GEO, AI Overviews & Content Strategy with AI: Complete 2026 Guide

50% of consumers now use AI search as their primary source for information. 44% prefer it to traditional search.

This isn't a prediction. It's McKinsey data from October 2025, validated across all age groups. While the industry debates whether SEO is dead—for the hundredth time—the shift has already happened. The question is no longer whether to adapt, but how.

In this article, I analyze three interconnected phenomena reshaping online visibility in 2026: Generative Engine Optimization (GEO), the impact of Google's AI Overviews on CTR, and content strategy with AI after two years of increasingly aggressive core updates. With concrete data, not buzzwords.

GEO: what is Generative Engine Optimization and why it matters

Generative Engine Optimization is the practice of structuring content and online presence to improve visibility in responses generated by AI systems: ChatGPT, Gemini, Perplexity, Copilot, Claude. Unlike traditional SEO, which aims for SERP rankings, GEO aims to be cited, referenced, or quoted within AI responses.

Where the term comes from

The term was formally introduced in an academic paper from November 2023 by Pranjal Aggarwal and colleagues from Princeton, Georgia Tech, Allen Institute of AI, and IIT Delhi. The paper was published at KDD 2024, the major international conference on data mining. It's the universally cited academic reference.

Related terminology—AEO (Answer Engine Optimization), LLMO (Large Language Model Optimization), AI SEO— refers to the same concept. As Lily Ray noted in her Substack, 2025 was "the year of GEO", with the familiar "SEO is dead" refrain resurrected to capitalize on fear and uncertainty.

Princeton study data: what actually works

The paper tested 9 optimization strategies. The three most effective:

1. Adding statistics and data: up to +41% visibility (Position-Adjusted Word Count metric)
2. Citing authoritative sources: +31.4% when combined with other techniques
3. Including direct quotes: +28% (Subjective Impression metric)

The best combination—fluency optimization plus data addition— outperformed any single strategy by 5.5%. Overall, GEO methods can increase visibility by up to 40%.

How AI engines select sources

Each AI platform has its own way of selecting and citing content. Understanding these differences is essential for effective GEO strategy.

The most significant data: the same brand can see citation volumes that differ 615-fold across platforms. GEO must be platform-aware.

An important evolution concerns Google AI Overviews sources. In July 2025, approximately 76% of citations came from pages in the top 10 organic results. In February 2026, this percentage dropped to 38%. Google decomposes queries into sub-queries (query fan-out) and pulls content from sources that may not rank well in traditional SERPs.

The 10 confirmed GEO ranking factors

1. Semantic completeness: the #1 factor. Content with 8.5/10+ scores are 4.2x more likely to be cited. Ideal passages are 134-167 words, self-contained.
2. Structured schema markup: pages with schema have 2.8x more AI citations. With FAQ schema, 3.2x more likely to appear in AI Overviews. Google in May 2025: "Structured data is critical for modern search features."
3. Statistics and citations in content: up to +41% visibility (Princeton study).
4. E-E-A-T signals: authors with transparent bios, citations from authoritative sources, consistent updates.
5. Third-party mentions and brand signals: LLMs heavily weight external sources. PR becomes an "external validation layer" for GEO.
6. Content freshness: LLMs prioritize the most recent version. Pages with "Last updated" timestamps perform better.
7. Multimodal content: pages combining text, images, video, and structured data see +156% selection rate. With complete schema: +317%.
8. Topical authority and content clusters: clusters beat isolated posts. Pillar pages linked to sub-topics with internal links.
9. Entity clarity: clear language about entities, schema defining who you are and what the page represents.
10. Quotability: answer-first format, direct definitions, evidence with bullet points. Definitive answer in first 50 words.

Google AI Overviews: rollout status and real impact

The timeline: from SGE to AI Overviews to AI Mode

Google launched the Search Generative Experience (SGE) in May 2023 as an experiment. In May 2024, rebranded as AI Overviews with official launch in the US. In March 2025, expansion to 9 EU countries including Italy. In July 2025, AI Mode—the conversational version—became available globally. By March 2026: AI Mode exceeded 100 million active users and 1 billion monthly queries.

