How do I get my product recommended by ChatGPT and other SI assistants? (2026)
To get your product recommended by ChatGPT and other SI assistants, let their search crawlers in, get mentioned by name on other sites, and answer buyer questions in quotable passages. In Ahrefs' 75,000-brand study, web mentions correlated with ChatGPT visibility at 0.664; backlinks only around 0.2.

To get your product recommended by ChatGPT and the other SI assistants in 2026, three things have to be true: their search crawlers can reach your pages, other sites mention you by name, and one of your pages answers a buyer's question in a passage the model can quote whole. Backlinks and domain authority matter far less than most guides claim.
That's the short version. The rest of this post is the evidence, the crawler settings, and how to check whether it's working, because almost every page ranking for this question skips the last part.
You probably found this the way your buyers will find you. You asked an assistant something, it named a few products, and you wondered why yours wasn't one of them. Fair question. The answer is less mystical than the "GEO agency" pitch decks make it sound.
What makes a product recommended by ChatGPT, Claude or Perplexity?
An assistant recommends what it can find, what others vouch for, and what it can quote. In practice that means its search index has your page, the wider web mentions your brand in the same breath as your category, and your page states plainly what you do and who it's for. Missing any one of the three, you rarely get named.
It helps to split what the model does into two jobs. Some of what it says comes from training data, which is frozen and only changes when a new model ships. The rest comes from a live search it runs while it's writing the reply to you. Google describes this for AI Overviews and AI Mode as "query fan-out": the system issues several related searches across subtopics, then builds one answer from what comes back.
You can't do much about the first job. You can move the second one this month, so that's where the rest of this post goes.
Does the "super intelligence" rename change how you get recommended?
No. The rename changed words, not systems. On 29 September 2026 the White House ordered federal agencies to say "Super Intelligence" and "SI" instead of "AI" in official documents. ChatGPT, Claude, Perplexity, Gemini and Copilot still crawl, retrieve and cite exactly as they did on 28 September.
What did change is the vocabulary your buyers type. Some now ask for "SI tools" or "the best SI assistant for X". If your pages only say "AI", you match half the phrasings. Add "SI" alongside "AI" once on the pages that define your category, and keep everything else as it is. Our guide to what SI means covers the rename in full, and SI vs AGI vs ASI sorts out which term means what.
One data point from our own site. In the 30 days to 5 October 2026, getclerion.com received 11 visits from AI assistants (8 from ChatGPT, 2 from Perplexity, 1 from Gemini), up from 1 the month before. Every single one landed on one of two pages: what SI means and what to know before buying a .si domain. Both answer questions people only started asking after the rename was announced in late September. Those are small numbers, and we won't pretend otherwise. But the pattern is clear enough: assistants send people to whoever answers the new question plainly and early.
Which crawlers have to reach your site for ChatGPT, Claude and Perplexity to cite it?
Each assistant runs a separate search crawler, and that bot is the one that decides whether you can be cited. The training crawlers are a different decision. You can block GPTBot or ClaudeBot from training on your content and still appear in answers, as long as the search bots below are allowed in robots.txt.
As of October 2026, from each company's own documentation:
| Bot | Company | What it does | Block it and… |
|---|---|---|---|
| OAI-SearchBot | OpenAI | Surfaces sites in ChatGPT search | You drop out of ChatGPT search answers |
| GPTBot | OpenAI | Collects data for model training | Training stops, search is unaffected |
| ChatGPT-User | OpenAI | Fetches a page when a user asks | Not used for search inclusion |
| PerplexityBot | Perplexity | Indexes sites for Perplexity results | You lose Perplexity citations |
| Perplexity-User | Perplexity | Fetches pages for a user's query | Generally ignores robots.txt |
| Claude-SearchBot | Anthropic | Indexes content for Claude's search | Visibility in Claude may drop |
| ClaudeBot | Anthropic | Collects data for model training | Training stops |
| Googlebot | Normal Search index, also feeds AI Overviews and AI Mode | You disappear from both |
Sources: OpenAI's crawler docs, Perplexity's bot docs, Anthropic's crawler help page, Google's AI features page.
