The May 2026 Core Update by Google has marked a point of no return in modern SEO: visibility no longer just means appearing in traditional organic search results, but being Cited and synthesized from AI models In AI Mode, AI Overviews, and alternative search platforms like Perplexity, Claude.ai, and ChatGPT. For Italian publishers and companies with a global audience, this represents both an unprecedented threat and opportunity.
The massive adoption of AI-powered synthesis systems has rendered traditional SEO strategies focused solely on ranking positions and click-through rates obsolete. “Real citability” — that is, the ability of content to be extracted, cited, and amplified by AI overviews — has become the new strategic axis on which to build traffic, brand authority, and long-term visibility.
This article analyzes the technical and strategic context of the May 2026 Core Update, dissects empirical research on citability in AI systems, and presents a tested operating framework to build content and strategies that truly position themselves in AI synthesis flows, without resorting to gimmicks, keyword stuffing, or temporary tactics.
The Paradigm Shift: From “Position in Results” to “Citations in AI Systems”
In 2025, the discussion about visibility was still polarized around established concepts: ranking position, click-through rate, bounce rate, and dwell time. In May 2026, Google has silently redefined ranking factors integrating “extractability and semantic coherence” metrics for AI models.
As analyzed in the previous article on Zero-Click Permanent and AI Overview Citations, “synthetic” traffic—that is, implicit traffic generated by citation in AI Overviews—now represents a significant share of overall visibility, especially for B2B, informational, and news sectors.
The practical implications are three:
- A website might not appear on the first page of Google for a specific query but be cited in 5-7 global AI Overviews.;
- Visibility in AI Mode does not strictly follow the same organic search result ranking hierarchy.;
- AI citation factors require a parallel and distinctive optimization Compared to traditional SEO.
Analysis of the May 2026 Core Update: What Really Changed
Google has not released detailed official notes on the May 2026 Core Update as it has for previous updates. However, comparative analysis of Search Console data, SERP tracking, and citation monitoring has revealed three macro-trends:
1. Downranking of “Generic Summary” Content and Upranking of “Specific Extract” Content”
Purely informational content, written to cover broad intent with a straightforward structure (introduction → body → conclusion), has experienced a general drop in rankings of 15–22%, even when domain authority remains the same. In contrast, content structured to semantic extraction — with clear data points, comparison tables, original data, and a list of resources — have gained visibility in both traditional rankings and AI citation.
The difference: “synthesizable” content has sections that an AI model can extract and summarize coherently without additional context. Generic content does not.
2. Boost of “Entity Authority” compared to “Domain Authority”
As discussed in detail in the article about Entity Authority in 2026, Google began to weigh the authority of’content entity (author, organization, source) instead of just domain authority. An article on a low-domain authority site, but written by an established entity and frequently cited in AI Overviews, is more likely to rank and be cited than an article on an authoritative site but attributed to no one.
3. Integration of “Query Fan-Out” into Ranking Metrics
The May 2026 Core Update has boosted the query fan-out — the ability of content to serve the semantic sub-intentions of a broad query. As explored in Surfer AI and the 500+ Ranking Factors, Content that answers not only the main query but also 3-5 related semantic variations receives visibility boosts from both AI ranking and citation.
This metric is critical because AI overviews synthesize more sources to answer a single query. If your content only covers the main query, you have a lower probability of being cited. If you cover the main query + semantic fan-out, the probability increases exponentially.
The CRAFT Framework for Real Citability in AI Mode: 5 Operational Pillars
Based on empirical analysis and testing on dozens of Italian websites in Q2 2026, a recurring AI citability success pattern emerged. The framework has been named CRAFT framework:
C — Clarity (Structural Clarity)
AI models extract information from textual structures predictable and consistent. Content with clear headings, numbered lists, tables, and code blocks is inherently more extractable than a dense wall of text.
Operational actions:
- Use H2 and H3 headings that reflect query undertones (not clickbait titles);
- Structure each section with an introductory sentence, followed by bullet points or a table;
- Include at least one comparison table or a numbered list for each main section;
- Use the tag
<strong>for key concepts that are worth extracting.
Example of a NON-AI-optimized structure:
“Domain authority is an important factor. Many SEOs think it's the most relevant factor, but actually 2026 changed the rules. Other metrics now count too...”