How widespread are AI Overviews

Data varies based on keyword sample analyzed, but the trend is clear: steady growth with fluctuations.

Sector expansion has been massive. According to BrightEdge, from February 2025 to February 2026 the percentage of queries with AIO went from 44% to 88% in Healthcare (+100%), from 18% to 83% in Education (+361%), and from 10% to 78% in Food & Restaurants (+680%).

Italy and Europe: rollout with brakes

In Italy, AI Overviews have been active since March 25, 2025, along with Austria, Belgium, Germany, Ireland, Poland, Portugal, Spain, and Switzerland. But with important limitations: available only for logged-in users 18+, due to GDPR, DMA, and DSA. EU expansion is slower than in the US.

The European Commission opened an antitrust investigation on December 9, 2025 into Google's use of web content for training AI and generating AI Overviews. The regulatory environment could further slow expansion in the European market.

The CTR impact: numbers to reflect on

Here the data is unambiguous. When AI Overviews appear, click-through rates plummet.

A data point worth attention: even on queries without AI Overviews, organic CTR fell 41%. The decline is structural, not just tied to AIOs.

Zero-click searches—where the user doesn't click any result— went from 56% to 69% between May 2024 and May 2025. With AI Overviews, the rate rises to 83%. Less than 1% of users click links inside AI responses.

Which queries are most affected

AI Overviews don't appear on all searches. Informational queries are hardest hit (57% with AIO), but the most concerning growth is in commercial queries (8% to 19%) and transactional queries (2% to 14%). Longer queries (7+ words) show AIO in 46% of cases, versus 9.5% for single-word queries.

Winners and losers

The losers: publishers and content sites. Organic traffic to news sites dropped from 2.3 billion monthly visits (mid-2024) to under 1.7 billion (May 2025): 600 million visits lost per month. Forbes and HuffPost lost 50% of traffic, Business Insider lost 55%.

The winners: Reddit (+450% AI citations), Wikipedia (11.22% of all AIO citations), and YouTube (9.51%). Sites with strong brand and direct audience are more resilient.

An interesting paradox: visitors from AI platforms convert at double the rate of traditional organic visitors and spend 68% more time on sites. But volume is still marginal: AI referral traffic represents 1.08% of total.

How to be cited in AI Overviews

The overlap with top 10 organic is in steep decline: from 76% to 38% in less than a year. Ranking well in traditional SERPs helps, but isn't enough anymore. Here are the key factors for being cited:

1. Direct answer in first 40-60 words of content
2. Data density: at least one statistic every 150-200 words
3. Structured schema markup: +3x likelihood of being cited (with multimodal + schema: +317%)
4. Strong E-E-A-T: authors with names, credentials, first-hand information
5. Unique content: original research, proprietary data, insights unavailable elsewhere
6. Freshness: recent updates with visible dates

Content strategy with AI in 2026: what Google rewards and what it penalizes

Google's official position

Google doesn't penalize AI-generated content as such. The official stance is clear: focus is on content quality, not how it's produced. As John Mueller reiterated: Google's systems don't care if content is created by AI or humans— what matters is that it's useful to users.

But practice tells a more nuanced story. In fact, the last two years' core updates have hit almost exclusively sites with high-volume low-quality AI content.

The March 2024 Core Update: first major impact

The stated goal was to reduce unhelpful content in results by 40%. The results: over 800 sites completely deindexed in the first phase. An Originality.ai study found that 100% of deindexed sites showed signs of AI content, and 50% had over 90% of posts AI-generated. Impact: 21 million monthly visits lost and $440,000 in ad revenue vanished.

The December 2025 Core Update: E-E-A-T everywhere

The latest significant core update extended E-E-A-T requirements beyond YMYL topics to practically all competitive searches. AI-generated content at scale without expert supervision: traffic drops up to 60-95%. But—and this is crucial— AI-assisted content with expert oversight remained stable or improved.

The distinction that matters: AI-generated vs AI-assisted

The line isn't between human and AI. It's between content produced at scale without added value and content where AI is a tool in the creative process. The Search Quality Rater Guidelines updated in January 2025 explicitly instruct quality raters to assess whether content is AI-generated and assign "lowest" ratings if it lacks original insight or unique value.

