No matter how good your SEO is, no matter where you rank on Google.

That's where GEO begins.

What Exactly Is GEO?

GEO stands for Generative Engine Optimization.

In plain terms, it's getting your content cited by AI search engines.

Traditional SEO was about "ranking on Google's first page." GEO is about "showing up inside the AI's answer."

You might say — isn't that just SEO with a new name?

Not really. The difference is big. I'll get to that.

The industry hasn't settled on one term for this yet. Some call it AI SEO, some call it AEO (Answer Engine Optimization), others call it LLMO. Different names, same goal: getting AI to mention you.

How AI Search Actually Works

To understand how to do GEO, you first have to understand how AI search works.

It's nothing like what you might imagine — "tossing your question into Google."

AI search runs in four steps.

Step one: break the question apart.

Say you ask, "Which VPN is best for watching Netflix in Europe?" The AI doesn't search that whole sentence. It splits the question into smaller ones — "best VPN 2026," "VPN for Netflix," "European VPN servers" — and searches them separately.

This is called a fan-out query.

Step two: pull information.

The AI takes those smaller questions, searches the web, and also rummages through its own knowledge base. The technique it uses is called RAG (Retrieval-Augmented Generation) — basically, it pulls relevant passages out of web pages and feeds them to itself as reference material.

Step three: synthesize the answer.

The AI weaves together information from multiple sources, reorganizes the language, and hands you a coherent answer. It doesn't copy and paste — it rewrites, merges, and integrates.

Step four: attach citations.

At the end of the answer, it labels the information sources. The sites that get cited have a chance to earn traffic.

You see it now, right? If you want to be cited by the AI, you first have to be scooped up by it inside those "broken-down smaller questions." That's a completely different game from traditional SEO, where you optimize a single long-tail keyword.

One More Thing You Have to Know

LLMs (large language models) have a property: they're non-deterministic.

What does that mean?

Ask ChatGPT the same question five times, and it'll give you five different answers.

So in AI search, there's no such thing as "ranking first." There's no position #1.

Then what should you chase?

Frequency.

How many different questions, how many different conversations does your brand show up in? The more it appears, the higher the chance users see it.

You can think of it as a "mention rate." The higher the mention rate, the more exposure you get inside AI.

What's the Real Relationship Between GEO and SEO?

A lot of people hear "GEO" and their first reaction is: so SEO is dead, right?

Quite the opposite.

GEO and SEO sit on the same foundation. The SEO work you've already done — E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness), technical optimization, content quality, backlinks — all of it still counts in GEO. AI search engines also crawl the web to grab content; if your site is well-built, your content is high-quality, and your backlinks are strong, the AI is just as likely to find you.

But a few things really are different.

Traditional search gives you a list of links, and the user has to click in and dig. AI search gives you a synthesized answer, and the user just takes it.

Traditional search keywords average about 4 words — short, fragmented. AI search questions average about 23 words — users describe their full need the way they would in a conversation.

Traditional search looks at rankings, click-through rates, traffic. AI search looks at citation count, brand mentions, share of voice.

The core question of traditional search is "Am I on the front page?" The core question of AI search is "Am I in the answer?"

There's one more thing worth paying attention to: ranking well on Google does not mean AI will definitely cite you. GEO firm Brandlight did a study — the overlap between Google's front-page links and the sources AI cites has already dropped from 70% to under 20%. And that gap is still widening.

The logic AI uses to pick whom to cite and whom to skip is no longer the same as Google's.

How Big Is This in 2026?

You might think AI search sounds cool, but are that many people really using it?

Here are a few numbers to feel the scale.

ChatGPT has over 800 million weekly active users. Google's AI Overviews covers billions of searches every month. Perplexity handles millions of queries a day. Apple is already integrating AI search (including Perplexity and Claude) into Safari.

These aren't "future trends" — they're happening right now.

And user behavior has changed too.

On Google, you search and leave, dwelling for a few seconds on average. In AI search, an average session lasts 6 minutes. Users follow up, refine, describe their full situation.

Even more interesting is conversion. Traffic from AI search is smaller in volume than Google, but higher in quality. The user has already gotten a recommendation from the AI, so when they click through, they're basically coming straight for you. Vercel once reported a figure: 10% of their new signups came from ChatGPT referrals.