Example structure optimized by AI
“In 2026, Entity Authority has surpassed Domain Authority as a primary ranking factor. The three critical factors are: (1) Recognized author, (2) Citations in AI Overviews, (3) Semantic consistency. Domain Authority remains relevant for 35% of the overall signal.”
R — Original Research (Proprietary Data)
AI systems are trained to prioritize sources with original datasurveys, empirical research, proprietary metrics, case studies with specific numbers. An article that includes “we tested 150 Italian websites and found that...” is drastically more likely to be cited than one that repeats generic insights.
Operational actions:
- Run small surveys or polls with your users (even 50 respondents are valid);
- Track proprietary metrics relevant to your industry;
- Document case studies with concrete figures (X% increase, Y hours saved, Z% growth);
- Repeatedly cite these original data, with sample and methodology specifications;
- Include a “Methodology” section for articles containing empirical research.
Original research doesn't necessarily mean expensive study. An analysis of 100 Italian websites in your industry, with public metrics collected via SEMrush or Ahrefs, is already considered “original” by AI systems.
Attribution & Authority
AI quotes prefer content with explicit attribution The author and their credentials. An article signed by “Andrea Rossi, 10 years of SEO experience, published on SearchEngineJournal” is 3-5 times more likely to be cited than one “By Admin” or without a byline.
Operational actions:
- Create a detailed author profile in WordPress: bio, professional photo, LinkedIn/Twitter links, previous publications;
- Use schema markup Author e Organization (JSON-LD);
- For critical content, include a final line: “Written by [Name], [Credential]. Last updated: [Date]. Verified by [Editor/Reviewer].”
- Build entity authority signals: appear in others' articles, be cited in Italian podcasts, participate in public conferences.
Entity authority is the new E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) in 2026. It’s not enough to have experience; the AI system must check your authority through external signals.
F — Fact-Checking & Updateability
AI models are trained to prefer content updated and verifiable. An article published 2 years ago without updates is less likely to be cited, even if it ranks well. Conversely, an article updated monthly, with citations and links to primary sources, is highly retrievable.
Operational actions:
- Update key articles at least every 2-3 months; record changes in a visible changelog;
- Include publication and last modified dates in ISO 8601 format (2026-05-15);
- Link to primary and authoritative sources (Google, government agencies, peer-reviewed studies);
- Use the tag
HTML5 for marking important dates; - For time-sensitive content (core update analysis), update the title with the date (e.g., “May 2026 Core Update Analysis”).
AI overviews synthesize the most recent and verifiable content. If your article is stale, it's less likely to be cited, regardless of organic ranking.
Topic Saturation & Differentiation
In 2026, AI systems have access to thousands of articles per topic. Differentiation is critical. An article that repeats verbatim the points of other top-ranking articles will not be cited; one that introduces a new perspective, a novel framework, or constructive criticism is much more extractable.
Operational actions:
- Perform an analysis of the top 10 articles for your keyword, identify what they all say and what no one says.;
- Build your content around a distinguishing thesis Everyone's talking about Entity Authority, but no one explains how to measure it in Python.;
- Include a critical or counterintuitive section (e.g., “Why Query Fan-Out Doesn't Always Improve Rankings” or “The Limits of Proprietary Data in AI Citability”);
- Offer one original framework (CRAFT, as in this article) rather than a simple list.
AI overviews rarely cite redundant sources. If you're the fifth article saying the same thing, the AI system will likely cite the first three and ignore the rest.
Practical Implementation: Step-by-Step for Your WordPress Site
The CRAFT framework is not theoretical; it is implementable with standard WordPress tools. Here is an operating procedure:
Step 1: Current Citability Audit
Before optimizing, measure the current state. Use the techniques described in the article on Testing AI Mode with Search Console API to identify which of your articles are already cited in AI Overviews and AI Mode.
Recommended tools:
- Google Search Console API (native, free): track impressions in AI Mode;
- Semrush or Ahrefs: monitor citability in AI Overviews (new “AI” section in 2026);
- Replit + ClaudeCreate a custom script to monitor quotes in real time (see Article on GEO Monitoring with Claude and Replit).
