Mention beats link: becoming an entity AI recognizes

Let me start by knocking down a belief that was worth fifteen years and now works against you. In the old Google game, the link was currency. The more sites pointed to yours, the more Google understood you were important, and the higher you climbed. The entire SEO industry revolved around this: chasing backlinks, trading links, buying links. Well. For AI, that game changed shape. What it values most today isn't the clickable link, it's the mention: your name, your brand, cited around the web, in sources it trusts, even without any link pointing to you at all. This lesson is about understanding why that happened, and what you do about it.

01Why the mention started to matter more than the link

Think with me about what AI is doing when it recommends you. It's not sending you traffic, like Google used to. It's answering the person right there, on the spot, and to do that it needs just one thing: confidence that the name it's about to say is real, taken seriously, and won't make it look foolish. And how does it gain that confidence? By seeing your name show up, repeated, in places where real people genuinely discuss your market.

Notice the difference in nature. The link is a technical vote: a site decided to point to another. The mention is a human vote: someone cited you because you came to mind when the topic came up. AI, which was trained reading people's conversations, learned to weight the second one more heavily. Search Engine Land covered a startling data point: in Google AI Overviews' answers, on prompts like best X, AI recommends a competitor, not the reference brand, about 69% of the times a listicle shows up (Lily Ray's analysis). In other words, most brands simply aren't in the conversation, and don't even know it.

The practical reading is this: you can have a flawless site, full of backlinks, and still not be cited by AI, because nobody outside talks about you. And the opposite is also true: a brand that shows up hammered in the right conversations gets recommended, even with a modest site. The link didn't die, but it lost its seat at the head of the table. What runs the show now is the mention. Fair enough?

02The consensus signal: AI needs to see you hammered, not just once

Now the mechanism I most want you to take away from this, because it explains almost everything. A single, isolated mention doesn't convince AI. It's skeptical by design: one single place speaking well of you could be you praising yourself, could be an ad, could be luck. What convinces AI is repetition across independent sources. When it sees your name in a Reddit thread, in a YouTube comparison video, in a G2 review, and in a third party's LinkedIn post, it cross-references everything and concludes: "this name the market genuinely cites". That's the consensus signal.

And it's not just a guess on my part. When Leapd analyzed where ChatGPT, Google AI Overviews, and Perplexity pull information from in 2026, Reddit showed up as one of the most consulted sources. On Perplexity specifically, it reaches about 46.5% of citations (Frase's research). Why? Because that's where real people, with no marketing script, say what they think. AI learned to weight that kind of organic conversation precisely because it's hard to fabricate.

This turns into a concrete guide for where to invest your effort. Instead of spending it all on your own blog (which AI reads as "you talking about you"), you spread presence outward: become a topic in a forum, show up in a review comparing your market, get cited in a video, get mentioned by third parties on LinkedIn. Frase, in a practical playbook on getting cited by AI, sums up the golden rule: AI favors earned media, what third parties say about you, above what you say about yourself. You don't control others' mentions. But you control whether you deserve them and whether you make them easier to happen. Fair enough?

03Becoming an entity AI recognizes (not loose words)

There's a third piece that ties the other two together, and it's the most technical-sounding one, but simple to understand. AI, and Google behind it, no longer thinks in terms of "keywords". It thinks in entities. An entity is a name the system understands as a real thing, classified, with facts tied to it: "this brand IS a performance agency, founded in such year, specialized in e-commerce, based in Brazil". When AI has that clear sheet about you, it cites you with confidence. When it doesn't, it hesitates, confuses you with someone else, or simply invents a wrong fact to fill the gap (and that hurts you).

In the jargon it's called Entity SEO, and behind it there's a structure called the Knowledge Graph, which is the "map of real things and how they connect" that Google maintains. Google itself, in its official guide to optimizing for generative AI features in Search, is direct: it recommends content with consistent identity, verifiable facts, and clear structure, precisely so the system can understand who you are. Becoming an entity means doing the homework of being easy to understand:

• Consistency. The same name, the same description, the same facts, everywhere you show up. If every corner describes you differently, AI can't build a single sheet.
• Verifiable facts. What you do, for whom, since when, where. Things third parties can confirm, not adjectives about yourself.
• A brand facts page. A clean, direct place that answers "who is this brand, what does it do, for whom", which AI can read as a source of truth.

This connects directly to what you saw back at the beginning of the course about context (lesson 2.1): AI is only as good as the context it has about you. If the context about your brand, scattered around the internet, is confusing and contradictory, the answer about you will be confusing too. Becoming an entity means, deep down, giving AI a context clean enough for it to talk about you without getting it wrong.

04The dangerous temptation: trying to force the mention

Now I need to give you a warning, because wherever there's opportunity, a shortcut appears, and this particular shortcut blows up in your hand. Since AI reads content to decide what to recommend, someone's going to think: "so I'll hide an instruction in a text AI reads, telling it to recommend me". This has a technical name, prompt injection, and we've already covered it in the course's security section (lesson G.2). What you need to know here is that, in marketing, this shortcut isn't just unethical, it technically makes your situation worse.

There's a 2026 study (the so-called "Injection Paradox") that measured exactly this. When you plant a hidden instruction in a document a safety-trained model (like Claude) is going to read, its safety classifier notices the manipulation attempt and does something cruel: it downranks your entire brand, silently, without warning. You vanish from the ranking and don't even know why. Having the poisoned document ends up worse than having no document at all. And there's the human side: in the United States, the FTC already bans fake reviews and testimonials, with heavy fines. Fabricating consensus is the fastest path to being punished by AI itself and by the law.

The good news is that the honest path and the path that works are the same path. Building real presence (being good enough that third parties talk about you, making that conversation easier, keeping your facts clean and consistent) is exactly what AI rewards. There's no hack that beats the real thing here. The consensus signal was designed, on purpose, to be hard to fake. Use that to your advantage: stop looking for the shortcut and build what AI can't ignore.

05Where this takes you in the rest of the module

This lesson was the why. You understood that the currency shifted from link to mention, that AI needs to see you hammered to trust you, and that the goal is becoming a recognized entity, with no fraudulent shortcuts. The rest of this GEO/AEO module is the how: measuring whether you're being cited (your share in AI answers), writing content AI can extract and cite verbatim, and tracking how your brand shows up in answers over time. Each lesson picks up one of these levers and gives you the hands-on work.

For now, what I want stuck on your wall is a single sentence: in the new game, you don't buy ranking, you earn consensus. And consensus is built from the outside, with real presence, not from the inside, with tricks. Let's move to the workbench.

Do it now

Practice

To take with you: the currency of marketing in AI shifted from link to mention. What gets AI to recommend you isn't whoever links to you, it's your name cited by third parties, hammered across several independent sources (the consensus signal), with consistent facts that make you an entity it recognizes. A fraudulent shortcut (hiding an instruction for AI) isn't just illegal, it downranks you silently. The game is earning consensus, not buying it. Fair enough? Next up.

Let's close the point together, because it changes where your marketing effort goes. You spent years in a logic of chasing links and dominating keywords. Forget that as a priority. Going forward, the question that pays off is different: when someone asks AI about my market, does my name show up, and does it show up correctly? For it to show up, you need to be on third parties' lips, hammered, in places AI trusts, with a facts sheet so clean it can't confuse you with anyone else. This isn't fabricated with a trick, it's built by being good and being cited for it. Whoever understands that mention beats link stops shouting about themselves and starts becoming the name the market repeats on its own. And that name, and only that name, is what AI recommends.

For the board