How to Get Cited by ChatGPT (2026): What My Tracking Data Says Actually Works

TL;DR: ChatGPT Doesn't Cite Experts. It Cites Institutions, Crowds, and Price Tags.

Almost every "how to get cited by ChatGPT" guide is written by an SEO tool vendor who has never been cited by ChatGPT. Here's a different source: my own tracking data, including the embarrassing part.

I run a weekly citation monitor across two engines and 90+ commercial buyer queries — the methodology and full dataset are in who do AI engines actually cite. The finding that matters for this post: my expert blog holds 14 citations on Perplexity and zero mentions in ChatGPT's citation dataset. Meanwhile my productized-services site, which mostly publishes prices, is cited by ChatGPT on 10 commercial queries. Same owner, same week, opposite outcomes. That split isn't bad luck — it's the clearest possible demonstration of what ChatGPT rewards.

What ChatGPT Actually Cites (the Numbers)

From DataForSEO's LLM-mentions dataset (US, English, pulled July 31, 2026), two prompt spaces I track as of 2026:

• "AI consultant" space (405 mentions): businessinsider.com 131 · reddit.com 110 · bls.gov 58 · timesofindia.indiatimes.com 54 · axios.com 41.
• Bookkeeping space (24,130 mentions): reddit.com 5,963 · indeed.com 1,992 · forbes.com 1,585 · quickbooks.intuit.com 1,474 · nerdwallet.com 1,449.

Read those lists again: press, forums, government statistics, job boards, product documentation. Not one individual practitioner. But before you despair — niche sites do break through: legalclarity.org holds 550 mentions in the space where Forbes holds 1,585, and accountinginsights.org holds 130. Unknown sites out-cite famous brands when their content is shaped like the answer. The bar isn't fame. It's citability.

The 4-Move Playbook, In Order of Evidence

1. Publish your prices. This is the strongest signal in my entire dataset: every ChatGPT citation my business holds is a "how much does X cost" page with real numbers on it. When a buyer asks ChatGPT what something costs, a page that states a price is the citable answer, and almost none of your competitors will publish one. This works today, and it converts the humans who arrive, too.

2. Be genuinely useful on Reddit. Reddit is the only domain in the top three of both engines I track — 110 mentions in the AI-consultant space, 5,963 in bookkeeping — and it has licensing deals with the major AI companies, so the content flows into models in a way most forums' doesn't. Answering questions honestly in the subreddits where your buyers already ask them is a legitimate channel. Being promotional there gets you removed, which is precisely why it still works.

3. Earn press, even small press. Business Insider, Axios, Forbes, trade outlets — institutional bylines dominate ChatGPT's source mix on informational queries. One quotable data point pitched to one trade publication does more for your ChatGPT visibility than a year of posting on your own blog.

4. Publish original data. The only content type that press AND language models both have a reason to cite is data nobody else has. Track something in your industry monthly and publish the numbers. (This post you're reading is that strategy, executing itself.)

Micro-action for today: pick your highest-intent "how much does X cost" question, and put a real number on a real page before dinner. That's move one, done.

"ChatGPT SEO" Is Two Different Jobs — Be Clear Which One You're Doing

A note on the phrase everyone searches: "ChatGPT SEO" gets used to mean both "using ChatGPT to do SEO work" and "doing SEO so ChatGPT recommends you." This post is the second job — the visibility one — and it has almost nothing in common with classic SEO tactics. There's no keyword density to tune and no backlink exchange to run; there's a source-selection layer deciding whom to trust, and the levers are the four above. The broader discipline (AEO/GEO, covering all the engines) is mapped in my AI search optimization guide and answer engine optimization; the writing-level mechanics are in LLM SEO.

Why Your Perplexity Wins Won't Transfer (and Vice Versa)

Perplexity rewards named experts answering specific questions; ChatGPT rewards institutional authority and crowd consensus. My 14 Perplexity citations came from answer-shaped expert posts — question as the H2, answer in the first paragraph, specific numbers, real author bio. That exact body of work is invisible to ChatGPT. If your buyers research on Perplexity, keep doing that (it works, and I've documented how in the citation data post). If they ask ChatGPT — and on commercial "what should I buy" queries, they increasingly do — you need the playbook above. Different engines, different games, and as of 2026 you have to pick deliberately rather than hoping one strategy covers both.

Measure It or You're Guessing

Everything in this post exists because I measure instead of assuming: a weekly Perplexity check across my money queries at about a penny a query, plus the LLM-mentions dataset for ChatGPT's side. The full monitoring setup — including what to track and what it costs — is in how to track your brand mentions in AI.

Want the shortcut? I'll run your domain through the same citation checker I use on client engagements — where AI engines mention you, where they recommend competitors instead, and the three fixes that matter most for your business. Free, and considerably less humbling than finding out the way I did.

Related guides: who do AI engines actually cite, track your brand mentions in AI, AI search optimization, answer engine optimization, LLM SEO.