// Network Troubleshooting · 7 Min Read

Why ChatGPT Search Isn't Citing Your Website

A systematic diagnostic runbook: how firewalls, robots directives, and client-side JavaScript walls silently turn your website into a blank canvas for OpenAI.

Published: 2026-08-26 Author: CiteDo Research Category: Search Indexing & Crawlability

Hundreds of website owners reach out to us with the same frustrating scenario: "Our brand ranks #1 on Google for our primary industry keywords, but when someone asks ChatGPT for product recommendations in our category, our competitors are cited and we don't exist."

The reason is almost always technical: **ChatGPT Search does not use Google's search index.** It conducts its own live HTTP crawls using OpenAI-owned user-agents. If your server returns an HTTP 403, an empty HTML shell, or an explicit Disallow header, ChatGPT's reasoning engine simply moves on to the next competitor.

Test Your Site for OpenAI Access (30s)

Our live probe tests your domain against robots.txt directives and live GPTBot connection headers.

Run GPTBot Test →

The 5 Technical Culprits (And How to Fix Them)

1. robots.txt Is Confusing GPTBot with OAI-SearchBot

Common Mistake

Many sites added User-agent: GPTBot / Disallow: / in 2024 to prevent model training. However, when OpenAI launched ChatGPT Search, they introduced OAI-SearchBot specifically for search attribution. If you used a generic wildcard rule, you accidentally banned search traffic too.

The Fix: Explicitly Allow Search Crawlers

# Allow ChatGPT Search & Perplexity to cite your site
User-agent: OAI-SearchBot
Allow: /

User-agent: ChatGPT-User
Allow: /

# Optionally block model pre-training data scraping
User-agent: GPTBot
Disallow: /

2. Cloudflare & CDN WAF Interception

42% of Failures

Features like Cloudflare's "Bot Fight Mode" or AWS WAF automated protection block non-browser user agents by default. When OAI-SearchBot queries your server, it receives an HTTP 403 Forbidden or a JavaScript challenge, causing ChatGPT to log the site as unreachable.

The Fix: Create a WAF Custom Rule

In Cloudflare Dashboard: Security > WAF > Custom Rules > Create Rule:
(cf.client.bot) or (http.user_agent contains "OAI-SearchBot") -> Action: Skip (All Security Features)

3. The Client-Side JavaScript Wall

Unlike Googlebot, which runs an expensive headless rendering queue (sometimes delayed by days), real-time AI search bots fetch HTML over standard HTTP sockets. If your React or Vue application only serves <div id="root"></div> without Server-Side Rendering (SSR) or Static Site Generation (SSG), OpenAI sees 0 words of content.

The Fix:

Implement Static Site Generation (SSG) or Server-Side Rendering (SSR) for core product pages and articles, or deploy an /llms.txt file.

4. Low Information Density & Missing Schema Entity

When ChatGPT parses a site, it looks for clean key-value facts: pricing, features, specifications, and supported platforms. Fluffy marketing copy without structured schema.org markup makes it difficult for vector search to index your brand as a concrete solution.

Get the Full Picture on Your AI Search Presence

Crawlability is step one. Discover where your brand actually ranks across 12 AI search engines when real buyers ask questions.