Two things catch people out. A CDN or hosting firewall can quietly block these bots even when robots.txt lets them in, so check both places. And Google is blunt that there are "no additional requirements" for AI Overviews: if a page is indexed and eligible for a snippet, it's eligible. No special file, markup or schema unlocks it.
What gets a brand mentioned more: backlinks or mentions elsewhere?
Mentions win by a wide margin. Ahrefs studied 75,000 brands in December 2025 and found that branded web mentions correlated with ChatGPT visibility at 0.664, while backlinks sat around 0.2 and Domain Rating at 0.266. YouTube mentions were the strongest single factor, at 0.737 for ChatGPT.
Read that table as a founder, not an SEO. A backlink is a vote for a page. A mention is a sentence somewhere saying "for X, people use Y". That sentence is what a model picks up and pulls back later, because it ties your name to your category in words a person actually used.
Some ranked factors from that study:
- YouTube mentions: 0.737 (ChatGPT), 0.740 (AI Mode)
- Branded web mentions: 0.664 (ChatGPT), 0.709 (AI Mode)
- Branded anchor text: 0.511 (ChatGPT)
- Branded search volume: 0.352 (ChatGPT)
- Domain Rating: 0.266 (ChatGPT)
Correlation isn't cause, and big brands top every column anyway. Still, it tells you where a free weekend goes furthest. One honest comparison post on someone else's blog, a demo video, or a useful reply in the subreddit your buyers read will do more than chasing links. Shipping in public on X counts too, as long as your posts say the category word next to your product name.
How should a page be written so an assistant can quote it?
Answer first, in a block that makes sense on its own, then back it with numbers and named sources. Assistants lift passages, not pages. If your answer only makes sense after three paragraphs of setup, the model will quote someone else who put it up top.
This isn't just a hunch. The GEO paper from Princeton and collaborators, published at KDD 2024, tested rewrites on generative search answers. Adding quotations, statistics and cited sources lifted a page's visibility by up to 40%. Keyword stuffing, the old SEO reflex, barely moved the needle, and on one measure it did worse than leaving the page alone.
What that looks like on your own site:
- Name the category in the first sentence. "Acme is a meeting-notes tool for remote sales teams", not "Acme helps you reclaim your time".
- Use the buyer's question as the heading. "Does Acme work with Zoom?" beats "Integrations".
- Put your price and plan limits in text, not only inside an image or a toggle.
- Publish the comparison page yourself, and be fair to the competitor. A one-sided table reads as an ad, and models tend to skip ads.
- Date anything that changes. "As of October 2026" tells a model your page is current.
Say you run a small AI tool that turns sales calls into CRM notes. Your homepage says "Close more, type less." A model reading that learns nothing it can repeat. Change the first line to "Acme writes CRM notes from Zoom and Google Meet sales calls, from $19 a month", and every assistant now has a quotable sentence that contains your category, your integrations and your price.
Can a small startup own its category in ChatGPT?
Yes, more often than you'd expect, because most categories have no clear winner yet. Semrush and Kevin Indig tracked 1,094 US categories from January to June 2026. Only 15.2% had a clear brand owner in ChatGPT's answers. 53.7% had no consistent leader at all.
Two more findings from that study matter for a small team. Traditional SEO metrics predicted who owned a category only about half the time, so a weak domain isn't a death sentence. And once a brand owned a category, it kept first place in 90.4% of month-over-month checks. Early position tends to stick.
So the plan is simple, if not easy: pick the narrowest category you can honestly win ("CRM notes from sales calls", not "AI productivity"), write the best answer to the five questions buyers ask about it, and get mentioned wherever those buyers already talk. Being first to a clear answer beats being tenth to a vague one.
How do I know if ChatGPT recommends my product?
