← Part of the full guide: Why AI Doesn’t Recommend Your Brand
TL;DR: ChatGPT reaches your site through three separate bots, and each one can be blocked independently. The most common reason a website never shows up in ChatGPT is not weak content — it is that the raw HTML the bot receives is empty, because the page builds itself with JavaScript and no AI crawler runs JavaScript. You can test both of these yourself in about ten minutes with two terminal commands. Or get a free AI visibility scan and skip the manual work.
What “can ChatGPT see my website” actually means
It means asking whether OpenAI’s crawlers are allowed to request your pages, and whether the HTML they get back already contains your content. Both have to be true. If either fails, ChatGPT cannot cite you no matter how good the writing is.
This is a different question from whether AI knows your brand. A model can describe your company from training data it absorbed years ago and still be unable to open your website today. If you want to test brand recall rather than site access, that is covered in how to check whether AI knows your brand. This page is about the plumbing underneath it.
Why your website is not showing up on ChatGPT — 5 causes
1. Your robots.txt blocks OpenAI’s crawlers
OpenAI honours robots.txt for its automated bots. Plenty of sites added blanket AI blocks during the 2023–2024 scraping panic and never revisited them. Cloudflare has also blocked AI crawlers by default for new domains since 2025, which means the block can exist without anyone on your team having written it. A single Disallow line is enough to make you permanently uncitable.
2. Your content only exists after JavaScript runs
This is the big one, and it is invisible in a browser. As of 2026 no major AI crawler executes JavaScript — not GPTBot, not OAI-SearchBot, not ClaudeBot, not PerplexityBot. One analysis of over 500 million GPTBot fetches found no evidence of JavaScript execution at all; GPTBot downloads .js files in roughly 11% of requests and never runs them. Googlebot renders JavaScript with headless Chrome, so a client-side-rendered site can rank perfectly well on Google while returning a near-empty document to every AI engine. That gap is why “we rank on page one” and “ChatGPT has never heard of us” coexist so often.
3. Your server or CDN blocks the bots before robots.txt matters
Bot-protection rules, rate limits, and WAF policies reject requests at the edge. The crawler receives a 403 and moves on. Your robots.txt could be fully permissive and it would change nothing, because the bot never got far enough to read it.
4. You are not in Bing’s index
ChatGPT search leans on Bing’s index for retrieval. If Bing has not indexed a page, ChatGPT search has a much harder time surfacing it, even when your Google presence is healthy. Bing indexation is one of the few ChatGPT ranking factors you can verify in a free dashboard.
5. Nothing on the page is quotable
Access is necessary but not sufficient. A page that never states a claim in a single liftable sentence gives the model nothing to extract. Research on AI citation behaviour keeps landing on the same pattern: adding statistics raises AI visibility by around 22%, adding quotations by around 37%, and roughly 55% of citations come from the top third of a page. Burying the answer costs you the citation.
GPTBot vs OAI-SearchBot vs ChatGPT-User
These three names appear in your server logs and they do different jobs. Blocking the wrong one costs you citations for no benefit.
| GPTBot | OAI-SearchBot | ChatGPT-User | |
|---|---|---|---|
| Purpose | Collects content for training future models | Builds the search index ChatGPT cites from | Fetches a page live when a user asks about it |
| Obeys robots.txt | Yes | Yes | Not always — the visit is user-initiated |
| Blocking it removes you from citations | No, but you lose long-term model familiarity | Yes — this is the one that matters most | Yes, for live look-ups |
| Runs JavaScript | No | No | No |
| Reasonable to block | Only if you object to training use | Almost never | Almost never |
How to check whether ChatGPT can see your site
Step 1 — Read your robots.txt (2 minutes)
- Open
yourdomain.com/robots.txtin a browser. - Search for
GPTBot,OAI-SearchBot,ChatGPT-User, and anyUser-agent: *block with aDisallow: /under it. - If the file does not exist at all, that is fine — no file means no restrictions.
