Jonathon's AI Wiki

SEO & Content

9 min read

An applied SEO and content stack pulls four loosely-coupled pieces into one feedback loop: per-query search optimization driven off Google Search Console data, multi-agent content generation that produces and quality-checks drafts, weighted competitive intelligence that watches what other players in the niche are publishing, and rule-based site audits that catch technical regressions.

The leverage isn’t in any single tool — it’s the loop. A query trending in GSC becomes a content brief, the brief produces a draft, the draft is quality-checked against on-page rules and the competitive set, and once published its impact lands back in the next GSC sweep. Layered on top, AI-search-visibility frameworks (e.g. FLUQs) shape content for citation by LLM answer engines as well as classical search.

Official sources

  • Google’s Official Generative AI Search Optimization Guide (AI Overviews + AI Mode) — Official Google Search Central documentation. Resolves the “AEO” / “GEO” terminology debate: from Google’s perspective, optimizing for generative AI search is still SEO — AI Overviews + AI Mode rely on the same Search index via RAG + query fan-out. To be AI-eligible, a page must be indexed + snippet-eligible. Reinforces FLUQs’s core-SEO-first emphasis. Primary source for any AEO-related claims; avoid third-party tactical guides where this doc contradicts them.

Independent research & data studies

This is an actively-tracked thesis cluster on AI search citation mechanics — what makes AI engines (Google AI Overviews, AI Mode, ChatGPT, Perplexity, Brave, Gemini) cite some pages and not others. The cluster splits cleanly along methodology — causal vs. correlational vs. engine-specific vs. meta-analytical vs. market-context — and the framing matters for what each study supports. The AI SEO hub surfaces the whole cluster in one methodology-grouped view.

Causal evidence (matched difference-in-differences):

  • Ahrefs Schema → AI Citations Causal Study (Linehan & Guan, 2026-05-11) — Only matched-DiD study in the cluster. 1,885 pages adding JSON-LD schema vs. 4,000 matched controls, Aug 2025-Mar 2026, three surfaces (AIO / AIM / ChatGPT). Adding schema produced no statistically meaningful citation lift on any surface. AIO 4.6% decline statistically significant but inside larger declining trend, not claimed schema-caused.

First-party controlled experiments (causal evidence at the format level):

  • Marketing to AI Agents — Ramp’s 5-Week Controlled Experiment (Grace Cummins, 2026-04-30) — First published B2B controlled experiment on marketing tracked incentives directly to AI agents. Three concurrent content-format variants (markdown / stripped HTML / schema) served via Cloudflare Workers across ~50 ramp.com pages. Markdown wins decisively over schema (the format literally designed for machines). Per-model behavior wildly different: Claude ~6 matches/day with exact offer + booking URL + claim instructions; Perplexity fast but vague forever (“some channels offer bonuses”); ChatGPT zero matches for 32 straight days. Day-21 step-change: Claude’s daily relay rate jumped 4× overnight with no observable external cause. 5-week totals: 1,300+ bot visits → 370 agent relays → peak 33 citations/day. Anthropic’s crawler more aggressive than all other named AI crawlers combined on Ramp’s content — independently corroborated by Anthropic’s own Measuring AI Agent Autonomy usage data. “Agent trust” emerges as the load-bearing signal — analogous to domain authority but the signals differ; pages LLMs already cite frequently are where new content surfaces, regardless of format. Format-level causal evidence + per-model differential behavior measured against one constant payload = two structural contributions no prior study supplied.

Citation geometry (longitudinal re-measurement):

  • Ahrefs 38% AI Overview Top-10 Update (Linehan & Guan, 2026-03-02) — re-run of the July 2025 “76% from top 10” study. 863K keyword SERPs / 4M AIO URLs. Top-10 overlap fell 76% → 38% in seven months, attributed to the Gemini 3 rollout (Jan 27, 2026) + expanded query fan-out. ~Two-thirds of AIO citations now originate outside the query’s own top-10 — fan-out coverage moved from optional to load-bearing.

