|
|
|
Issue #73 · July 21, 2026
When You Ask ChatGPT Which AI to Use, It Quietly Steers You to OpenAI — 2.0x More Than Other Engines Do.
AI Self-Citation Bias · July 2026 · 5W AI Communications
5W AI Communications analyzed 32,200 prompts across five major AI engines and found a consistent self-promotional pattern: ChatGPT recommends OpenAI models 2.0x more often than other engines recommend the same models. Gemini recommends Google DeepMind 1.7x more often. Google AI Overviews steers toward Google products 1.6x more often. Perplexity surfaces itself in adjacent queries at 2.4x baseline.
Claude showed the smallest bias, landing within 1.2x of a neutral baseline — the only engine approaching equal treatment. The market concentration finding adds context: OpenAI captures 24.6% of all citations across AI assistants when answering questions about the AI industry — nearly a quarter of every AI-category answer surface. For brands competing in any space where AI engines also have a stake, this data suggests the recommendation layer is not neutral.
|
|
|
|
|
|
Citation Research · 2026 · CiteLens
SEO Decides 93% of Google AI Citations and 89% of Perplexity's. For ChatGPT, It's Only 30%.
CiteLens analyzed 320 queries across four AI engines and found the platforms split into two groups. Google AI Mode and Perplexity are effectively SEO-driven: 93% and 89% of their citations respectively come from Google's top-10 organic results. ChatGPT is almost the opposite: only 30% of its citations overlap with Google's top 10, with near-zero correlation to both Google ranking and brand popularity. Claude sits in between at 53%, with a distinct preference for Wikipedia-backed sources (58% of citations). "Marketers are told to optimize for AI as if it's one thing — our data shows it's at least three."
|
|
Read the study ->
|
|
|
|
|
|
|
|
Citation Formats · 2026 · DeltaV Digital
25,337 Citations Across 8 Industries: Comparison Pages Earn 45% More Citations. LinkedIn Is #1 for B2B.
DeltaV Digital tracked 25,337 citations across 21,075 AI responses in 8 industries over 90 days (April-July 2026). Comparison pages emerged as the highest-performing format at 1.87 citations per retrieval — 45% above the average. Industry format preferences diverge sharply: B2B Tech favors listicles (61% of citations); healthcare cites articles exclusively (54%). For B2B services specifically, LinkedIn was the single most-cited domain with 736 citations — outranking industry publications. "Listicles earn 61% of AI citations in B2B services. In healthcare, they earn 0%."
|
|
Read the full study ->
|
|
|
|
Technical Signals · 2026 · Onely
73% of Google Page-1 Brands Have Zero AI Mentions. LCP Over 4 Seconds Cuts Citations 72%.
Onely's 2026 AI visibility guide surfaces the scale of the gap between Google ranking and AI presence: 73% of brands ranking on Google page one have zero mentions in AI responses. The platform gap is also stark — despite ChatGPT's 81.84% market dominance, it has only a 0.59% citation rate for brands; Perplexity cites at 13.05%. Technical performance is a hard threshold, not a ranking factor: pages with LCP over 4 seconds see a 72% reduction in citations, and 85% of AI-cited pages pass all Core Web Vitals. Schema implementation (Organization, Brand, AboutPage) boosts citations up to 3x in some verticals.
Read the full guide ->
|
|
Navarro's Take
"The CiteLens split is the most important tactical finding in this issue. Google AI Mode and Perplexity are SEO-driven — rank well on Google and you'll likely get cited. ChatGPT operates on different logic entirely, with only 30% overlap with Google's top 10. This means your GEO strategy for Google AI is effectively your SEO strategy, while your ChatGPT strategy needs to be something else: earned media, brand mentions, Wikipedia-adjacent authority. The 5WPR self-citation data is a warning for anyone trying to build visibility in a category where the AI engine has a competing product. You are not optimizing on a neutral surface."
- Ido Navarro, LLMagnet
|
|
|
