Signal — AI Knowledge: ASO & GEO Insights
This Week in Brief
AI referral traffic is consolidating around three platforms — ChatGPT, Gemini, and Perplexity — with Statcounter data confirming Gemini overtook Perplexity for the No. 2 spot in March 2026. Meanwhile, practitioner consensus is hardening around a core GEO principle: structured, answer-led content is the primary lever for citation in AI-generated responses. Issue #1 maps the foundational signals every ASO and GEO practitioner needs to track.
AI Lab Signals
Google Gemini Claims No. 2 Position in Global AI Referral Traffic, Edging Perplexity
Statcounter's March 2026 data shows ChatGPT commanding 78.16% of global AI chatbot referrals, with Gemini at 8.65% now ahead of Perplexity at 7.07%. For practitioners, this confirms that content optimised solely for ChatGPT citation is leaving meaningful Gemini-driven referral share on the table — Google's own AI surface is becoming a material traffic source distinct from traditional Search.
ChatGPT Processes Over 100 Million Daily Search Queries as AI Search Becomes Mainstream
LLM Intel reports that ChatGPT now handles over 100 million daily search queries, a figure that contextualises the scale of AI-mediated information retrieval. Practitioners should treat ChatGPT Search not as an experimental channel but as a primary discovery surface requiring deliberate citation-optimisation strategy.
Perplexity's 'Deep Research 3.0' Surfaces Multi-Source Synthesis as a Standard Interaction Pattern
A March 2026 Perplexity update described as 'Deep Research 3.0' enables multi-step retrieval across hundreds of pages to produce cited, structured reports. Note: this characterisation originates from a non-primary source and specific performance claims should be treated as unconfirmed until verified by Perplexity's official communications. The practitioner implication stands regardless: longer, comprehensively cited documents are more likely to be ingested and quoted by deep-research retrieval pipelines.
Training Data & Crawl
No significant developments this week.
AI Search & ASO
Google AI Overviews Now Present on 65% of Queries, Cutting CTR by an Estimated 40%
LLM Intel cites figures indicating AI Overviews appear on 65% of Google queries and that a 40% click-through rate drop is observed when an AI Overview is present. These figures are attributed without a named primary study — practitioners should treat the specific percentages as indicative rather than definitive, pending a primary source. The directional signal is consistent with broader industry observation: zero-click outcomes are rising, and being cited inside the Overview is increasingly the only visibility available on affected queries.
Structured Content Receives 3x More Citations in AI-Generated Answers, Per Practitioner Reports
LLM Intel reports a 3x citation advantage for structured content in AI search responses. The primary study underlying this figure is not named in the source — treat as (Unconfirmed) pending verification. The finding is directionally consistent with GEO guidance from Incremys, which specifies that answer-led, clearly structured pages increase the probability of selection by generative engines.
Research Radar (arXiv)
This technical guide details how RAG systems chunk, embed, and retrieve source documents — covering chunk size trade-offs, vector store selection (ChromaDB vs. Qdrant), and query optimisation configuration. For GEO practitioners, understanding chunking behaviour is directly actionable: documents broken into semantically coherent, self-contained sections are more likely to survive the retrieval pipeline intact and surface as cited sources in AI-generated answers.
Practitioner Takeaway
Audit your highest-value pages for 'chunk survivability': each major section should be able to stand alone as a complete, accurate answer to a discrete question. Given that RAG systems retrieve and embed content in fixed-size chunks, and that structured content attracts an observed citation advantage, restructuring pages around discrete question-answer units — with clear headings, defined terms, and embedded citations — is the single highest-leverage GEO action available this week. Prioritise pages targeting queries where Google AI Overviews already appear, as those represent queries where citation-or-invisibility is already the operative dynamic.
The 6-phase framework used to structure this newsletter is available as a complete methodology guide — including audit tools, templates, and implementation checklists.
Get the Framework — $299New to AI knowledge publication? Download the free briefing flyer — helping you make the case for why your organisation cannot wait.