When ChatGPT Talks About Your Brand, Is It Praising You — or Not Mentioning You at All?

A while ago, a friend who runs a B2B software company came to see me with an aggrieved look on his face.

He said, "Run, our SEO isn't bad. Our core keywords sit on Google's front page year-round, and website traffic is steady. But lately the sales team keeps hearing that more and more customers ask ChatGPT before buying: 'What are the best project management tools?'"

ChatGPT rattled off a list.

He wasn't on it.

Not a single mention.

Frustrating, right? The rankings are still there, the traffic is still there — but the moment of decision has moved. Moved to a place where you can't even find an optimization lever: the AI's mouth.

That got me thinking. So today, let's unpack it properly.

The Referee of Search Has Changed

What was search before? A library.

You asked the librarian a question, and he handed you a stack of books: "Go dig through them yourself. The answer is probably in the third one, or the seventh." So everything companies fought over was the spot of "being handed to the user." Number one ate meat, number ten sipped soup, and page two basically didn't exist.

That was the rule of the SEO era: fight for position.

And now? Search today is a consultant.

You ask ChatGPT, Perplexity, or Gemini a question, and it doesn't hand you books. It reads the whole stack itself, synthesizes an answer, and reads it out loud to you. The user listens, feels they understand, and can't even be bothered to click a link.

The referee has shifted from "the one who hands you books" to "the one who reads out the answer." This is the biggest changing of the guard in search in over a decade.

So here's the question: why should you be in the answer it reads out?

What Is GEO?

The industry has a name for this new playbook: GEO, Generative Engine Optimization.

Sounds intimidating. Strip it down, and it's one thing:

Getting AI, when it answers a question, to be willing to mention you — and to mention you correctly.

Note that this is a completely different skill from SEO.

SEO fights for a ranked list. First place, second place, third — the slots are limited, and it's a zero-sum war. GEO fights over "how the AI talks about you." AI answers are generated on the spot; the same person asking today and tomorrow may get different answers. There's no fixed leaderboard for you to climb.

So staring at rank-tracking tools every day? That game is over. What you need to watch now is: does the AI's answer include you, where does it place you, and is what it says about you praise, a knock, or lukewarm nothing.

So how do you get AI to "know" you? First understand how it thinks about you.

How AI "Gets to Know" a Brand

AI relies on several streams of signals.

The first stream: the content it has read. During training, the model consumed a massive amount of web pages — your website, your articles, other people's posts about you, all of it. Shallow content, watery content, the same recycled phrases over and over — it swallows those without chewing and they leave no impression.

The second stream: its understanding of "entities." AI doesn't look at keyword stuffing. It recognizes: who this company is, what it does, who it keeps company with, and whether people trust it. It's like matchmaking — what matters isn't how many words are on your résumé, but whether the matchmaker sees you, the actual person, as reliable.

The third stream, and the one most easily ignored: how people online talk about you.

Discussions on Reddit, forum posts, word of mouth on social media. The buzz in these places seeps into the model's impression of your brand. What your own website says is self-promotion; what everyone across the web says — that's reputation.

Once you get this, you understand the weight of one sentence:

Community discussion is the new backlink.

Back in the SEO days, every authoritative site that linked to you added a point to your weight in Google's eyes. Today, with GEO, wherever a crowd of people genuinely discusses you and recommends you, the AI mentally notes down a point for you. The game has changed; human nature hasn't: what everyone agrees is good, the AI believes too.

Two Kinds of AI, Two Playbooks

There's a nuance here: AI engines come in two camps, and they call for different tactics.

One camp relies on "credentials." Think ChatGPT's base model, Claude, Llama — their knowledge has a cutoff date; they're graded on what they've absorbed over time. With them you can't rush. You do the long-term work: steadily produce content that stands the test of time, wait for the next model update, and when it looks at you favorably, it will bring you along.

The other camp relies on "live lookups." Think Perplexity, Gemini, and ChatGPT with web access enabled — they fetch fresh content from the web on the spot when answering. With them, results come fast: clean up your content structure, standardize your data markup, update whatever needs updating, and it can see you today.

One is planting trees; the other is watering flowers. You need to do both.

Three True Stories

Principles alone won't convince you. Let me tell you three cases that stuck with me.

The first. A B2B company selling project management software found that ChatGPT's knowledge base barely contained them — ask it anything, and they went unmentioned. Instead of throwing money at ads, they went back and went all-in on content: explained the product clearly, covered the industry thoroughly, standardized their structured data, and then tended the discussions in communities. After the next model update cycle, when ChatGPT listed "the best project management software," they made the recommendation list. Branded search volume rose 25%.

The second. An e-commerce shop selling eco-friendly lifestyle products noticed their target customers loved using Perplexity — sustainability-minded people gravitate to search with cited sources. So they optimized specifically for Perplexity's live-lookup mechanism: fresh content, clean structure, clear labeling. In under two months, their name showed up consistently in recommendations like "best eco-friendly stores to buy from," and monthly revenue rose 18%.

The third. A consulting firm with a bigger appetite wanted to show up in both Google's Gemini and web-connected ChatGPT at the same time. So they trained both skills at once: accumulating authoritative content long-term to feed the credential camp, updating daily to feed the live-lookup camp.

Different emphases, same foundation: content solid enough, structure clear enough, and real people actually talking about you in communities.

How to Know Whether You Exist in the AI's Eyes

One last practical question: how do you measure this?

It is genuinely more slippery than SEO. SEO rankings at least hold steady on a weekly basis; AI answers differ from today to tomorrow, and rephrase the same question and the answer changes again.

So don't count on a fixed ranking. Watch a few things instead: how often you get mentioned for common questions, whether the context of those mentions is favorable or not, and whether the mentions translate into rising branded searches and direct visits.

Dedicated tools have appeared for exactly this — Goodie AI, for example. It helps you monitor what your brand looks like in the answers of various generative AIs, turning "you in the AI's eyes" into a report you can track and optimize.

If you can't see it, you can't fix it. Don't skip this step.

Back to the Beginning

Back to that friend ChatGPT never mentioned.

He eventually understood: the customers didn't change, the demand didn't change — what changed is how customers "ask for directions." They used to ask Google; now they ask an AI. Your shop is still on the old street, but the foot traffic has rerouted through a new intersection.

Keep the storefront on the old street — and put up a sign at the new intersection too.

SEO gets you found; GEO gets you remembered. Over the next decade, many user questions will be answered straight inside a chat window. Whether that answer contains your ideas and your name may well decide whether you gain one more customer — or were never known at all.

The best time to plant a tree was ten years ago. The second-best time is now.