The AI Search Era — GEO / AEO

Google's core message Confirmed: "Optimizing for generative AI search is optimizing for the search experience, and thus still SEO." AI features are rooted in core Search ranking/quality systems. There is no separate "GEO" ranking system; E-E-A-T signals are processed by the generative model.

Eligibility — nothing special to do

• Requirement: the page must be indexed and eligible to show in Search with a snippet. That's it. "There are no additional requirements to appear in AI Overviews or AI Mode."
• What you do NOT need (explicitly debunked by Google): ❌ machine-readable/AI/Markdown files ("Google Search ignores them") ❌ special schema.org for AI ❌ content chunking ❌ AI-only writing style ❌ inauthentic "mentions." ⚠️ Avoid creating "separate content for every possible variation" → trips scaled content abuse.

How links surface

• RAG: core Search ranking retrieves relevant pages, then AI shows prominent clickable links.
• Query fan-out: the system issues multiple related concurrent queries → a wider set of links than a single SERP. AI Overviews/AI Mode still fan out to traditional search underneath — organic SEO remains the substrate (SOTR ep. 109).

What actually gets you cited (measurable tactics)

• Answer-first capsule [CONFIRMED foundation + COMMUNITY position data]: open every page AND every major H2 with a 40–60 word paragraph that directly answers the query — AI extracts these near-verbatim. ~44–55% of AI citations come from the first 30% of a page.
• Credibility elements [CONFIRMED — Princeton GEO study]: quotations (+41%), statistics (+32%), inline citations (+30%) lift citation likelihood.
• Write definitively Community: cited text is ~2× more likely to use definitive language; cut hedging ("may", "might").
• Format to intent [COMMUNITY, Evertune]: listicles win commercial/comparison (63% of LLM citations are listicles, 71–86% numbered "Top-N"); long-form articles win informational; clean product pages win transactional.
• Schema Community: schema-marked pages cited ~2.3× more.
• Word count is NOT a factor [COMMUNITY, Ahrefs]: correlation ≈ 0.04; ~1,000–1,500 words is fine — don't pad.
• You must already rank [CONFIRMED-ish]: 76%+ of AIO citations come from page-one organic; ~96% from strong-E-E-A-T sources. Freshness matters. Earned media/PR is a disproportionately strong citation source (journalism = 25–27% of cited sources).
• Being cited is "the new position one": brands cited inside AI Overviews earn ~35% more organic clicks.

AI crawler robots policy (three-function model)

1. Training crawlers (GPTBot, ClaudeBot, Google-Extended token, Applebot-Extended token, Bytespider, CCBot, Meta-ExternalAgent) — blocking keeps you out of training data, NOT live answers.
2. Search/indexing crawlers (OAI-SearchBot, Claude-SearchBot, PerplexityBot) — ALLOW these if you want citations.
3. User-triggered fetchers (ChatGPT-User, Claude-User, Perplexity-User, Google-Agent) — inconsistent vendor policies.

Recommended robots.txt for a site that WANTS AI visibility:

# Allow AI SEARCH / RETRIEVAL bots (drive citations)
User-agent: OAI-SearchBot
Allow: /
User-agent: Claude-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /

# Optionally block training-only crawlers
User-agent: GPTBot
Disallow: /
User-agent: ClaudeBot
Disallow: /
User-agent: Google-Extended
Disallow: /
User-agent: CCBot
Disallow: /
User-agent: Bytespider
Disallow: /

Caveats: (1) blocking training bots does NOT remove you from live AI answers — those come from search bots (most common mistake). (2) Google-Extended/Applebot-Extended are opt-out tokens, not crawlers — disallowing them opts out of Gemini/Apple training only, NOT Search ranking or AI Overview eligibility. To opt out of AI Overviews without killing Search, use the GSC toggle (§12). (3) robots.txt is unenforceable against Bytespider and Perplexity stealth crawlers [CONFIRMED — Cloudflare delisted Perplexity as a Verified Bot, Aug 2025]; use WAF/server 403/CDN. (4) Watch the Cloudflare Sept 15, 2026 deadline — training/agent crawlers blocked by default on ad-serving pages behind Cloudflare.

llms.txt — reality check Confirmed

Do not invest in llms.txt expecting citation/traffic gains. 97% of llms.txt files received ZERO requests (Ahrefs, May 2026); AI retrieval bots were 1.1% of the requests that did occur — top requesters were SEO audit tools. Google does NOT use it — Mueller (2026-06-02): "purely speculative… none of the AI systems use it," compared to the deprecated keywords meta tag; it "lacks a discovery mechanism." The one legitimate use case: developer-product docs consumed by coding assistants (Cursor, Continue). Google points to WebMCP as the more promising direction; the most basic agentic optimization is "don't block agents."

Bing & AI

Bing added GEO to its official guidelines (Feb 2026) and aligned its AI-content stance to Google (quality over production method; target = "large-scale content generated without oversight"). New named abuse: AI-citation-engineered language + prompt injection. Bing AI-usage meta controls: nocache (only URL/title/snippet in AI+training), noarchive (excluded from AI answers/training but still ranks). Bing runs an AI Performance report (Total Citations, Grounding Queries) and uses IndexNow.

Related on Crawlinx

• GEO & AEO: How to Get Cited in AI Search (2026)
• Control AI Crawlers: GPTBot & robots.txt (2026)
• Google Search Updates 2025-2026: What Changed
• content.thin

Sources

1. Google Search Central — AI features and your website
2. Google Search Central Blog — Succeeding in AI search
3. GEO: Generative Engine Optimization (Princeton, arXiv)