Users trust the answers AI gives. That's why GEO is worth doing.

How to Actually Do It: Six Things

Alright, that's enough theory — now for the practice.

I've broken it into six things, each one a tactic that genuinely works in 2026.

Thing One: Make Sure AI Can Actually Read Your Content

This sounds obvious, but it's the most common pitfall.

A lot of websites have quietly shut AI crawlers out without realizing it. Cloudflare changed its default settings a while back to block AI crawlers outright. If you're using Cloudflare, your content might already be invisible to AI — and you wouldn't even know.

So what do you do?

Go check your robots.txt file. Check your server logs for the ChatGPT-User user agent. If you're on Cloudflare, go look at the AI Crawl Metrics page in the dashboard.

There's another easily overlooked point: if your important content loads via JavaScript, AI can't see it.

AI crawlers aren't humans. They don't click, they don't scroll, they don't play with the interactive slider on your pricing page. They only look at the HTML the server returns. Anything you hide inside tabs, collapsible menus, or behind a login wall is invisible to AI.

If you want to be cited, put that content into the HTML.

Thing Two: Structure Your Content So AI Can Grab It

The AI needs to pull specific information out of your page. The easier you make it to grab, the more willingly it cites you.

How?

Keep your heading hierarchy clear. H1, H2, H3, nested one inside the next, each section covering only one topic. The AI uses headings to judge what each paragraph is about.

Use lists and numbered items more often. Someone analyzed ten thousand real queries and found that pages with structured lists, citations, and statistics had 30% to 40% higher visibility inside AI answers.

Give the answer first, then the background. Don't bury the conclusion under three paragraphs of preamble. The AI is looking for answers it can extract directly.

Keep paragraphs short. Two or three sentences at most. A dense wall of text is hard for the AI to parse, and its citation probability drops sharply.

Thing Three: Target the Sub-Questions AI Actually Searches

This is where GEO and SEO diverge the most.

Remember the fan-out queries from earlier? A user asks one long question, and the AI splits it into several short ones to search.

For example, a user asks ChatGPT: "Which email marketing platform should a small e-commerce store with under ten thousand subscribers use?" The AI might split it into "best email marketing platforms 2026," "e-commerce features for email marketing," "small-team pricing for email marketing."

You need to make sure you have content that answers each of those short questions. Use the words AI might search, and cover every angle of the breakdown.

Think about it — if you had to break this question down yourself, what would you search? Those fragmented short queries are exactly what you need to optimize for.

Thing Four: Give AI the Authority Signals It Trusts

Before the AI decides whom to cite, it evaluates whether the information source is trustworthy.

You have to give it signals.

Add expert quotes. Real names, real titles, real companies — the AI reads these as strong authority signals. Label the sources of your data. "According to Semrush's clickstream data" carries far more weight than an unsourced assertion.

Show first-hand experience. Your own case studies, your own observations, your own data — this is the E in E-E-A-T (Experience), and AI values it more all the time.

Write author info clearly. Who wrote it, and what makes them qualified. This helps not just SEO but GEO too.

Thing Five: Stay Fresh

AI has a strong "recency bias."

From the data I've seen, once content is over 3 months old, the frequency with which AI cites it falls off a cliff.

Refresh your important content at least once a quarter. Update the data, swap in new cases, add new developments. This isn't just a good SEO habit — it directly determines whether AI searches can find you.

Thing Six: Build a Presence Outside Your Own Website

This goes beyond the scope of traditional SEO, but it's crucial.

AI learns about your brand not just from your official site — it looks at the whole internet.

Unlinked brand mentions count too. When someone offhandedly mentions you on Reddit, in a forum, in a blog — even without a link attached — the AI still counts it.

An even sharper move: find the pages AI is already citing, and figure out how to get yourself into them.

How do you find them? Look at the topics you care about and see which sources the AI's answers cite. Then figure out how to appear in those sources — comment on that Reddit thread, email the author of that blog post, take part in that forum's discussion. Someone used this method to go from a brand AI couldn't see at all to earning their first AI mention in under an hour.

Reddit, YouTube, all kinds of forums show up frequently in AI answers. Engage seriously — it works. Spam and flood — it doesn't.

