LLM Citation vs Organic Visibility

🔬

What This Research Looks At

This research cross-references GA4 LLM referral traffic (sessions from ChatGPT, Perplexity, Gemini, Copilot, Claude and Kagi) against Google Search Console coverage data to understand the relationship between organic visibility and LLM citation. The headline finding across both clients: Google indexation is the strongest predictor of LLM citation — indexed pages are 5–7x more likely to receive LLM traffic than non-indexed ones. Getting content indexed remains the foundation. The secondary finding is that a small number of specific, factual pages do get LLM citations without Google indexation — this is an interesting find since most SEO experts today claim that LLMs can't cite what is not indexed.

Double Win

Indexed by Google and cited by LLMs — the gold standard

LLM Only

Cited by LLMs but Google refuses to index — hidden opportunity

GEO Gap

Indexed by Google but zero LLM citations — visible but AI-invisible

💡

Cross-Client Takeaways

Client A

Double Wins15

LLM-Only (not indexed)2

LLM citations as % of indexed9%

Total LLM sessions233

Top LLM sourceChatGPT (80%)

Client B (ContentForge)

Double Wins34

LLM-Only (not indexed)3

LLM citations as % of indexed52%

Total LLM sessions122

Top LLM sourceChatGPT (52%)

Indexation Is the Strongest Predictor of LLM Citation

Across both clients, indexed pages are 5–7x more likely to receive LLM citations than non-indexed ones. Client B: 52% of indexed pages cited vs 7.7% of non-indexed. Client A: 9.4% vs 1.8%. Getting Google to index content remains the single most important step for LLM visibility — GEO does not bypass SEO fundamentals, it builds on them.

Structured, Factual Content Wins Both Channels

The same content types dominate both Google rankings and LLM citations: cost comparisons, regulatory guides, "X vs Y" formats, and data-led pages. Both algorithms reward content that directly and specifically answers a question. ContentForge's structured output achieved a 52% Double Win rate from indexed pages in 3 months versus Client A's 9% with manually produced content.

LLMs Occasionally Override Google's Rankings

A small but notable exception: Client B's /services/drones-for-surveying-and-mapping sits at avg position 29.7 on Google yet attracted 8 LLM sessions. Across both clients, 5 pages received LLM citations despite not being indexed. These are almost always highly specific, factual posts — the edge case worth fixing, not the rule to build strategy around.

AI Is Already a Commercial Referral Channel

Both clients see LLMs referring traffic to commercial and brand pages — /apply-now and /founder-profile for Client A; /contact-us, /who-we-are, and /services for Client B. AI is acting as a bottom-of-funnel channel, not just a research tool. These sessions need conversion tracking separate from organic.

📋

Client Analysis 1 of 2

Client A

MBA & Masters Admissions Consultancy · Data: April 2026

Site Indexation Snapshot

Total Pages

273

Known to Google

Indexed

160

59% of site

Crawled, Not Indexed

48

Google visited, rejected

Discovered, Not Indexed

65

Never crawled

Indexation Context

41% of the site (113 pages) is invisible to Google. The bulk of non-indexed pages are question-format posts published in late 2025 — Google is treating them as thin or duplicate content. This makes the LLM citation picture below more striking: AI models are citing content that Google has explicitly declined to rank.

LLM Traffic Snapshot

Total LLM Sessions

233

Dec 2025 – Feb 2026

Pages Cited

48

Unique URLs receiving LLM traffic

Top Source

ChatGPT

186 sessions (80%)

Other Sources

4

Perplexity, Gemini, Copilot, Kagi

Citation vs Indexation Classification

✅ Double Win

15

Pages indexed by Google and cited by LLMs — capturing both channels

⚡ LLM Only — Google Rejected

2

LLMs cite these pages but Google refuses to index them

⚠️ GEO Gap

145

Indexed by Google but receiving zero LLM citations

🔕 Dark Pages

111

Not indexed by Google, no LLM citations either

The 2 Pages LLMs Cite That Google Won't Index

Page URL	GSC Status	Last Crawled	Google Clicks	LLM Sessions
/post/insead-mba-interview-preparation-guide	Crawled – Not Indexed	Nov 2025	0	~3
/post/how-to-discuss-setbacks-and-challenges-in-your-insead-mba-interview	Crawled – Not Indexed	Nov 2025	0	~2

What This Means

Both are INSEAD interview posts — question-format content Google is rejecting as thin. Despite that, ChatGPT has trained on them and is sending real sessions. These are the exception that proves the rule: specific, factual content can get LLM pickup without Google indexation, but the far more valuable opportunity is fixing the indexation so these pages capture both channels rather than just one.

