The people who need your expertise are asking a chatbot, not a search bar. Here is what the measured evidence — not folklore — says about becoming the source the answer cites.
TL;DR — how to get cited by AI search, in five lines:
I spend a lot of time telling Founders that packaged expertise beats hourly expertise. But a package nobody finds is inventory, not income. And the finding layer is changing underneath us right now: the people who need your twenty years of judgment increasingly don't type keywords into a search bar. They describe their whole situation to a chatbot and read one synthesized answer. Either your name is in that answer, or the transaction happens without you.
This dispatch is my working file on that problem. Every mechanism claim below traces to published research or the engines' own documentation. Where the honest answer is "nobody knows," I say so, because the AI-visibility space is already filling up with people selling certainty they do not have.
The hype version says search is dead. The measured version is more interesting. TechCrunch, reporting Similarweb data, counted 1.13 billion AI-platform referrals to the top 1,000 websites in June 2025 — up 357% year over year. In the same month, Google Search sent 191 billion. AI referrals were well under 1% of Google's volume. ChatGPT alone drove over 80% of the AI share.
So why care about a stream that small? Because of what's in it. Seer Interactive's case study — one large site, October 2024 to April 2025, roughly 11,000 AI sessions against 14 million Google organic sessions — found ChatGPT referrals converting at 15.9% versus 1.76% for Google organic. About nine times better. Perplexity converted at 10.5%, Claude at 5%. Seer's own explanation is the sober one: the premium is intent, not anything mysterious about the channel. Visitors do their qualifying inside the chat — comparing, objecting, narrowing — and click through only when they're nearly decided.
And the stream is compounding. Similarweb's generative-AI statistics show total AI referral traffic growing more than 3x between September 2024 and September 2025, with ChatGPT referrals converting at 7.1% — second only to paid search at 7.8%, and ahead of organic, direct, social, email, and display. Semrush's clickstream study (over a billion lines of US clickstream data) found ChatGPT's outbound referrals up 206% year over year even as total ChatGPT visits plateaued — and the number of unique domains receiving those referrals grew from roughly 71,000 to 170,000. The pie of cited sources is widening. There is room at this table for individuals, not just publishers.
Strip the mystique and the pipeline is knowable. When web search is engaged, an answer engine runs retrieval — usually leaning on a conventional search index — pulls a shortlist of pages, and synthesizes an answer with citations to the pages it leaned on. Two hard facts fall out of the published evidence:
Fact one: access is binary. OpenAI's crawler documentation is explicit: OAI-SearchBot is what surfaces websites in ChatGPT's search features, and "sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers." Note the separation of concerns — GPTBot handles training data, ChatGPT-User handles user-initiated browsing, and neither determines search inclusion. I have watched experts block every bot with "AI" in the name out of principle and then wonder why they're invisible. Check your robots.txt before you do anything else on this page.
Fact two: AI citation is mostly organic ranking, restated. Ahrefs studied 1.9 million citations across a million Google AI Overviews: 76.1% of cited pages already rank in Google's top 10, and about 86% appear somewhere in the top 100. The single most-cited source in an answer has a median organic rank of position 2. The romantic story — that AI search resets the game and lets unknowns leapfrog incumbents — is not what the data shows. The engines trust what search already trusts. Everything you've done for organic visibility carries over; nothing is wasted.
Given retrieval, the page-level question becomes: what does a machine prefer to quote? Here is my position, and I'll argue it: answer engines cite pages that behave like sources, not pages that behave like ads. A source defines its terms, states numbers, names where the numbers came from, and answers the question in the first screen instead of after four paragraphs of throat-clearing.
The logic is mechanical, not aesthetic. A language model assembling an answer needs extractable material: a clean definition it can compress, a statistic it can attribute, a step sequence it can restate. A page that opens with "In today's fast-paced world..." gives the model nothing to grab. A page that opens with "X is Y; the measured effect is Z, per this named study" is quotable in one pass. This is also why one complete page beats five shallow ones — Semrush's same clickstream study found 65–85% of ChatGPT prompts don't match traditional search keywords. People ask long, situational, conversational questions. The page that survives that shift is the one that fully covers a problem, not the one tuned to a two-word keyword.