The E-E-A-T framework applied to AI content:

• Experience: add direct experiences, personal case studies, proprietary data
• Expertise: bylines with real credentials, author photos, verifiable biography
• Authoritativeness: citations of primary sources, links to studies, mentions in external publications
• Trustworthiness: transparency on AI use, documented fact-checking, update dates

Can Google detect AI content?

Not directly—Google doesn't have a "ChatGPT detector". But SpamBrain detects manipulative patterns: repetitive introductions, generic anchor text, missing citations. Independent detection tools have accuracy ranging from 19% to 61% in real-world scenarios. At Google I/O 2025, watermarking was confirmed to be "under review" for automated detection.

The most interesting data comes from an Ahrefs study of 600,000 pages: 86.5% of pages in the top 20 contain AI-generated content, but the correlation between AI percentage and ranking position is nearly zero (0.011). Only 4.6% of top-ranking pages is entirely AI-generated. The message: AI as support doesn't penalize, AI as total replacement does.

The recommended editorial workflow in 2026

1. Research: AI for competitive analysis, keyword research, SERP analysis
2. Outline: AI for structure, human for editorial angle and unique insights
3. Draft: AI for initial draft
4. Expert review: human adds direct experience, fact-checks, inserts unique data
5. Optimization: SEO tools for semantic gaps
6. Quality assurance: fact-checking on numbers, statistics, citations, references

Information Gain: the signal that rewards originality

Google has a patent (filed 2018, granted June 2024) for Information Gain Score: measures how much a document adds new information relative to the existing corpus. The May 2024 Google API leak confirmed an OriginalContentScore and possible effortScore estimating creation effort.

In practice: copying the structure of the top result and rewriting with AI doesn't add information gain. Adding proprietary data, personal tests, original analysis, and unique perspectives does.

The big picture: AI search market numbers in 2026

Before moving to operational strategies, it's worth looking at the full picture to understand the scope of ongoing change.

The volume paradox: AI traffic grows 500%+ year-over-year but still represents just 1% of total. AI visitors convert at double rates but are still few. This is an embryonic market with explosive growth signals.

GEO vs SEO: competitors or allies?

The hottest debate in the industry is whether GEO is something radically new or simply good SEO with a different name. The answer, as often happens, is somewhere in between.

The foundations are the same: structured content, real expertise, claims backed by data, clear answers to real questions. Companies maintaining strong SEO + GEO strategy capture 60-70% of available visibility.

Lily Ray's contrarian take: "Blog and industry site articles were cited by LLMs without doing anything new or different from standard SEO best practices." Ray warns that many GEO recommendations risk "flooding search indexes with low-quality automated content."

Fishkin of SparkToro adds another important data point: according to SparkToro/Datos/Semrush data, traditional search usage remained stable throughout 2025—LLM adoption didn't erode Google usage. But McKinsey says 50% of consumers prefer AI search. Both can be right: usage is additive, not substitutive. For now.

Tools to monitor AI visibility

A new ecosystem of tools is emerging to track AI citations. Here are the main ones as of March 2026.

Tools dedicated to GEO monitoring

SEO platforms with GEO features

Lily Ray's important warning: LLM tracking tools are "fundamentally limited" because AI responses are non-deterministic and personalized. The same prompt can generate different citations at different times. Use them as trend indicators, not absolute truth.

How to adapt: operational checklist for 2026

Enough theory. Here are concrete actions to adapt your content and visibility strategy to new scenarios.

Optimizing content for GEO and AI Overviews

1. Answer the main question in first 40-60 words of each piece. AI engines extract snippets from early paragraphs.
2. Insert at least one statistic or concrete data every 150-200 words. Princeton study shows +41% visibility with added statistics.
3. Structure content with clear, hierarchical headings (H2 > H3) that work as question-answer pairs.
4. Create self-contained passages of 134-167 words that can be cited without additional context.
5. Cite primary sources with links: studies, papers, official reports. Not unsupported opinions.
6. Add proprietary data, personal tests, original case studies— this maximizes Information Gain Score.

