Can you actually rank in ChatGPT the way you rank in Google?

Not in the same mechanical way: there's no ranked list of ten blue links. But ChatGPT does choose which businesses to name, and that choice is driven by signals you can influence: how clearly your entity is defined, how many trusted third-party sources cite you, and how extractable your own pages are. “Ranking” in ChatGPT means becoming one of the handful of businesses the model reaches for when someone asks for a recommendation in your category.

How does ChatGPT decide which businesses to recommend?

Answers may draw on existing model knowledge and live retrieval. An outside operator cannot inspect or change the training data or model weights, so the controllable work is narrower: clear public identity, crawlable attributable content, credible corroboration, and repeated observation of which sources appear for the agreed prompts.

How long does it take to start showing up in ChatGPT?

Entity and on-page fixes can register within a few weeks once the pages are re-crawled and the browsing layer picks them up. The slower work is third-party citations, getting named in the publications, directories, and review sites ChatGPT trusts, which compounds over months. Treat it like SEO: the foundation is fast, the authority is a campaign.

Does traditional SEO still matter for ChatGPT visibility?

Yes, more than most people expect. ChatGPT search leans on a live web index, so being crawlable, indexed, and structured well still determines whether your pages can be fetched and quoted. Strong technical SEO, clean schema, and topical authority are the price of entry. AI visibility is built on top of them, not instead of them.

Do I need to be on Wikipedia to rank in ChatGPT?

It helps but it isn't required. Wikipedia is a high-trust source models lean on for entity facts, so a legitimate presence there strengthens recognition. But most businesses win visibility through the broader citation network, industry publications, credible directories, review platforms, podcasts, and roundups, long before they'd ever qualify for a Wikipedia article.

How do I measure whether I'm ranking in ChatGPT?

Run the real prompts your buyers would type, “best [category] in [city]”, “who should I hire for [service]”, and record whether you're named, how you're described, and who's named instead. Do it repeatedly, because answers vary. This is what share-of-model tracking measures; Joel co-founded Outrigger with Andrew partly to run it at scale instead of by hand.

Is ChatGPT SEO different from Google SEO?

The mechanics differ but the foundations overlap. Google ranks pages; ChatGPT names entities and quotes sources. That shifts weight toward being cited by others, being unambiguous about who you are, and structuring content for extraction rather than for a click. It's a discipline that sits alongside SEO, often called generative engine optimization, not a replacement for it.

Should I hire an agency for ChatGPT SEO or do it in-house?

You can start in-house: fix your entity, structure your pages, and reclaim obvious citation gaps. The part that's hard to run solo is the digital-PR engine that earns trusted citations and the instrumentation that tells you whether any of it moved the answer. That's where an agency with its own tooling earns its keep: you're paying for the measurement loop as much as the work.

All Posts

AI Search / GEO

How to Rank in ChatGPT: An Operator's Guide for 2026

Joel HouseForbes Agency Council

Founder, Xpand DigitalJuly 11, 202614 min read

“How do I rank in ChatGPT?” is the question I get on nearly every sales call now, and most of the answers online are written by people who have never watched the numbers move.

I run an agency that got tired of guessing. Joel co-founded Outrigger with Andrew as a separate AI-visibility joint venture, specifically to measure who ChatGPT names and why. When digital PR is in scope, Xpand may use PressForge to organize research, pitching, and follow-up. I also wrote a book about turning AI into a revenue channel rather than a novelty. So this guide is not theory. It's written from the operator's chair.

Here is the uncomfortable baseline. Outrigger's April 2026 AI Visibility Index , 95,392 data points across 1,000 businesses, found that 65.9% of businesses are effectively invisible in AI search. Two of every three companies never get named when a buyer asks ChatGPT for a recommendation in their category. That is the gap. This guide is how you climb out of it.

One framing point before the steps. There is no ranked list of ten results inside ChatGPT. “Ranking” here means becoming one of the small set of businesses the model reaches for when someone asks it to recommend, compare, or shortlist. Your job is to make yourself the obvious, well-evidenced answer, and then to measure whether it worked. That last part is where almost everyone quits.

Disclosure: I co-founded Outrigger with Andrew as a separate joint venture. Xpand may use PressForge to organize digital-PR workflow. I'll give you the principle first every time: each move works whether or not you use either tool.

What “ranking” in ChatGPT actually depends on

Before the tactics, you need the mechanism, because it dictates every move that follows. ChatGPT names businesses through two layers, and you have to win both.