15 Double Win Pages — Indexed + LLM Cited

Page	GSC Clicks	Impressions	LLM Source(s)
/post/insead-mba-fees-cost-breakdown	52	54,845	ChatGPT Perplexity
/post/insead-mba-deadlines	36	14,408	ChatGPT
/post/which-european-mba-is-best-for-consulting-careers-insead-lbs-or-iese	54	10,126	ChatGPT Perplexity
/post/insead-mba-gmat-gre-scores	49	7,628	ChatGPT
/post/finding-value-a-guide-to-europe-s-most-affordable-top-mba-programmes	39	17,227	ChatGPT
/post/insead-mba-cv-resume-formatting-tips-templates	47	2,591	ChatGPT
/post/insead-vs-lbs-vs-oxford-mba	42	5,276	ChatGPT Perplexity
/post/best-mba-programs-in-europe	23	10,095	ChatGPT
/post/insead-scholarships-application-guide	7	642	ChatGPT
/post/insead-mba-essays	3	97	ChatGPT
/post/insead-vs-lbs	7	1,510	ChatGPT
/founder-profile Founder profile page	—	—	ChatGPT
/apply-now Commercial page	—	—	ChatGPT
/emba/cambridge	—	—	ChatGPT
/post/will-lbs-and-cambridge-judge-accept-my-gre-score-instead-of-gmat	15	1,602	Perplexity

Pattern in the Double Wins

The 15 Double Win pages share three characteristics: they are data-led (fees, GMAT scores, deadlines), comparative ("X vs Y" or "best for Z"), or structured factual guides. These formats answer AI queries directly and concisely — exactly what both Google and LLMs reward. The founder profile and /apply-now pages being cited by ChatGPT is notable: AI is acting as a sales referral channel for branded search.

📋

Client Analysis 2 of 2

Client B

Drone Inspection & Services · ContentForge (Jan–Mar 2026) · Data: April 2026

Context

Client B is a GEOforge ContentForge client — an automated content production programme publishing AI-assisted blog posts targeting drone inspection, insurance, energy, and construction use cases. This report covers data from January to end of March 2026, showing what structured automated content delivery produced across both Google and LLM channels during that period.

Site Indexation Snapshot

Total Pages

104

Known to Google

Indexed

65

63% of site

Crawled, Not Indexed

21

Google visited, rejected

Discovered, Not Indexed

18

Never crawled

Indexation Context

37% of the site (39 pages) is not indexed — a similar ratio to Client A. The crawled-not-indexed pages are predominantly newer ContentForge posts from Feb–March 2026, suggesting Google is applying quality filtering to the higher-volume automated content. The 63% index rate on a programmatic content site is a reasonable baseline and reflects well-structured, topic-specific content.

LLM Traffic Snapshot (Jan–Mar 2026)

Total LLM Sessions

122

Across 5 AI sources

Pages Cited

37

Unique URLs receiving LLM traffic

Top Source

ChatGPT

64 sessions (52%)

Notable Source

Claude

27 sessions (22%) — 2nd largest

Claude.ai as a Referral Source

Claude.ai sent 27 sessions — making it the second-largest LLM referral source, ahead of Gemini (18) and Perplexity (12). This is notable for a ContentForge client: AI-assisted content is being indexed and cited by the same AI models used to help produce it. Claude referred traffic to the homepage, disaster relief posts, and the /services page — covering both informational and commercial intent.