Practically, for an individual expert, that means: pick the ten questions you have answered most often in your career. Write one definitive page per question. Lead each with the answer. Load each with the specific numbers, edge cases, and failure modes only someone with your mileage knows. Your unfair advantage against big publishers is not volume — it's that your pages can contain judgment they cannot fake.
Engines cite entities they can resolve. If you're "Jane Doe" on your site, "J. Doe Consulting" on LinkedIn, and "the Ops Whisperer" on YouTube, a retrieval system sees three weak entities instead of one strong one. The fixes are boring and free:
This is the same infrastructure thinking I keep coming back to from the book:
Get great at building systems and you can build almost anything. — The Mentor Economy, Chapter Two
Entity consistency is a system. You set it up once, and every future mention, interview, and page compounds into the same identity instead of leaking into three.
This dispatch is the field summary. The full playbook — page templates, the crawler-access checklist, and the entity-building sequence in order — is Get Found by AI, one of the books in the library. Free; you cover shipping.
See the books →Now the section most guides omit. There is no lever that makes an answer engine cite you. Retrieval behavior changes without notice: web search was active on only 34.5% of ChatGPT queries as of February 2026 (Semrush), meaning most answers still come from model memory where no page tweak reaches. Engines change their citation formats overnight — when ChatGPT added clickable brand links in May 2026, Similarweb measured total referrals up 157.7% in a single week and homepage referrals up 354.7%. That windfall landed on people who had built citable pages before the update, and it could reverse just as fast. (Encouragingly, the post-update visitors viewed 24% more pages and spent 11% more time on site — the quality of this traffic is rising, not just its quantity.)
So the honest posture is portfolio thinking, not lever-pulling. You cannot control whether next month's model update favors forums, or news sites, or your niche. You can control whether, whenever an engine goes looking for a trustworthy source on your topic, you are the least risky page to quote: accessible to the crawlers, already ranking, definitionally written, evidence-dense, and attached to one unambiguous human being. That asset appreciates under every scenario — including the one where AI search stalls and classic search keeps its 191-billion-referral crown.
This discovery layer is one piece of a larger machine. The pages that get you cited are the same pages that feed the AI clone that scales your judgment, fill the top of the book → course → mentorship ladder, and compound the numbers I laid out in the mentor economy, measured. Get found. Then have something worth finding.
One non-negotiable: your site must allow OpenAI's OAI-SearchBot crawler. OpenAI's own documentation states that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers. Beyond that, the strongest predictor of being cited is ranking well in classic organic search — Ahrefs found 76.1% of AI Overview citations already rank in Google's top 10.
The volume is small but the intent is unusually high. Seer Interactive measured ChatGPT referrals converting at 15.9% versus 1.76% for Google organic on the same site — roughly nine times better — because visitors pre-qualify inside the chat before they ever click. Similarweb's data shows ChatGPT referrals converting at 7.1%, above organic search, social, email, and display. Small stream, dense with buyers.
It helps machines understand you; it does not guarantee citations. Google and Microsoft have both said structured data gives their AI systems an advantage, but an independent analysis found no correlation between schema coverage and citation rates, and OpenAI, Perplexity, and Anthropic have not disclosed whether they use schema at all. Ship it as cheap insurance, not as a strategy.
Nobody can promise a timeline, because retrieval behavior shifts with every engine update. What compounds predictably is the underlying asset: definitional pages that answer one question completely, consistent entity signals (same name, same bio, same claims everywhere), and organic rankings — which remain the strongest measured predictor of AI citations. Build those and every engine update works in your favor.
Author of The Mentor Economy and co-founder of MentorMe. He writes about turning hard-won expertise into AI-leveraged one-person businesses.
The Mentor Economy is the full system — free, you just cover $9.95 shipping.
Get Your Free Copy