Existing model knowledge may affect an answer, but an outside operator cannot inspect, edit, or verify the underlying training data or weights. Treating publicity as a direct model-training lever claims access nobody outside the provider has.

The browsing and search layer , ChatGPT search, which grounds answers with a live web index and real-time page fetches, reads current pages at answer time. This is the layer you can influence fastest, and it's why classic SEO still matters: if your page can't be crawled, indexed, and cleanly parsed, it can't be retrieved and quoted.

The controllable work has three parts: be unmistakably clear about who you are, earn credible public corroboration, and make useful answers easy to retrieve and evaluate. The moves below make those inputs concrete without claiming control over the resulting answer.

The seven moves that get you named in ChatGPT

Run them roughly in order. The early moves are the foundation the later ones stand on: there is no point chasing citations if the model can't tell which business you even are.

Nail your entity before anything else

An entity is who you are in the model's eyes: your name, your category, your location, what you do, and who you're distinct from. If that's fuzzy or contradictory across the web, ChatGPT hedges, and a hedged model reaches for a competitor it's more certain about. Ambiguity is the single most common reason a real business stays invisible.

Do this

Make your name, category, and description identical everywhere they appear: your site, Google Business Profile, LinkedIn, Crunchbase, industry directories, and review platforms. Add Organization and sameAs schema to your homepage tying every profile together. Write one crisp sentence, “X is a [category] in [place] that does [outcome] for [buyer]”, and use it verbatim in every bio. Consistency is the signal.

Joel's take

Do this first, always. I've watched businesses double their mention rate on nothing but entity cleanup, because they weren't losing to competitors: they were losing to their own contradictory descriptions. Boring work, biggest lever.

Earn credible third-party corroboration

Independent sources can support search authority, corroborate public facts, and appear in retrieval or citation sets. They cannot make a business the default answer on a guaranteed basis.

Do this

A responsible digital-PR program can include original commentary, data hooks, expert roundups, podcast guesting, and relevant editorial outreach. Xpand may use PressForge to organize research, pitching, and follow-up, but editors decide what is published and engines decide what is retrieved or cited.

Joel's take

Skip mass low-quality listings. Prioritise accurate profiles and editorially credible sources your buyers would genuinely use. Publication and citation remain separate decisions.

Structure your pages so a machine can lift the answer

When ChatGPT's browsing layer fetches your page, it's hunting for a clean, self-contained answer it can quote without editing. Pages built to make a human scroll and “discover” the point are hostile to extraction. Pages that answer the question in the first two sentences of each section get quoted.

Do this

Lead each section with a direct-answer paragraph: the claim, stated plainly, before the elaboration. Use real questions as your H2s (the exact phrasing buyers use). Add a genuine FAQ section with FAQPage schema. Keep facts in short, self-contained sentences with named numbers rather than buried in prose. Every page on this site follows the same pattern: it's not a coincidence.

Joel's take

Write the sentence you'd be happy to see quoted back to you in a ChatGPT answer, then put it first. If you have to read three paragraphs to find your own point, so does the model, and it won't bother.

Publish original data with named numbers

Models gravitate to primary sources they can attribute. A statistic that only exists on your page, a benchmark, a survey result, a proprietary number, forces every downstream article to cite you, which feeds both the training layer and the live layer. It's the most valuable content you can make for AI visibility.

Do this

Answer a question your industry argues about with data no one else has: survey your customers, aggregate your own campaign results, or analyze a public dataset from a fresh angle. Package it as a citable report with a clear headline number, then get it in front of journalists. The AI Visibility Index we published is exactly this play: it earns citations precisely because the figure lives nowhere else.

Joel's take

One good study outperforms a year of generic blog posts for AI visibility. Specificity wins: “median cost per lead across 500 campaigns in [niche]” beats “the state of marketing” every time.

Fix your presence on the sources ChatGPT actually reads

ChatGPT leans disproportionately on a predictable set of high-trust sources for category recommendations: Wikipedia, established review platforms, credible directories, community threads, and the “best [category]” roundups that dominate the SERP. If a competitor owns those and you're absent, the model repeats what it reads.

Do this

Audit the pages that surface when you search your category, then close the gaps: claim and complete your review profiles, get accurate listings in the directories that rank, earn inclusion in the legitimate roundups, and answer the recurring questions in the communities where your buyers gather. A defensible Wikipedia presence helps if you genuinely qualify, but don't force it; the broader citation network does most of the work.