Ask it, then check your analytics for the visits it sends. Run the same five buyer questions through ChatGPT, Claude, Perplexity, Gemini and Copilot once a month, and log whether you're named. Then look for AI referrals in your traffic, landing page by landing page, so you know which answer is working.
The asking part is manual but free. Use a logged-out or fresh session so your own history doesn't skew the answer, and keep the wording identical each month.
The measuring part is harder than it should be. Default setups in many tools fold assistant visits into "direct" or a generic referral bucket, which we covered in does Google Analytics track AI traffic from ChatGPT. Some visits never carry a referrer at all, because certain apps strip it, so whatever number you see is a floor.
Microsoft is ahead of the others on citation data. Since February 2026, Bing Webmaster Tools has shown an AI Performance report in public preview, with how often your content appears in AI-generated answers. If you haven't verified your site there, do it: it's free, and the numbers come straight from the platform rather than from a third-party estimate.
If you'd rather not set this up by hand, Clerion reports ChatGPT, Perplexity, Claude, Gemini and Copilot as their own channel, with the pages each one sent people to. You can also run any page through the free AI crawler checker to see what a model reads on it.
Do this tonight
- Open your robots.txt and your CDN's bot settings. Make sure OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot are allowed.
- Rewrite the first sentence of your homepage so it names your category, your buyer and your price.
- Write down the five questions a buyer asks before choosing a product like yours. Ask all five in ChatGPT, Claude and Perplexity, and note who gets named.
- Turn the question you lose most clearly into a page, with the answer in the first 60 words.
- Post one useful, specific answer where your buyers ask that question, and name your product in it once.
Related reading
- What is SI (super intelligence)?: what the rename means and what it doesn't.
- SI vs AGI vs ASI: what's the difference?: the three terms, side by side.
- Does Google Analytics track AI traffic from ChatGPT?: why assistant visits hide in "direct".
- How to track ChatGPT traffic to your website: the setup, step by step.
Frequently asked questions
Does blocking GPTBot stop ChatGPT from recommending my product?
No. GPTBot only collects training data. ChatGPT search uses OAI-SearchBot, so you can block GPTBot and still be cited, as long as OAI-SearchBot is allowed in robots.txt and your firewall.
Do I need an llms.txt file to get recommended by ChatGPT?
None of OpenAI, Perplexity, Anthropic or Google list llms.txt as a requirement for search citations. Google says AI Overviews and AI Mode need no special files: an indexed page that can show a snippet is eligible.
How long does it take for ChatGPT to start recommending a new product?
Answers built from live search can pick up a new page once the assistant's search crawler has indexed it. Answers that come from training data only change when a new model ships, so expect the two to move at very different speeds.
Is SI the same as AI for getting recommended?
Yes. The US executive order of 29 September 2026 changed the term federal agencies use, not the technology. ChatGPT, Claude, Perplexity, Gemini and Copilot crawl and cite exactly as before. Mention both terms on category pages.
Do backlinks still matter for AI recommendations?
Less than you might think. In Ahrefs' December 2025 study of 75,000 brands, backlinks correlated with ChatGPT visibility at around 0.2, while branded web mentions scored 0.664 and YouTube mentions 0.737.
How can I see if Copilot cites my website?
Verify your site in Bing Webmaster Tools. Since February 2026 it has offered an AI Performance report in public preview that shows how often your content appears in AI-generated answers from Microsoft Copilot.
Sources
- Fact Sheet: President Donald J. Trump Inaugurates The Era of Super Intelligence (White House)
- Overview of OpenAI crawlers
- Perplexity crawlers
- Does Anthropic crawl data from the web, and how can site owners block the crawler?
- AI features and your website (Google Search Central)
- Top Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews (Ahrefs)
- AI visibility is a topic-level game: A study of 50,000 brands in ChatGPT (Semrush)
- GEO: Generative Engine Optimization (KDD 2024)
- Bing Introduces AI Performance Reporting in Webmaster Tools
- What is SI? Super Intelligence explained (Clerion)
- Things to know before you buy a .si domain (Clerion)