Andrea Volpini

Good morning 🦞 WordLift Claw is your SEO execution layer in Slack, helping you spot AI Overview opportunities at robot speed.

@wordliftit has been selected by ICSC 4 Startup, with our work on #SEOcrate supported by Italy’s National Research Centre for HPC and Quantum Computing. Ontology-led RL for training trustworthy Small Language Models for Agentic AI is getting funded 🦾🥳 wor.ai/GCZrGV

Google just put out its first official article on optimizing for AI search (AEO/GEO)! TL;DR: ⭐️ SEO is still the foundation for AI search. ⭐️ Create non-commodity, "people-first" content. ⭐️ Ignore most "GEO/AEO hacks like chunking and llms.txt. developers.google.com/search/docs/fu…

The mistake many AI visibility solutions made: confusing scale with durability. Scalable SEO/GEO templates may win temporarily, but they leave footprints that search systems eventually learn to detect.

AI visibility is not a content-volume game. It is an evidence game. In e-commerce, the winners are not the brands publishing more pages, but the ones giving AI systems stronger, clearer, and more verifiable evidence. Great insights from @aleyda aleydasolis.com/en/ai-search/e…

@mpedrao #05132026" target="_blank" rel="nofollow noopener">support.google.com/analytics/answ…

@suganthan I’m usually very selective about what I read on schema 😀 but I knew I would appreciate your clarity on the topic.

@cyberandy I love the graphic. (Made me look much younger too lol) Thanks for sharing. Yes great point on copilot. I will add it to the article. 🙏💪

Worth reading today. “Think of schema markup as business registration, not advertising.” I’d add: Bing relies heavily on schema too, and the semantic layer you build for search is the layer that aligns your AI workflows. Same schema. Three lives: indexing, training, retrieval.

@ziodave @cyberandy @lilyraynyc @jdevalk We bet on schema.org several years ago as the canonical data model for our “observed entities” in content. It’s been a valuable bet since now we offer that as a context layer for AI agents, pre-indexed, and with standards-based entity layer.

"You're not scaling content. You're scaling disappointment" Go read this 👇🏻

Super interesting catch by @natzir9, which seems confirming a suspect I had about Google grounding also using its Shopping Graph because, let’s be honest, it would be strange Google not using it if it aims to expand transaction via AI Mode

The semantic layer is all you need.

Checkout the blog post here: alphaxiv.org/blog/reinforce…

One of my pet peeves is that many SEOs haven't updated their mental model of how a mondern search engine works. So, I wrote it down for you. serprecon.com/blog/how-a-sea…

I spent the last 3 weeks running what might be the most comprehensive LLM ranking factors analysis to date. 29,562 unique domains tracked and scored across 145 industries, 1,595 buyer personas, and 105k+ ChatGPT prompts. Over 500TB of data, and 12 external signals correlated against rank-weighted LLM recommendation scores. This is a large-scale correlation study: what external signals actually predict whether a brand gets recommended by ChatGPT, across 145 industries and 1,595 buyer personas. -- Research Process 145 industries from 500 candidates. 11 personas each (10 targeted + 1 neutral). 25 runs per persona, rank-weighted scoring. 29,562 unique domains tracked. Data collected: - Common Crawl: 1.15B pages, domain mentions + phrase co-occurrences - Reddit: 5B+ posts and comments scanned - Google Search: 15,697 queries, top 100 results; 1.5M+ results captured - SERP HTML: parsed for outbound links and phrase presence - Wikimedia: 300M+ Wikidata entities + Wikipedia citations - Backlinks (Common Crawl Web Graph): PageRank + Harmonic Centrality; 4B+ - Top Site Homepages: parsed for persona-specific phrases -- Analysis Process 13 signals per domain. Spearman ρ vs. LLM recommendation score, per-industry and globally. R² shows variance explained. Lift measures over-representation in the top 10% most-recommended domains. Tiered: Dominant (ρ ≥ 0.30) down to Baseline (< 0.05). -- Key Findings SERP appearances, SERP rank, and outbound links from search results pages are the three strongest signals. Traditional SEO is the dominant measurable influence on LLM recommendations. Backlink authority (PageRank, Harmonic Centrality) follows. Combined, these point to one thing: established search authority drives LLM visibility. Signal hierarchies vary by industry. Wikidata dominates in established categories (hotels, ERP, furniture). Reddit drives community-driven ones (enterprise AI, live entertainment). No universal strategy. 80–85% of recommendation variance is inside the model. All external signals combined explain under 20%. You cannot infer LLM visibility from search rankings; you have to test it directly. -- The Two Conclusions That Matter 1. SEO is the foundation. OpenAI is using search data today and building their own index. As that matures, the connection between search authority and LLM visibility deepens. Traditional SEO principles are not obsolete, they're the starting point for LLM visibility too. 2. Persona is the measurement unit. The #1 airline for a frequent flyer is a different site from the #1 for a student flying abroad. Same model, same industry, different person, different result. You don't have one LLM rank, you have a rank per buyer segment. Monitor by persona or the number is meaningless. -- Full Report and data for all 145 industries and 1,595 personas available here: oppalerts.com/LLM-Ranking-Fa…

The mouse pointer as context.

@ziodave @glenngabe @lilyraynyc @cyberandy @wordliftit I also wrote a thing: iloveseo.net/the-ahrefs-sch…

1/6 There’s a lot of discussion around the Ahrefs AI citations report, but several concepts are getting mixed together. Schema.org and JSON-LD are not the same thing. ahrefs.com/blog/schema-ai… cc @glenngabe @lilyraynyc @cyberandy @wordliftit

More details to replicate the analysis: wordlift.io/blog/en/how-ai…

I began using Gemma 3 as an open interpretability lab. Prompt → layer-32 residual vectors → Gemma Scope SAE activations → Natural Language Autoencoder explanations. The goal is to map how a model internally represents a brand, its products, and competitors before it answers.

Context Graphs help the agentic workforce understand the “why” behind decisions. In most cases, these graphs need to be mined from conversations. But the missing layer is not the graph technology itself, RDF and OWL can support that. What is missing is the knowledge architecture.

Local open-weight AI on a laptop has been improving more than twice as fast as Moore's Law! Between May 2024 and May 2026, the most expensive MacBook Pro you could buy stayed at 128 GB of unified memory. The hardware ceiling barely moved. But the smartest open-weight model from @huggingface you could actually run on it went from a score of 10 (Llama 3 70B) to 47 (DeepSeek V4 Flash on @antirez's mixed-Q2 GGUF) on the @ArtificialAnlys Intelligence Index. That is 4.7× in 24 months, or a doubling of intelligence every 10.7 months. Moore's Law (transistor count) doubles every 24 months. Local open-weight AI on a laptop has been improving more than twice as fast as Moore's Law, on completely unchanged hardware.