Joel's take

Type the prompt your buyer would type, read the top ten sources ChatGPT could pull from, and ask: am I on these pages, and am I described the way I'd want? That list is your to-do list.

Instrument it: measure share of model, not vibes

Here is where two-thirds of the market fails: they do the work and never check whether it landed. AI answers are non-deterministic: ask the same question twice and the wording, and sometimes the names, change. You cannot manage that by asking ChatGPT once and screenshotting a good day.

Do this

Build a fixed set of buyer prompts, “best [category] in [city]”, “who should I hire for [service]”, “alternatives to [competitor]”, and run them on a schedule, logging whether you're named, how you're described, and who's named instead. Track the trend, not the snapshot. This is what “share of model” measures, and running it by hand across dozens of prompts is exactly why Joel co-founded Outrigger with Andrew.

Joel's take

If you can't answer “what percentage of my target prompts name us this month versus last?” you're not optimizing, you're decorating. The measurement loop is the difference between a strategy and a hope.

Feed the live search layer: stay crawlable and current

ChatGPT search grounds answers in a live index and real-time fetches, so the technical fundamentals that decide whether Google can read you decide whether ChatGPT can too. A brilliant page that's slow, blocked, or unindexed is invisible to the layer most likely to name you this week.

Do this

Confirm your key pages are crawlable and indexed, keep them fast, and don't block the AI crawlers you actually want reading you. Publish an llms.txt that points to your best source pages, keep your facts current (models distrust stale numbers), and make sure the pages that answer buyer questions are the ones that rank: the live layer often pulls from the same results a search engine would show.

Joel's take

AI visibility isn't a replacement for technical SEO: it's built on top of it. If the plumbing's broken, none of the first six moves reach the tap.

“You don't rank in ChatGPT by writing for ChatGPT. You rank by making the trusted web certain about who you are, and then measuring until it shows.”
Joel House · Founder, Xpand Digital

What a realistic timeline looks like

The honest sequence, from the campaigns we run:

Weeks 1–3: entity cleanup (Move 1) and page structure (Move 3) ship. Once pages are re-crawled and the browsing layer picks them up, you can start appearing in grounded answers that pull live sources.
Weeks 3–12: source-page fixes (Move 5) and the first citations (Move 2) accumulate. Mentions become more frequent and more consistent across repeated prompts.
Months 3–9: the citation network and any original data (Move 4) start influencing the model's default framing of your category. This is the slow, durable win, and the one that survives model updates.

The measurement loop (Move 6) runs the entire time, because without it you can't tell whether you're trending up or just having a good screenshot day.

The mistakes that keep businesses invisible

The 65.9% invisibility rate in the Index isn't random. It clusters around the same handful of errors:

Optimizing content, ignoring citations
Rewriting your own pages forever while never earning a single trusted third-party mention. Your pages help the live layer; only citations move the base model. You need both.
Treating one good answer as proof
Asking ChatGPT once, seeing your name, and declaring victory. Answers vary run to run. Without repeated, logged prompts you have an anecdote, not a signal.
Contradicting yourself across the web
A different category on LinkedIn than on your homepage, a different service list on your directory profiles. The model reads the contradiction and hedges toward a competitor it's sure about.
Chasing volume over trust
Buying mass listings and low-quality placements. They do not establish credible corroboration and may create inconsistent or spammy public records.
Skipping the technical fundamentals
Blocking crawlers, shipping slow or unindexed pages, letting facts go stale. The live layer can't quote what it can't fetch or trust.

Ranking in ChatGPT is one engine in a larger discipline. The same entity, citation, and structure work pays off across every AI answer surface: the tactics just get pointed at different models:

ChatGPT SEO how we run this as a managed program, with the citation engine and measurement loop built in.
Generative engine optimization the umbrella discipline covering ChatGPT, Perplexity, Gemini, and Google's AI answers together.
Answer engine optimization structuring content so machines can extract and cite it, the on-page half of Move 3.
AI visibility audit the fastest way to see which of your target prompts name you today, and which name a competitor instead.
How to get found in AI search the broader 2026 guide across ChatGPT, Perplexity, and Gemini.

Frequently Asked Questions

Joel House

Founder, Xpand DigitalJuly 11, 2026

Want to know if ChatGPT
names you or a competitor?

We'll run your real buyer prompts through the major AI models, show you where you stand against the businesses being named instead, and map the moves that close the gap. Measured, not guessed.

Get an AI Visibility Audit See how ChatGPT SEO works

Put it to work

Work with the team that wrote the playbook.

Where to start