Citation vs Indexation Classification

✅ Double Win

34

Pages indexed by Google and cited by LLMs — 92% of all LLM-cited pages

⚡ LLM Only — Google Rejected

3

LLMs cite these pages but Google has not indexed them

⚠️ GEO Gap

31

Indexed by Google but receiving zero LLM citations

🔕 Dark Pages

36

Not indexed by Google, no LLM citations either

3 Pages LLMs Cite That Google Won't Index

Page URL	GSC Status	Last Crawled	LLM Source	Sessions
/blog/blue-uas-compliance-requirements-for-critical-energy-infrastructure-inspections	Crawled – Not Indexed	Recent	Claude	1
/blog/how-drones-preserve-disaster-scenes-for-legal-investigations	Discovered – Not Indexed	—	Claude	1
/blog/drone-assisted-roof-inspection	Discovered – Not Indexed	—	Claude	1

What This Means

Only 3 of 37 LLM-cited pages (8%) are not indexed — a significantly healthier ratio than Client A's. All three were cited exclusively by Claude.ai. The topics (Blue UAS compliance, legal drone evidence, roof inspection) are specific and factual — exactly what AI training captures even when Google filters pages out. With content production stopped, fixing indexation on these 3 would recover the lost LLM referral potential at no additional content cost.

Top Double Win Pages — Indexed + LLM Cited

Page	GSC Clicks	Impressions	Avg Position	LLM Sessions	LLM Sources
/blog/drone-thermography	11	3,166	15.2	10	ChatGPT
/blog/roi-of-drone-pipeline-inspections-vs-manual-crews	8	9,073	9.2	8	ChatGPT Gemini
/services/drones-for-surveying-and-mapping	4	6,602	29.7	8	ChatGPT
/blog/drone-vs-manual-inspections-cost-for-adjusters	22	5,365	7.3	7	ChatGPT Gemini
/blog/drones-in-disaster-relief-drones-for-emergency-management	13	4,182	9.7	7	ChatGPT Claude Perplexity
/blog/can-ai-powered-drones-auto-detect-roof-hail-damage	40	6,910	7.3	6	ChatGPT Perplexity
/blog/drones-thermal-imaging-for-detecting-pipeline-leaks	30	21,124	8.5	5	ChatGPT Gemini Perplexity
/blog/state-regulations-for-drone-use-in-insurance-inspections	37	9,308	7.4	3	ChatGPT
/blog/insurance-companies-that-accept-drone-inspection-reports	36	6,508	9.6	4	ChatGPT
/who-we-are Company profile page	21	837	5.9	4	ChatGPT Claude

Pattern in the Double Wins

The top LLM-cited pages follow a consistent pattern: cost comparisons (drone vs manual, cost breakdowns), regulatory/compliance topics (FEMA guidelines, state regulations, OSHA), and sector-specific use cases (pipeline leaks, disaster relief, hail damage). These are exactly the types of factual, decision-relevant queries AI users ask. Notably, /services/drones-for-surveying-and-mapping is ranked at an average position of 29.7 on Google yet attracted 8 LLM sessions — LLMs are citing a page Google has buried on page 3.

🏗️

Study Findings

Proof Points

→ Google indexation is the single strongest predictor of LLM citation — indexed pages are 5–7x more likely to receive LLM traffic across both clients

→ Structured, factual content formats (cost comparisons, "X vs Y", regulatory guides) dominate both Google rankings and LLM citations — the same content wins both channels

→ LLMs are already acting as a bottom-of-funnel referral channel — commercial and brand pages received AI-referred traffic, not just informational posts

→ Contrary to prevailing expert opinion, LLMs can and do cite pages that Google has not indexed — 5 confirmed instances across both clients, all highly specific and factual in nature

Open Questions

! What content characteristics cause indexed pages to fall into the GEO Gap (indexed by Google but zero LLM citations)? Format, depth, and structured data are likely factors worth testing

! Does the 5–7x indexation multiplier hold across industries beyond admissions and drone services, or is it sector-dependent?

! How quickly do LLM citation rates decay when content production stops — and how much of the indexed base retains citations over a 6–12 month window?

LLM Citation vs Organic Visibility

platform

Resources

Compare Us

use cases

compare us