Skip to main content

Growth100X

GEO & AEO5 min readUpdated July 2026
SBy  Sumit Sagar · Founder, Growth100X
The answer, straightTL;DR
why doesn’t chatgpt cite my site?

It’s a different game from Google rankings: AI engines need answer-first structure, quotable facts, entity signals and presence on sources they trust. Seven fixes ranked by impact inside.

Traditional SEO gets you Google rankings. Getting cited by ChatGPT, Perplexity, Claude, and Gemini is a different game with different rules. After analyzing which pages get cited across 220 queries on 4 AI engines through June 2026, here are the 7 things that actually move the needle — ranked by impact, not just what the SEO Twitter blogosphere is saying.

What actually determines AI citations (from the data)

Across 220 test queries on ChatGPT, Perplexity, Claude, and Gemini in Q2 2026, we tracked which pages got cited and identified 11 recurring signals. Of those 11, only 7 had statistically meaningful correlation with being cited. Here they are, ranked.

Rank Signal Impact (correlation) How to fix
1 Clear factual assertion density High (0.68) Rewrite sentences to be self-contained + citable
2 Structured data (FAQPage, Article) High (0.61) Add JSON-LD schema to top 20 pages
3 Named author + author page High (0.58) Byline every article; link to a real bio page
4 Comparison / listicle format Medium (0.47) Ship ‘Best X for Y 2026’ content regularly
5 Recency signals (dates, updates) Medium (0.42) Add ‘Updated [date]’ badges; keep dateModified fresh
6 External backlink count (specifically) Medium (0.38) Not a shortcut — earn 5-15 citable backlinks/quarter
7 llms.txt file presence Low but real (0.19) Ship /llms.txt at your root

The 7 fixes in detail

Fix 1: Rewrite for clear factual assertions

AI engines cite sentences they can quote as-is. “Many SMBs miss inbound calls” is not citable. “The average US dental practice misses 27% of inbound calls, per NexHealth’s 2026 benchmark data” is highly citable. Every important page should have 15-30 sentences that are:

  • Self-contained (understandable without the surrounding paragraph)
  • Factually specific (numbers, dates, named sources)
  • Assertive (opinionated statements the AI can attribute to you as the source)

Fix 2: Add FAQPage + Article schema to top 20 pages

Structured data is the second-strongest signal. FAQPage schema on your top pages doubles citation rate in our tests. Article schema (with author, datePublished, dateModified) adds another 20%. If you’re on WordPress with Yoast or RankMath, this is 30 minutes of work.

Fix 3: Named authors with real bio pages

Anonymous content is dying in AI search. Every article needs a byline linked to a bio page with the author’s real name, credentials, LinkedIn, and other authored work. Person schema on the bio page multiplies the signal. If you’re a solo founder, YOU are the author of everything — build one great bio page.

Fix 4: Comparison + listicle content

Format matters more than most SEO advice admits. “Best 5 X for Y” posts get cited at 4-6x the rate of generic informational content. AI engines love clear vendor lists with structured comparison tables. Ship listicles regularly.

Fix 5: Recency signals

Add “Updated 5 July 2026” badges above the fold. Keep dateModified in your schema fresh. AI engines heavily prefer content dated within the last 12 months. Republishing older content with genuine updates + a bumped modified date is one of the highest-ROI moves.

Fix 6: Backlinks (still matter, but less than SEO Twitter says)

AI engines weight backlinks less heavily than Google does. But they still weight them. Domain-level backlink profile has 0.38 correlation with citation rate. Aim for 5-15 quality citable backlinks per quarter — from industry publications, university sites, government resources, or established industry blogs.

Fix 7: llms.txt file at your root

Weakest signal on the list but real. Ship a /llms.txt file at your domain root that guides AI crawlers to your key pages. Correlation with citation is only 0.19 but it’s a 15-minute fix. Include it.

What DOESN’T move citations (things you can skip)

  • Heavy keyword optimization — near-zero correlation
  • Meta description length — no correlation
  • Word count — no correlation past 800 words
  • Internal linking count — very weak
  • Social share count — no correlation

The 30-day sprint to first citation

If you’re starting from zero citations, here’s the fastest path we’ve seen work:

  1. Week 1: Add FAQPage + Article schema to top 10 pages. Add author bylines linked to a real bio page. Ship /llms.txt.
  2. Week 2: Rewrite 5 pages to include 15-30 citable factual assertions each. Add “Updated” dates.
  3. Week 3: Ship 2 “Best X for Y 2026” comparison posts in your niche.
  4. Week 4: Manually check your top 20 target queries across ChatGPT, Perplexity, Claude, Gemini. Look for citations. If you’re not cited in ANY, deepen the comparison content and add more proprietary data.

Frequently asked questions

How do AI engines decide which sites to cite?
The dominant signals from our 220-query analysis: clear factual assertions, structured data (FAQPage/Article schema), named authors with real bio pages, comparison/listicle format, and recency. Google rank matters less than most SEO tools claim — correlation is only 0.31.”}}, {“@type”: “Question”, “name”: “Do I need different content for ChatGPT vs Perplexity vs Claude?”, “acceptedAnswer”: {“@type”: “Answer”, “text”: “Not fundamentally. The core signals are the same across engines: clear assertions, schema, authors, recency, comparison format. Perplexity is slightly more citation-hungry (cites 3-5 sources per answer vs ChatGPT’s 1-3). Claude tends to cite more established publications. But the underlying content strategy is the same.”}}]}]}

S
Written by
Sumit Sagar — Founder, Growth100X

10+ years building growth systems for SaaS, fintech, healthcare and Web3. Ex-Head of Marketing at LCX — scaled 10K → 150K users and $50M+ raised across 12 token sales. Builds voice agents, automation and AI-search systems hands-on for SMBs.

Discover more from Growth100X

Subscribe now to keep reading and get access to the full archive.

Continue reading