Step 2: Structural Optimization of Existing Content
For the 10-20 articles receiving the most traffic:
- Add clear headings (H2/H3) that reflect semantic sub-intents;
- Create at least one comparison table or numbered list per section;
- USA
<strong>for key concepts that can be extracted; - Add a “Recap” section at the end with a list of 3-5 main points.;
- Update the modification date and add a visible changelog.
This optimization does not require a complete rewrite; it is a restructuring of existing content.
Step 3: Creating New Content with the CRAFT Framework
For new articles:
- Fan-Out Search: For the main keyword, identify 3-5 semantic sub-queries.
- Distinctive Thesis: Define what you will say that others don't.
- Original Research: even small (survey of 30 people, analysis of 50 sites).
- CRAFT Structure: Clarity (heading + table), Research (include data), Attribution (byline + credentials), Fact-checking (link to sources), Topic saturation (differentiation).
- Schema Markup: Use JSON-LD for Article, Author, Organization, and, where relevant, HowTo or FAQ.
Step 4: Configure Full Schema Markup
Schema markup is critic for AI citability. WordPress 7.0+ includes native improvements for block schema. If you don't have WordPress 7.0, use plugins like Yoast SEO o RankMath.
Required schemas:
- Articleheadline, datePublished, dateModified, author, image;
- Authorname, email, url, description;
- Organizationname, logo, url, sameAs (social profiles);
- BreadcrumbListfor navigation;
- FAQ: if the article contains questions/answers (like this one).
Check on Schema.org Validator e Google Rich Results Test.
Step 5: Continuous Monitoring and Iteration
Configure a monitoring dashboard con
- Monthly Insights in AI Mode (Search Console);
- Number of citations in AI Overviews (Semrush/Ahrefs or custom script);
- Entity Authority Score (derived from authority backlinks, mentions, AI citations);
- Update frequency (days since last update);
- Traditional ranking position (by correlation).
As described in Anti-Volatility Strategy with Looker Studio and Slack, integrate these signals into Looker Studio or Metabase for automatic alerts.
Case Study: An Italian Website That Increased Its Citation Index by 340% in 3 Months
An Italian content marketing publisher, with 200 articles and a domain authority of 35, implemented the CRAFT framework between February and May 2026:
Baseline (February):
- 15 articles cited in AI Overviews per month;
- 0.3 citations per article;
- Entity Authority Score (measured with custom signals): 4.2/10.
Implementation (February-May):
- Structural optimization of 30 existing articles (phase 1);
- Creating 12 new articles with original research and the CRAFT framework;
- Development of detailed author profiles for 5 main contributors;
- Full implementation of schema markup;
- Monthly update of the top 20 articles.
Results (May post-core update):
- 51 articles cited in AI Overviews per month (+240%);
- An average of 1.25 citations per article (+317%);
- Entity Authority Score: 7.1/10 (+69%);
- Estimated synthetic traffic (mentions in AI Overviews): +1,801 TP3T compared to traditional organic traffic, -151 TP3T (but total traffic +651 TP3T).
The paradox illustrated here is critical: traditional organic ranking traffic is descended (common in 2026 for sites not in vertical media/finance), but synthetic traffic from AI citability has grown exponentially, resulting in an overall net gain.
The Limits of AI Citability and Gimmicks to Avoid
The CRAFT framework is based on solid principles, not on temporary tactics. However, it is critical to understand what It doesn't work. in 2026:
❌ Keyword Stuffing with “AI-friendly keywords”: Some SEOs believe that stuffing articles with terms like “AI Mode,” “Citerability,” and “Query Fan-Out” improves AI citability. This is false. AI systems evaluate semantic coherence, not keyword density.
❌ Structural Clickbait Heading Headlines like “Shocking: How Google Uses AI Overviews” can boost CTR, but they damage AI extractability because the heading doesn't reflect the content below.
❌ Generation of False “Original Data”: Making up statistics (“73% of Italian marketers use AI”) is quickly identified by AI models through cross-checking and by identifying contradictions with other sources. It is counterproductive.
❌ Optimization Exclusively for AI, Ignoring Human Users AI overviews depend on end-user satisfaction. If your content is unreadable to humans but “crawlable” by AI, users will bounce quickly, dampening the ranking signal.
❌ Fake Updates: Modifying the update date without changing the content is a negative signal that AI models can easily detect by comparing content hashes.
