Why ChatGPT Is Not Citing My Website (8 Reasons + Fixes)

How ChatGPT picks what to cite

ChatGPT blends two information sources to assemble each answer: training-data knowledge and live retrieval through Bing when search is enabled. For your brand to appear, at least one of those sources needs to surface authoritative content about you. If neither does, ChatGPT either omits you entirely or, worse, hallucinates a wrong description.

Eight root causes account for roughly 90 percent of "why isn’t ChatGPT citing me" cases we see in audits. Most sites have two or three of them simultaneously.

1. GPTBot is blocked in robots.txt

The single most common cause. Many sites copied a robots.txt template from 2022 or earlier that blocks unknown crawlers by default, and GPTBot got caught in the blanket. Check your /robots.txt right now and explicitly allow:

User-agent: GPTBot
Allow: /

User-agent: ChatGPT-User
Allow: /

User-agent: OAI-SearchBot
Allow: /

Fix: add the explicit Allow rules. Most sites see indexing pick up within 7 to 14 days. The audit checks every AI crawler and shows you which to allow.

2. No FAQPage or HowTo schema on landing pages

FAQ content is the single highest-value real estate for AI citations. Sites that ship plain-prose FAQs without FAQPage JSON-LD lose the citation slot to competitors that wrap their FAQs in schema, even when the answers are identical.

Fix: ship FAQPage JSON-LD on the top three landing pages with eight to fifteen entries each. Use HowTo for procedural content. Validate with Schema.org validator before deploy.

3. The lede buries the answer

ChatGPT lifts the lede when it cites you. If your first paragraph is marketing wind-up, the model lifts a competitor’s lede instead. We audit pages where the actual buyer-question answer appears in paragraph six.

Fix: rewrite the first 60 words of every priority page to answer the buyer question directly. Three sentences, no qualifiers, no preamble.

4. Inconsistent entity data across the web

ChatGPT cross-references your business name, description, address, and founding date across the homepage, Google Business Profile, LinkedIn, Crunchbase, and Wikidata. Inconsistencies reduce confidence. We see this in roughly 40 percent of audits: three different business names, two different categories, four different founding years.

Fix: pick one canonical version and ship it everywhere. The audit produces a side-by-side mismatch report so you know exactly which sources to align.

5. The page does not rank in Bing

ChatGPT search uses Bing as the live retrieval layer. If your page does not rank in Bing for the buyer query, ChatGPT does not see it. Many sites optimize for Google and ignore Bing entirely; Bing has different signal weights and often ranks differently.

Fix: add Bing Webmaster Tools, submit your sitemap, fix Bing-specific indexing issues. Most pages that rank in Google top 20 will rank in Bing top 30 within 30 days of basic Bing hygiene.

6. No Person schema or AuthorCard on bylined content

ChatGPT uses E-E-A-T signals to pick between similarly-ranked pages. A bylined page with Person schema, a real bio, and sameAs links to LinkedIn outranks an anonymous page on the same topic. Many sites publish quality content without any author markup at all.

Fix: add an AuthorCard component with Person JSON-LD on every blog post and pillar page. Include jobTitle, knowsAbout, sameAs to LinkedIn and a personal site.

7. Stale dateModified

ChatGPT favors fresh authoritative content for time-sensitive queries. Pages with dateModified from 2022 or 2023 lose citation share to competitor pages updated in the last 90 days, even when the underlying advice is still correct.

Fix: update dateModified when you ship real edits. Do not fake-update dates - that fails when the model compares your archived version. A real refresh of a 2022 page restores citation eligibility.

8. The category is dominated by older training-data signals

Some categories are locked in at the training-data layer by older competitors with deep mention histories. New entrants struggle to crack the citation list even with perfect on-site optimization, until the next ChatGPT model cycle picks up the new entity signals.

Fix: shift effort to live-retrieval engines (Perplexity, AI Overviews) where training-data lock-in does not apply, while feeding the training-data layer through authoritative third-party mentions. Plan for 6 to 12 months on the ChatGPT side; expect 4 to 8 weeks on Perplexity.

How to prioritize the fixes

Most sites have three to five of these issues simultaneously. Fix in this order for fastest impact:

• Day 1: check robots.txt, add explicit Allow rules for GPTBot and friends.
• Day 2-3: audit entity consistency across the web and align to one canonical version.
• Day 4-7: ship FAQPage and HowTo schema on top three pages with eight to fifteen entries each.
• Day 8-14: rewrite ledes definition-first on top 10 pages.
• Day 15-21: add Person schema and AuthorCard to bylined content.
• Day 22-28: refresh dateModified on stale priority pages with real edits.
• Day 29-30: re-run citation tracker to measure deltas.