If your brand qualifies, put real work into a Wikipedia entry. Wikipedia makes up a big share of AI training data, and an accurate entry shapes how AI describes you.

Do Tactics Differ by Platform?

The core principles carry over, but each AI search platform has its own temperament.

ChatGPT holds about 70% of the AI search share, mixing real-time search with training data, and favors content that's comprehensive, with clear sources and professional signals. Its referral impact is becoming increasingly measurable.

Google AI Overviews blends traditional search ranking signals with AI synthesis. Content that performs well in organic search usually does well in AI Overviews too. Structured data (Schema) may influence whether you get picked.

Perplexity leans heavily on citations, searches in real time, prefers fresh content, and has the highest source transparency. SaaS product conversion rates perform well on Perplexity.

Google Gemini is growing the fastest, deeply integrated with Google's search infrastructure. Do Google SEO well, and Gemini visibility usually follows.

Claude excels at synthesizing information rather than citing directly, and prefers content that's cleanly structured and logically strong. Apple has announced it's integrating Claude into Safari — the impact of that move is worth watching.

How to Measure Results

The metrics from traditional SEO aren't enough for GEO.

Most AI search is zero-click — users get their answer and leave without ever coming to your site. Staring at Google Analytics for traffic means you're not seeing the full picture.

What should you look at?

Share of Voice. This is the most important GEO metric. Across a large batch of relevant AI questions, how often does your brand appear? I said earlier — AI search has no "rank first," only "shows up a lot or a little." This metric measures exactly that. The higher the share, the more exposure.

Competitive ranking. Just looking at yourself isn't enough — you have to compare with competitors. Where do you rank inside the AI's answers?

Citation tracking. Which of your pages got cited by AI? What kind of questions were they answering?

Brand description accuracy. How does AI describe your brand? Is the information correct? Does it say good things?

AI referral traffic. Check server logs for ChatGPT-User, or Cloudflare's AI Crawl Metrics.

No need to rush out and buy tools. You can measure it yourself:

Pick 10 to 20 questions related to your business — especially the ones users ask when making purchasing decisions. Take them to ChatGPT, Perplexity, Gemini. Record whether your brand shows up, how it's described, and who got cited. Run it once a month and watch the trend.

A Few Common Pitfalls

Let me close with a few mistakes that are easy to make.

On content: keyword stuffing doesn't work — AI penalizes over-optimization just like Google does. Content that's too thin or too watery, AI won't want to cite. Letting information go stale — past 3 months, citations drop. Having data with no source makes credibility take a hit. One giant block of text with no line breaks is hard for AI to parse.

On technical: blocking AI crawlers without realizing it (check robots.txt and Cloudflare). Important content rendered by JavaScript that AI can't see. Content hidden behind interactive elements (tabs, collapsibles, sliders) that AI won't click. A site that's too slow, dragging down AI crawler efficiency.

On strategy: treating GEO and SEO as two separate things. They're actually one — AI uses real-time search, so strong SEO means strong GEO. Using AI to mass-produce junk content and flood your site — that's poison for both SEO and GEO. Only staring at your own site and ignoring whether anyone mentions you on third-party platforms. Not tracking results — most AI search is zero-click, and if you don't specifically measure it, you'll see nothing at all.

Back to That Friend From the Beginning

After we finished talking, that friend was silent for a moment.

He said, "So did I waste those three years?"

I said, "No. Your SEO foundation is valuable. AI also crawls the web for content — your site is well-built, your content is high-quality, your backlinks are strong, and it'll find you just the same."

But relying on the old playbook alone really isn't enough anymore.

The foundation you laid before wasn't for nothing. But on those six things, you've got some catching up to do. Whether AI can read your content, whether it can pull information out of it, whether it can find authority signals, whether you're staying fresh online — all of that needs a once-over. And one more thing that's easy to miss: how much you exist outside your own official site — AI cares about that a lot.

You used to chase "ranking on the front page." Now you have to chase "showing up in the answer."

Those are two different things. But doing the first one can help you do the second.

SEO isn't dead. It changed its clothes and kept moving forward.

Keep up, and ahead of you is a blue ocean. Fail to keep up, and your competitors will catch up for you.

---

Back to that very first question: someone asks ChatGPT a question — is your company in the answer?

If not yet, today's a good day to start thinking about it.