Real citability emerges from authentically useful content, structured for clarity, supported by original research, attributed to credible entities, and kept up-to-date.
Integration with Complementary Strategies in 2026
AI citability isn't the only strategic axis. To maximize overall visibility, integrate the CRAFT framework with:
Social Signals (see Social Search vs Google): The 30% user group searches TikTok and Instagram first. Adapt your CRAFT insights into short-form and narrative formats to cover both channels.
2. Community Engagement (see Community Management 2026): AI overviews give weight to authentic engagement signals. Build micro-communities around your content on Threads, Discord, or private platforms.
3. Voice Search Optimization (see Voice Search Optimization 2026): Voice devices rely on AI Overviews and synthesis. Conversational FAQs and direct answers are increasingly relevant.
4. Entity Authority Building (see Entity Authority 2026): Parallel the content's quotability with building author/organization entity authority. Publish guest posts, participate in podcasts, get mentions in authoritative media.
FAQ
How do I measure my current quotability in AI Overviews?
Use Google Search Console to track impressions in AI Mode (Insights > Performance > Filter by “AI Mode” section). For traditional AI Overviews, manually monitor by searching your keywords on Google and checking if you are cited in the AI summary, or use tools like Semrush (AI section, introduced in 2026) or Ahrefs. For a custom solution, use the script described in the article at GEO Monitoring with Claude and Replit, which integrates programmatic research in Google Search with citation tracking.
Does the CRAFT framework completely replace traditional SEO?
No. Traditional SEO (link building, domain authority, technical SEO) remains critical for organic rankings. CRAFT is a complementary layer optimized specifically for AI citability. Sites that win in 2026 do both: traditional SEO + CRAFT. As analyzed in Zero-Click Permanent and AI Overview Citations, ...synthetic traffic now accounts for 25–35% of total traffic for news publishers; ignoring it means leaving a third of potential visibility on the table.
How long does the CRAFT framework take to show results?
The timeline is variable. An article optimized according to CRAFT can start being cited in AI Overviews within 2-4 weeks of publication or optimization, provided the site has a minimum of established authority. However, the maximum impact emerges over 2-3 months, as schema markup is indexed, entity authority signals accumulate, and frequent updates signal continuous relevance to AI systems. As seen in the Italian case study, significant improvements usually require 3 months of consistent implementation across 20-30 articles.
Should I completely rewrite my existing articles?
No. Most articles benefit from a light restructuringAdding clear headings, inserting a comparison table, emphasizing key concepts with <strong>, date update and changelog. A complete rewrite is only necessary if the article is radically outdated (e.g., a 2023 article on a Google Core Update that has never been updated). Prioritize the 20 articles with the most organic traffic and start with those.
How can I verify that my schema markup is correct?
Use three tools in series: (1) Google Rich Results Test (https://search.google.com/test/rich-results) — upload the URL and verify that schema is recognized without errors; (2) Schema.org Validator (https://validator.schema.org) — upload raw HTML code and check compliance with specifications; (3) Google Structured Data Testing Tool — less used today but still useful for detailed debugging. If you use WordPress 7.0+, plugins like Yoast SEO and RankMath include integrated validators in the block editor. Test a representative article before rolling out to all articles.
Conclusion: Real Citability as the Foundation for Sustainable Visibility in 2026
The May 2026 Core Update it has definitively tipped the scales: real citability in AI Overviews, AI Mode, and alternative search platforms is no longer an optional bonus, but a strategic necessity for publishers and brands with global visibility ambitions.
The CRAFT framework — Clarity, Research, Attribution, Fact-Checking, Topic Differentiation It offers a concrete, tested, and repeatable operating model for building this citability without resorting to temporary tactics or gimmicks.
The implementation steps are clear: (1) current citability audit, (2) structural optimization of flagship content, (3) creation of new content with CRAFT, (4) complete schema markup configuration, (5) continuous monitoring and iteration.
As seen in the case study, the result is an exponential increase in synthetic visibility—traffic that emerges not from ranking position, but from AI citation and synthesis. By 2026, this is the new dominant discovery channel for informational content and B2B research.
For further information, consult the related articles on Query Fan-Out and 500+ Ranking Factors, Entity Authority in 2026, E Testing AI Mode with Search Console API. Real citability is not a hypothesis; it is the new foundation of modern visibility.
