/seo-2026-sota

name: seo-2026-sota
description: "Knowledge base on SEO/GEO 2026 (trilogy 'SEO 2026 — Lo Stato dell'Arte', 'GEO 2026 — Generative Engine Optimization', 'SOTA 2026 Executive'). Use it to apply frameworks on Google technical SEO, Core Web Vitals, E-E-A-T, schema.org, AI Overviews, LLM optimization (ChatGPT/Claude/Gemini/Perplexity), GEO distribution, official Google sources/APIs/standards — studying the documents or referencing their concepts."
allowed-tools:
  - Read
  - Grep
argument-hint: [topic, framework, or chapter number e.g. ch13]

SEO / GEO 2026 — Lo Stato dell'Arte

**Sources**: trilogy "SEO 2026 — Lo Stato dell'Arte" + "GEO 2026 — Generative Engine Optimization" + "SOTA 2026 Executive" (fact-checked ed., May 2026) | **Chapters**: 20 (summary of 121 sub-chapters) | **Generated**: 2026-06-07

How to use this skill

• **Without arguments** — loads the core frameworks below as reference.
• **With a topic** — ask about `core web vitals`, `schema`, `AI Overviews`, `IndexNow`, `Reddit GEO`, `NavBoost`: finds and reads the right chapter.
• **With a chapter** — ask `ch13`: loads that file.
• **Browse** — ask "what chapters do you have?" for the full index.

When you ask about a topic not covered in the Core Frameworks below, the relevant chapter is read before responding.

---

Core Frameworks & Mental Models

The central thesis: "SEO doesn't die, it splits in two"

Two parallel programs with partly distinct levers:

• **Classic SEO** (Google) — readable site, authority, content for people and crawlers.
• **GEO** (generative engines) — readable and **citable** site by LLMs, which have their own rules and weights.

The real overlap between the two is **low (8-12%)**, not the "63%" of popular narrative. Treat them as distinct disciplines that reinforce each other only in certain segments (unique content, schema, E-E-A-T, proprietary data). See **ch04**.

The ToFu → BoFu pivot (GEO)

With zero-click (58.5% US, 83% with AIO) and the collapse of organic CTR (−58/−61%), shift budget to the **4 BoFu formats** that LLMs read to recommend: vertical listicles, "X vs Y" comparisons, choice guides with a matrix, "alternatives to [competitor]" pages. See **ch01**.

Heading-as-question + atomic answer

H2/H3 = real question; first sentence below = standalone answer 1-2 lines (≤100 chars for Gemini); detail below. This is the pattern that AIO and LLMs extract as a citation. Works for both SEO and GEO. See **ch02, ch13**.

Trust transfer

A recommendation from an LLM is processed as **editorial advice**, not advertising. Appearing as the top choice in an AI response is more like being cited by the NYT than buying a Google Ad: you don't pay the engine for it, you pay in editorial work. See **ch01**.

GEO allocation by model

Don't treat "LLMs" as a single entity. Map **customers → models → channels**: ChatGPT=Wikipedia+G2/Capterra (3×); Claude=technical docs+GitHub+papers; Gemini/AIO=YouTube+schema+recency (Reddit 0.1%); Perplexity=Reddit. Only Gemini/AIO gives measurable feedback in GSC. See **ch05**.

Core Web Vitals 2026

LCP ≤2.5s · INP ≤200ms · CLS ≤0.1 (p75, 28 days on CrUX). INP replaced FID (Mar 2024) and is the bottleneck: 43% of sites fail it due to post-load blocking JS. CWV are now the **minimum condition to be selected as a source by an AIO**. See **ch13, ch16**.

E-E-A-T with Experience dominant (Core Update Mar 2026)

First-hand experience (specific details, original data, own photos, "How we tested", author credentials with Schema Person+sameAs) outperforms "perfect" but generic pages. E-E-A-T is not a direct dial: it is the raters' lens (SQRG) that algorithms approximate. See **ch12, ch13, ch20**.

"AI Search is still SEO" (Google official, May 2026)

For Google's AI Overviews you do NOT need special markup, do NOT need llms.txt, do NOT need artificial chunking: you need classic SEO made more rigorous + **non-commodity** content. Note: this applies to Google's AIO; external LLMs (ChatGPT/Claude/Perplexity) have different mechanics (see tension in ch04 vs ch20). See **ch20**.

The 80/20 GEO rule

The 20% that delivers 80%: 3-5 BoFu listicles · G2+Capterra (20-30 reviews, 4.3+) · 3-5 long-form YouTube videos · 2-3 authentic subreddits · schema Article+Author+Organization. Black hat tactics (review farms, astroturfing) deliver short-term results with high reputational risk. See **ch06**.

Click signals are real (antitrust + 2024 leak)

NavBoost (re-ranking on clicks, 13 months), Glue, Tangram, SiteAuthority, NSR are confirmed. "Clicks don't matter" is demonstrably false → measuring CTR and dwell has strategic value. The leak provides the **vocabulary** to name what you observe in the SERPs. See **ch20**.

Verify numbers before deciding

The direction of industry GEO insights is right, the proportions are not: "17× conversion" → 4.4× median; "63% overlap" → inverted (8-12%); "70% ChatGPT" → 65-87% in erosion. Revalidate with Semrush/Ahrefs/Profound/Similarweb before committing budget. See **ch01, ch04**.

---

Chapter Index

| # | Title | Key Framework | |---|-------|---------------| | [ch01](chapters/ch01-geo-fondamenti-fiducia-bofu.md) | GEO: trust transfer, zero-click, BoFu pivot | Trust transfer, 4 BoFu formats | | [ch02](chapters/ch02-geo-playbook-6-step.md) | GEO Playbook in 6 steps | First response rule, weekly cycle | | [ch03](chapters/ch03-geo-distribuzione-off-site.md) | GEO off-site distribution | LLM source map, YouTube quick win, 3× B2B | | [ch04](chapters/ch04-seo-vs-geo-overlap.md) | SEO vs GEO: low overlap | Token-based ranking, 2 parallel programs | | [ch05](chapters/ch05-ottimizzare-per-modello.md) | Optimizing for each model | Levers for ChatGPT/Claude/Gemini/Perplexity | | [ch06](chapters/ch06-geo-topical-authority-blackhat-8020.md) | GEO topical authority, black hat, 80/20 | 80/20 rule, internal linking with Claude | | [ch07](chapters/ch07-seo-fondamenti-come-funziona-google.md) | SEO fundamentals, how Google works | 5 SE