Correlational evidence (cross-sectional observation, no causal claims):

  • AirOps + Kevin Indig Fan-Out Effect ChatGPT Study (2026-04-13) — Largest single-engine dataset to date. 16,851 queries / 50,553 ChatGPT responses / 353,799 pages. Headline: retrieval rank dominates everything (rank-1 cited 58.4% vs rank-10 cited 14.2%, 4.1× gap). Schema +6.5pp in stratified analysis. Query fan-out: 88.6% of ChatGPT queries trigger exactly 2 sub-queries. DA shows no positive correlation on ChatGPT. Refreshed 2026-05-19 with the “From Retrieved to Cited” commercial-content companion (comparison pages 3 tables +25.7%, validation pages +26.9%, 5-7 stats +20%, ≤10-word sentences +18.8%).
  • Digital Applied 1,000 AIO Citation Pattern Study (2026-04-26) — Within-query control study on Google AI Overviews. 1,000 AIOs / 4,243 cited URLs / ~50,000 control URLs. Headline: top 1% of domains capture 47% of all citations. Schema lifts 2.3× (Article + BreadcrumbList) → 2.8× (HowTo) after regression-style DA control. DA Pearson +0.61 on AIO (engine-divergent from AirOps’s ChatGPT-null finding).
  • SE Ranking — 50+ AI Mode Ranking Factors (2025-12-15) — Primary research scoring 50+ candidate factors for AI Mode citation, top 20 ranked. Global domain traffic is ~3× more predictive than content-quality factors — the AI-Mode-specific primary-data counterpart to Zyppy’s cross-engine meta-analysis.
  • GEO-16 Framework (Kumar & Palkhouski, arXiv 2509.10762v1, 2025-09) — First academic AEO/GEO citation study. 1,100 URLs / 1,702 citations / 70 prompts / 3 engines (Brave + AIO + Perplexity). Pages scoring GEO ≥0.70 + ≥12 of 16 pillar hits → 78% cross-engine citation rate. Top correlations: Metadata & Freshness r=0.68, Semantic HTML r=0.65, Structured Data r=0.63 (all p<0.001).

Engine-specific evidence (one AI surface isolated; cross-engine divergence):

  • Ahrefs — AI Mode vs AI Overviews (730K responses, Q1 2026) — Cleanest within-Google divergence data. AI Mode and AIO cite the same URLs only 13.7% of the time for the same query (16.3% for top-3). YouTube tops AIO; Wikipedia/Quora/Facebook over-index in AI Mode.
  • SE Ranking — Google Self-Citation in AI Mode (1.3M citations, 2026-03-06) — 68,313 keywords / 1,321,398 citations / 20 niches. Google.com is 17.42% of all AI Mode citations — more than YouTube + Facebook + Reddit + Amazon + Indeed + Zillow combined. Tripled from 5.7% in nine months; composition shifted from 97.9% Google Business Profiles to 59% organic Google SERPs.
  • SISTRIX AI Citation Drift (2026-05-01) — 82,619 prompts / 1,548,213 snapshots / 6 countries / 3 platforms / 17 weeks. “Fixed core + carousel”: 86% of prompts hold a stable 1-5 domain core; AIO rotates 56%/week, ChatGPT 74%/week. Reframes GEO from “am I cited?” to “am I in the core or the rotating set?”

Meta-analysis:

  • Zyppy AI Citation Ranking Factors Meta-Analysis (Cyrus Shepard, 2026-05-07) — Synthesis of 54 published experiments, patents, and case studies → 23-factor ranking with 0-10 evidence-based scores across ChatGPT / Gemini / Perplexity. Top tier (9+): URL Accessibility, Search Rank, Fan-out Rank, Preview Control, Query-Answer Match, Intent-Format Match. Structured Data scores 5.6 (#20). LLMs.txt scores 2.0 (#23) — the most overhyped 2025 tactic. Thesis: “win SEO, win AI citations (most of the time, with extra steps).”

