You opened ChatGPT, asked "what's the best [your category] for [your situation]" — and watched it recommend three competitors while skipping you entirely. If you're reading this, you've probably done that test more than once. The frustration is real and the lost pipeline is measurable. Most people Google "why isn't my website showing up in ChatGPT" and find vague articles telling them to "improve their SEO." That's not actionable. This is.
There's almost always a specific reason behind it, and it's almost never just one thing. We've run Reffed on hundreds of customer sites over the past six months, and the same eight problems come up over and over. Work through this list in order. If you fix three of them, you'll likely see your citation share improve within 30–60 days.
1. AI crawlers are blocked at robots.txt
This is the single most common — and most quietly destructive — issue. Many sites still have robots.txt files that explicitly block GPTBot, ClaudeBot, and PerplexityBot. Sometimes this was done deliberately during the 2024 "AI training" panic when developers added blanket bans as a privacy gesture. Sometimes it's a default in a Wix or Shopify theme. Either way, if those bots can't read your site, you literally cannot be cited.
How to check: visit yourdomain.com/robots.txt in any browser. Look for lines like User-agent: GPTBot followed by Disallow: /. If you see them, you're blocked.
How to fix: update your robots.txt to explicitly allow the major AI crawlers. At minimum, allow GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-Web, PerplexityBot, and Google-Extended. We have a detailed walkthrough in our robots.txt for AI crawlers guide.
2. Your site doesn't rank in Bing
Here's the secret most "GEO experts" don't tell you: when ChatGPT browses the web, it uses Bing's search index, not Google's. If your site ranks #1 on Google but doesn't appear anywhere in Bing's index, ChatGPT will simply never see it during live retrieval. This single fact disqualifies more sites from ChatGPT citations than any other technical problem.
How to check: open Bing and search for site:yourdomain.com. If you see fewer than a dozen pages indexed, Bing barely knows you exist. Also register for Bing Webmaster Tools and check your indexed-pages count there.
How to fix: submit your sitemap directly to Bing Webmaster Tools. It's free, takes 15 minutes, and unlocks citation potential most sites leave on the table. Bing indexes new sites slower than Google, but once you're in, you're in for ChatGPT browsing mode too.
3. Your content is hidden behind JavaScript
If your site is built as a single-page app — older React or Vue setups without server-side rendering — your actual content may only appear after JavaScript runs. Some AI crawlers execute JS, many don't. The safe assumption: if it's not in the raw HTML response, it doesn't count for citation purposes.
How to check: right-click your homepage and choose "View Page Source." If the body looks like <div id="root"></div> and nothing else, your content is JS-rendered and probably invisible to most AI crawlers.
How to fix: migrate to server-side rendering or static generation. Next.js, Astro, Nuxt, SvelteKit, and Hugo all handle this natively. If migrating isn't feasible immediately, pre-render at least your highest-value commercial pages.
4. No structured data (schema markup)
JSON-LD schema is how you explicitly tell AI crawlers what your site is, what your business does, and what offerings you have. Without it, engines have to infer everything from prose, and they often get it wrong or skip you entirely. Sites with proper schema get cited noticeably more than sites without — the gap is one of the most reliable signals in the entire GEO space.
How to check: paste your URL into Google's Rich Results Test. You should see at minimum an Organization (or LocalBusiness) entry plus a WebSite entry. If you see nothing, you have zero schema and you're invisible to the systems that score citation candidates.
How to fix: add at minimum: Organization or LocalBusiness, WebSite, and FAQPage schemas to your homepage and main marketing pages. We cover the specific JSON-LD blocks to use in our schema markup guide.
5. Your content is too thin to extract from
ChatGPT, Claude, and Perplexity all prefer to cite pages with substantive content — typically 800+ words on the topic. A homepage with three taglines and a button isn't extractable. AI engines need actual passages they can pull and cite. Thin content fails on both information richness and "answer capsule" availability.
How to check: count the words on your homepage. If it's under 400 words of real prose, it's too thin.
How to fix: add a substantive section to your homepage that answers the most common question your buyers ask. Format it as a clear question-answer pair (with proper FAQ schema), 250–400 characters per answer. AI engines extract these almost preferentially.
6. Your business has no authority signals
Technical readiness gets you eligible for citation. Authority signals are what makes engines actually pick you over a competitor. The key signals: backlinks from authoritative domains, brand mentions across the web, Wikipedia presence (when notability warrants), Crunchbase or LinkedIn company entries, and press coverage. Without these, even a technically perfect site will struggle to compete with established brands.
How to check: search for your brand name in Google. How many places mention you? A site mentioned in 5 places across the web is in a different league from a site mentioned in 500.
How to fix: this is the slow work. Pitch one industry publication per month. Get listed in 5 directories relevant to your category. Submit your founder story to local press. AI engines weight authority heavily, and authority compounds over months, not days.
7. You're asking ChatGPT for queries it doesn't browse for
ChatGPT doesn't always use browsing mode. For questions it can answer from training data — "what is X" or "how does Y work" — it often responds without retrieving live web results. Browsing is triggered for recent events, specific data, current comparisons, or when you explicitly enable web search.
How to check: open ChatGPT, enable web search, and re-run your query. Did you appear when browsing was on? If yes, you're optimized for live retrieval but invisible to training-data answers.
How to fix: for training-data citation, you need to be visible enough during the training window that your brand sticks in the model's weights. This is a multi-year authority play. For browsing-mode citation, optimize for current Bing rankings on the specific queries your buyers ask.
8. Your content isn't fresh
AI engines prefer pages that have been updated recently. A blog post from 2022 about "current trends" hurts more than helps. Engines look at modified dates, Article schema, and content freshness to decide whether a page is current enough to cite for time-sensitive queries.
How to check: when did you last update your main marketing pages? If the answer is "I built them and never touched them again," that's a problem.
How to fix: update your core pages every six months. Add new data, new examples, and refresh the modified date in your Article schema. This signals to engines that you're actively maintained.
Run the audit, then start fixing
If you want a faster diagnostic than working through this list manually, run the free Reffed audit. It checks all eight of these (plus four more) in 60 seconds and gives you a ranked list of fixes by impact. No signup, no credit card.
The honest truth: most of these problems take 1–3 hours to fix per item. The whole list is doable in a weekend for a motivated owner. Once you're done, the gains aren't instant — AI engines re-index on their own schedule — but you'll typically see citation share improve in 30–60 days.
If you've fixed everything and still aren't cited after 60 days, your bottleneck is probably authority signals, not technical readiness. That's a different conversation, and we cover it in our full citation playbook.