How to get your brand mentioned by ChatGPT

By Dmitry (Dee) Kargaev · Updated November 11, 2026

To get ChatGPT to mention your brand, let AI crawlers read your site, set up profiles on platforms that let AI crawlers in, link those profiles to your site with markup, and measure. You finish with a tally of which assistants name you. Nothing here guarantees a mention.

This page was written by Dmitry (Dee) Kargaev, who builds and maintains Awesome GEO, a free, open-source catalog for Generative Engine Optimization (GEO), and wrote the two working papers cited below. The free browser tools used in the steps belong to that catalog.

What the evidence says

The SEO-to-GEO Gap: Quantifying Ranking Factor Divergence Between Traditional and Generative Search is a working paper, and its findings are provisional. The first: visibility in generative search appears more tightly associated with entity and brand prominence than with classic link-proxy measures. The second: adding citations and statistics to content produces substantial gains in the GEO benchmark. The third: the overlap literature indicates that GEO still operates on top of organic search rather than replacing SEO outright. The paper argues for reweighting rather than replacement, and says some headline claims about SEO-to-GEO disruption are stronger than the evidence supports.

Other publishers add claims of their own. TrendlyAI states that ChatGPT names brands from two sources: what the model learned in training and what it finds when it searches the web live. It also states that no paid placement puts your brand inside ChatGPT’s recommendation text. QuickCreator makes the same point about paid placement and adds that there is no form for submitting your business to ChatGPT. These are the publishers’ statements, not findings of the paper.

Before you start

You need the public address of your site, a way to add a script tag to its pages, and access to ChatGPT, Claude, Gemini and Perplexity, because you ask each of them yourself. The Awesome GEO tools used here are free. Field and button names below are as the tools read on 11 November 2026.

Site check fetches your page from the Awesome GEO server, so local and private addresses are refused, and so are responses over 1,000,000 bytes. If the fetch fails, it offers “Paste HTML instead” and “Try again”. Paste HTML mode skips “Crawler access”, “Live fetch as each bot” and “Files”, because it has only the HTML to read.

Steps

Open Site check, choose “URL”, type your address into “Page URL”, press “Check”, and read “Crawler access”, where each AI crawler shows as allowed, partial, blocked or unknown.
Read “Problems” at the top of the results, which shows a count and a list or “No problems found”, then “Live fetch as each bot”, which shows the HTTP status code returned to each crawler. A footnote says this tests the user agent only, not the crawler’s verified IP ranges.
If a crawler you want shows as blocked, open the robots.txt tool, type your domain into “Site” and press “Load”. Choose a preset; the default is “Let AI search read me, keep training bots out”. Under “Bots”, click a crawler chip to switch it between allowed and blocked, then press “Copy” in the “robots.txt” section.
Go to the Presence planner, choose “Person” or “Organization” under “Type” and, if you like, type a few words into “Focus”. “Start here” then lists six platforms from the Awesome GEO catalog, each with its rank, name, description and score. The ranking uses your type, AI crawler access, link type, effort, indexability, domain rating and the words in “Focus”. The score is out of 100 and is built from crawl access, popularity (Tranco rank), authority (Open PageRank), link type and effort.
Create a profile on each of the six, with the same name, one-line description and category everywhere. TrendlyAI’s guide names that consistency as its first fix. Of the 157 platforms in the catalog, 136 allow at least one major AI crawler, and a script re-checks each platform every Monday at 06:00 UTC.
In the planner’s “sameAs” section, press “Copy”. The block is JSON-LD with a “sameAs” list of profile URLs that end in placeholders for your handle; replace each placeholder with your real handle.
If you want more fields, use the Schema tool instead: choose “Organization”, fill “Name”, “Website”, “Logo URL”, “Founder” and “Profiles (sameAs)”, and press “Copy”. The JSON updates as you type and empty fields are left out. For an individual, choose “Person” and use “Name”, “Job title”, “Website”, “Works for” and “Profiles (sameAs)”. A note in the tool says the markup makes clear which entity you are.
Paste the copied JSON inside a script tag of type application/ld+json on your site. The copy has no script tags around it.
Run Site check again with “Check again” and read the “JSON-LD types” and “Parse errors” rows under “Structured data”.
Fill the Answer panel fields “Product” and “Category”, which are required, and add up to three “Competitor” fields. Press “Write questions” for five questions: a comparison, an alternatives, a best-of, a how-to and a pricing question.
Press “Copy all” and ask the five questions in ChatGPT, Claude, Gemini and Perplexity yourself, because the Answer panel never queries them.
In the “Tally” table, click each cell to switch it between absent, named and cited only; the “Named” row holds counts such as “2 of 5”. The TrendlyAI guide separates a mention, your name in the answer text, from a citation, a source link attached to the answer. Your tallies stay in your browser.
Keep the five questions unchanged and ask them again later. The same guide advises a fixed set of prompts checked on a schedule, so you see change over weeks rather than guess from one answer.

If ChatGPT still does not name you

The sources give a few reasons a brand goes unnamed. Each is the publisher’s own statement.

The Darl states, from its own brand audits, that the most common gap is zero third-party mentions ChatGPT can draw from, and that ChatGPT ignores brands it cannot verify quickly. TrendlyAI states that if a brand is new, or described five different ways, the link between your category and your name is weak or missing. Sightivo states that one mention on one blog won’t register, while being referenced repeatedly across many credible sources will.

If your tally shows cited only, TrendlyAI states that ChatGPT can cite a blog post on a how-to question without ever recommending your product.

The working paper SEEN: A Four-Layer Framework for Generative Engine Optimization asks whether an entity is easy for AI systems to find, read, verify, resolve and recommend. It treats owned content, platform-hosted first-party content and third-party corroboration as one body that must read as a coherent, verifiable entity. It does not claim the framework guarantees citations, recommendations, rankings or traffic.