User behavior + market share (macro context):

  • Datos + SparkToro State of Search Q1 2026 — Clickstream panel, millions of US/EU/UK desktop users, 12 months. The reality check: AI tools are <2% of total desktop visits; US zero-click fell 24.5%→22.4%; organic click share rose to 44.9%. Tempers any “AI search is already the channel” framing.
  • Similarweb 2026 Generative AI Brand Visibility Index — Brand-mention share across ChatGPT/Gemini/Copilot/Perplexity across 6 sectors. AI referral traffic plateauing even as platform usage grew +28.6% — in-answer visibility matters more than chasing AI referral clicks. Companion to the most-cited-domains study below (different angle). The Stanford HAI 2026 adoption data and Pew sentiment survey complete this layer (both in ai-industry-research, surfaced via the hub).

Practitioner frameworks & companion data:

Industry trends guides (practitioner perspective):

  • Conductor 2026 Q1 AEO & Content Marketing Trends Guide (Reinhart + Vize, Q1 2026) — 22-page vendor practitioner trends doc; layers Clutch 2026 State of Content survey data (41% brand reputation = primary goal, 75% expanded AI tools, 77% creating for LLMs, 52% increasing video) onto Conductor commentary. Contributes the five-indicator AEO ROI framework (share of AI citations / sentiment / prompt-level visibility / competitive model share / AI-influenced conversions) — the cleanest articulation of the post-traffic measurement stack in the cluster. Anti-pattern callouts: skip LLMs.txt-style markdown duplication; avoid thin self-promotional listicles (Lily Ray cited 30-50% traffic loss); don’t treat AI referral traffic as primary KPI.
  • O & Google Marketing Live 2026 Recap (May 2026) — CMO-level strategic synthesis reading I/O 2026 + GML 2026 (~May 21-22) as one move: Google rebuilding discovery/shopping/advertising around Gemini agents. Load-bearing thesis: “authority becomes distribution” as AI mediates discovery (“Does the AI trust your brand?”). Canonical list of the GML 2026 ad products (Ask Advisor, Asset Studio, AI Mode ad formats, UCP + Universal Cart), corroborated against Google’s official announcements. Opinion, not data — the article reconciles the thesis against the empirical cluster (supported by SE Ranking + Ramp “agent trust”; tempered by Datos’s <2% reality check and AirOps’s ChatGPT-DA-null divergence).
  • GEO Profitable — Lessons From 100+ Campaigns (NP Digital) — Neil Patel / NP Digital operator playbook from 100+ AEO/GEO client campaigns across 28+ countries. Core argument: AI-search visibility (mentions, citations, impressions) is a vanity metric unless tied to revenue. Load-bearing rules: do GEO and SEO (ranking organically makes an LLM citation more likely); the 80/20 rule — ~80% of what AI cites comes from sites that aren’t yours (off-site mentions, PR, third-party reviews, roundups), so “if you only exist on your domain, you’re invisible to LLMs”; a three-layer signal stack (retrieval readiness → authority → distribution); four citation-worthy content types (comparison/alternatives pages, first-party research, bottom-funnel education, FAQ frameworks); and a reordered measurement frame — put a business-outcome metric next to every visibility metric. Ships a concrete 90-day plan (audit+foundation → create+distribute → convert+measure) plus distribution mechanics (LinkedIn + YouTube are the heaviest LLM pull sources; review velocity beats raw total). The operator/measurement companion to Conductor’s AEO trends guide.

The cross-study tension to know: The correlational studies (AirOps, Digital Applied, GEO-16, SE Ranking, Zyppy meta-analysis) all show schema-using pages get cited more — by magnitudes ranging from +6.5pp to 2.3× to r=0.63. Ahrefs’s causal study (matched DiD) found no causal lift from adding schema. The reconciliation: schema is a marker of editorial / technical / publication-infrastructure maturity that correlates with citation, not a causal lever in isolation. Practitioner implication: ship schema (the cost is low, the parseability benefits Google’s classical surfaces); don’t expect it to be the lever that moves AI citations on its own.

19 items under this folder.