Step 2 — Fetch your page the way a bot does (5 minutes)
- Run:
curl -A "OAI-SearchBot" https://yourdomain.com/ > test.html - Open
test.htmlin a text editor and search for a sentence you know is on the live page. - If the sentence is missing, or the file is a few kilobytes of empty
<div>tags and script tags, your content is JavaScript-dependent and AI engines see nothing. - A 403 or a challenge page instead of HTML means your CDN or firewall is blocking the bot.
Step 3 — Confirm Bing has you (3 minutes)
- Search
site:yourdomain.comon Bing and count the results. - Register the site in Bing Webmaster Tools and check the index coverage report for pages that were crawled but excluded.
Caution: “View source” in Chrome is not the same test. Some browsers show you the rendered DOM rather than the raw response, which makes a JavaScript-dependent page look fine when it is not. Use curl with the bot’s user-agent string, or disable JavaScript entirely before viewing the page.
What to fix first
1. Fix access before you fix content
Rewriting pages while OAI-SearchBot receives a 403 changes nothing. Unblock first, verify with curl, then work on the writing.
2. Server-side rendering beats every other technical fix
If step 2 came back empty, nothing else on your GEO list matters yet. Server-side rendering, static generation, or prerendering for bot user-agents all solve it. Which one is realistic depends on your platform, and some hosted site builders simply do not expose the option.
3. Give the crawler something worth lifting
Once the HTML arrives intact, put a direct 40–60 word answer under each heading, name your sources in the text, and keep the strongest claim in the first third of the page. That is where most citations are drawn from.
4. Recheck after every deploy
A theme update, a new WAF rule, or a framework upgrade can silently reintroduce the block. Treat the curl test as a release check, not a one-off audit.
CiteLogics — find out what AI crawlers actually receive from your site
The AI Visibility Audit fetches your site with real AI crawler user-agents, shows you the exact HTML each engine gets back, and scores crawlability alongside four other dimensions of the AIVS Score.
- AI Visibility Audit — full AIVS Score report + 30-minute call. THB 2,000, delivered in 24 hours.
- Big Win (Audit + WebFix) — audit plus full implementation, including server-side rendering fixes. THB 20,000.
Get your free AI visibility scan, or contact sales@citelogics.com.
Frequently asked questions
Can ChatGPT read my website in real time?
Yes, through the ChatGPT-User agent, which fetches a page live when a user asks about it. It still cannot run JavaScript, so it only sees whatever is in the raw HTML response.
Does ChatGPT render JavaScript?
No. As of 2026 none of OpenAI’s crawlers execute JavaScript, and neither do Anthropic’s or Perplexity’s. Only Googlebot renders JavaScript, which is why a site can rank on Google and stay invisible to ChatGPT.
Should I block GPTBot?
Only if you object to your content being used for model training. Blocking GPTBot does not stop ChatGPT from citing you, but blocking OAI-SearchBot does — those are separate decisions and should be made separately.
How long before ChatGPT picks up a fix?
Recrawl timing varies by site and is not published by OpenAI. Sites with regular publishing cadence tend to be revisited sooner. Plan on weeks rather than days, and verify with server logs rather than by asking ChatGPT repeatedly.
My site is on Wix or Shopify — can I still fix this?
Partly. Hosted builders serve server-rendered HTML for most page types, so the JavaScript problem is usually smaller than on a custom SPA. What they limit is how much structured data and how many raw HTML controls you get. An audit tells you the realistic ceiling for your platform before you spend money on content.
Written by
Nipun Kasevayuth
Founder & CTO, CiteLogics
An engineer and AI enthusiast who reverse-engineers how AI models decide which brands to cite. He proved the method on his own websites first — then delivered the same results for brands like Coldtubb. He and the CiteLogics team have scanned 1,000+ websites for AI visibility.
Connect on LinkedIn →
