AI ignores your brand because it has nothing it can verify to say about you. When someone asks ChatGPT or Gemini “who is the best in Houston,” the engine does not rank a list, it names a source it can find, understand and trust. Most expert firms have left no machine-readable trace of the thing that makes them the best, so the machine reaches for whoever it can stand behind instead. You do not force the mention by gaming anything. You force it by becoming the most verifiable answer to the question. Ahrefs studied 75,000 brands and found that mentions of a brand across the web predict its visibility in AI answers roughly three times more strongly than backlinks.
If you have asked an AI tool the exact question you get paid to answer and watched it recommend someone else, you know the feeling. You are the best at what you do. The machine just cannot see it. That is not a verdict on your work. It is a verdict on the trail you have left, and the trail is the part you can fix.
Here is the uncomfortable part. The old playbook does not force the mention. Ranking first on Google, stacking backlinks, rewriting meta descriptions: none of it reliably moves an engine to say your name. Ahrefs found that only about 38 percent of the pages cited in Google’s AI Overviews now rank in the top ten organic results, down from 76 percent a year earlier. Being on page one and being the name AI repeats have quietly become two different games.
At Probably Genius this is the whole job: building a version of your firm that AI can read, then making it the source a cautious machine is willing to recommend. This guide covers why AI skips you, why the mention has come loose from the ranking, and what actually forces it.
Why AI Skips You, and What It Reaches For Instead
Three things have to be true before any engine names you: it has to find you, understand you, and trust you. A brilliant practitioner with a thin, marketing-shaped website fails all three at once. There is nothing structured to find, nothing specific to understand, and no verifiable proof to trust. So the engine does what a cautious system always does. It reaches for the source whose expertise is written down in a form it can stand behind.
This is why asking a model to generate a few articles about you backfires. In 2026 that produces what we call Training Data Echo, content that reads like the material the model was already trained on. The machine recognizes its own reflection and rarely cites it. The firms AI skips are not the ones with too little content. They are the ones whose real expertise never made it into a form a machine can read, the phenomenon we call the Black Box Effect.
The fix is not louder. It is clearer and more verifiable. For a fuller look at how that decision gets made, read how ChatGPT decides who to recommend, and for the discipline underneath it, what Generative Engine Optimization actually is.
The Mention Has Come Loose From the Ranking
For two decades the goal was to rank. Get to position one and the clicks followed. AI answers broke that link. When Ahrefs analyzed 863,000 search results and 4 million AI Overview citations, it found that only about 38 percent of the cited pages still ranked in the top ten organic results, down from 76 percent the year before. The engine increasingly splits a question into smaller ones and assembles its answer from sources scattered far below the first page.
Read that the right way and it is good news. You no longer have to outrank a national competitor with a ten-year head start to be the name AI says. You have to be the clearest, most verifiable answer to the specific question. A single store can be cited beside a national chain. Mattress Warehouse, one shop in San Marcos, Texas, is now cited by AI alongside Forbes and the American Sleep Association, not because it outranks them, but because it answers the question better in a form the engine can lift.
The catch is that the win does not travel. Yext found only about 11 percent of cited domains overlap across engines. The firm ChatGPT loves can be invisible on Gemini and Perplexity, because each engine retrieves differently. Forcing the mention is not a single trick. It is building a presence that holds up everywhere AI looks, which is a reputation problem, not a ranking one.
What Actually Forces the Mention
Start with the strongest signal in the data. In that 75,000-brand study, the factors most associated with AI visibility were not links. They were mentions: a brand named across the web correlated with AI visibility at 0.664, more than three times the 0.218 for backlinks. Correlation is not proof of cause, but a gap that wide is hard to ignore. It is starker at the top. Brands in the highest quartile for web mentions averaged 169 AI Overview mentions, against 14 for the quartile below. Being talked about, in public, in places a machine can read, is what moves the needle.
The second signal is ownership. Yext’s analysis of 17.2 million citations found 86 percent came from sources a brand controls, chiefly its own site and its listings. That is the part most experts never claim. The surfaces AI cites most are the ones you can fix this week: your own pages, your structured data, your consistent details across every directory that feeds an engine.
The third signal is information gain. Engines favor content that says something they have not already absorbed, which is why citations, quotations and concrete statistics lifted source visibility by up to 40 percent in the Princeton study, while keyword stuffing dragged it below baseline. What forces the mention is the combination: be mentioned across the web, claim and structure the surfaces you control, and publish the one thing only you can say.
- Earn mentions in public, in writing, where a machine can read them. Unlinked mentions in neutral, credible places now outweigh link-building.
- Claim and structure what you control: your site, your schema, your listings, your details, consistent everywhere.
- Publish proof, not prose: the methodology, the real numbers, the case outcome no competitor can reproduce.
- Build for every engine, not one, because the win on ChatGPT does not carry to Gemini or Perplexity.
The fastest place to start is with the mentions you have already earned and never collected. Every supplier that lists its stockists, every association with a member directory, every podcast that still has your episode in its notes, every conference that published its speaker list: each one is a neutral, credible page where your name can sit in machine-readable text. Ask for the listing. Then make sure the name on it matches your website and your schema exactly, because an engine treats “Smith & Co” and “Smith and Company” as two weak entities rather than one strong one. No link required. The mention itself is the signal.
How We Force It on Purpose
Probably Genius does not start with your website. We start with you. The Genius Interview maps the expertise that has never been published, the judgement and the frameworks that only exist in your head and your client work. We structure that into Entity Infrastructure a machine can read, connected by a single consistent narrative we call the Golden Thread, so every engine assembles one coherent, authoritative picture of who you are instead of fragments.
Then we publish Net New Intelligence, content built on the gap between what AI already knows and what only you know, produced through a multi-model architecture of specialized agents that research, write, engineer and check each other’s work. None of it ships unproven. Every piece must pass the AI Integrity Standard, our 100-point gate that does not publish anything scoring below 85, which is how we avoid adding to the Training Data Echo.
The pattern repeats across client builds. BY Projects, a Melbourne architecture firm, became the number one heritage architect Melbourne and now fields 14 qualified leads a month, with clients calling daily to say “ChatGPT told me to call you.” Harmony Group went from invisible to recommended in three weeks. PSS went from zero to recommended for NDIS queries in the same window, ahead of providers with a ten-year head start. A practice that Gemini once described as untrustworthy was repositioned into a credibly cited expert with 29 years and 50,000 clients behind it. The payoff is the Researched Buyer: people who arrive already recommended, which is why AI-referred visitors convert far better than organic search, around 50 percent higher in Shopify’s 2026 data. Scores run 0 to 109. Ninety and up means AI treats you as a primary source. Under 35 means you are effectively invisible. For most firms the mention is not out of reach. It is unclaimed.
Want to Learn More?
Probably Genius was built by Jacquie Baker and Christopher Shaw, who spent more than 40 combined years and two agencies translating what makes an expert the best in the room for the system that decides who gets chosen. The audience changed from investors to AI. The skill did not. We map genius, build the infrastructure that makes it verifiable, and publish the proof that gets it mentioned.