Schema markup and structured data

1. Implement JSON-LD schema for every content type: Article, FAQPage, HowTo, Organization.
2. The data: pages with schema have 2.8x more AI citations. With FAQ schema: 3.2x. Not optional.
3. Update existing schema: BrightEdge measured +44% AI citations after updating structured data.

Brand and authority

1. Build third-party brand mentions: PR, guest posts, interviews, citations in industry publications. LLMs use these as authority signals.
2. Curate author pages: verifiable bio, real credentials, photos, links to professional profiles.
3. Create complete topic clusters: pillar page linked to sub-topics with coherent internal links. Clusters beat isolated posts.

Managing AI crawlers

1. Allow access to search AI crawlers (ChatGPT-User, PerplexityBot, Claude-SearchBot).
2. Consider blocking training crawlers (GPTBot, ClaudeBot) if you don't want your content used for training.
3. Consider llms.txt: 844,000+ sites implemented it, but no AI platform officially confirmed reading it. Risk of "cargo cult"—only implement if low cost.

Content strategy with AI

1. Use AI as assistant, not substitute. Workflow: AI for research and draft, human for insight, verification, added value.
2. Don't scale content production just because AI makes it possible. 1 in 3 sites with programmatic implementation suffer traffic collapse within 18 months.
3. Focus on uniqueness: proprietary data, first-hand experience, original analysis. This is what Google measures with Information Gain Score and what AI can't generate alone.
4. Diversify channels: reduce dependence on Google organic traffic. Newsletter, community, social, owned brand—every channel you own is one AI Overviews can't take.

Legal context: EU AI Act and AI content

Key obligations of the EU AI Act take effect August 2, 2026. For Italian content creators, practical implications: AI-generated content intended to inform the public on matters of public interest will require mandatory disclosure. For marketing and commercial content, the situation is less defined, but transparency is recommended. Monitor closely.

The Italian context: what changes for businesses and professionals

The Italian market has important specificities. AI Overviews in Italian have been active for less than a year (March 2025) and EU restrictions (logged-in users 18+ only) limit adoption. But growth signals are there.

At the Microsoft AI Tour in Milan (March 10, 2026), estimates showed AI could contribute +18% to Italian GDP by 2040: 336 billion euros annually. SEOZoom published a complete GEO guide in Italian. The ecosystem is moving.

For the Italian market specifically: ChatGPT favors Wikipedia and encyclopedic content (47.9% of primary citations)—optimizing Italian Wikipedia entries is concrete action. Perplexity cites Reddit heavily (46.7%), but Reddit usage in Italy is significantly lower than in the US: platform-specific strategies need local adaptation.

For local businesses, GEO integrates with Local SEO: optimizing your Google Business profile and local citations strengthens visibility in geolocalised AI responses.

Frequently asked questions

What is GEO and how does it differ from traditional SEO?

Generative Engine Optimization is the practice of optimizing content and online presence to be cited in responses from AI engines (ChatGPT, Gemini, Perplexity). Traditional SEO targets SERP rankings; GEO targets being quoted inside AI-generated responses. The foundations are the same (quality content, E-E-A-T, structured data), but GEO adds focus on semantic completeness, answer-first format, and multi-platform visibility.

Are AI Overviews active in Italy?

Yes, since March 25, 2025. Available for logged-in users 18+. Restrictions are due to GDPR, DMA, and DSA. EU expansion is slower than in the US, and the European Commission opened an antitrust investigation into Google's use of web content for generating AI Overviews.

Does Google penalize AI-generated content?

Not automatically. Google's official position is that focus is on quality, not production method. But in practice, 2024-2025 core updates hit almost exclusively sites with high-volume low-quality AI content. The Ahrefs study of 600,000 pages found no correlation between AI content and penalties, but only 4.6% of top-ranking pages was entirely AI-generated. The key: AI as assistant, not as total substitute.

How much traffic is lost with AI Overviews?

Data varies by study. Seer Interactive measures -61% organic CTR on queries with AIO. Ahrefs reports -58% for position #1. GrowthSRC -32%. Zero-click searches hit 83% when an AI Overview is present. But decline isn't uniform: some sectors like shopping (only 3.2% AIO) are still less affected.

Do I need to invest in GEO monitoring tools?

Depends on size and sector. For SMBs and freelancers, starting with free tools like HubSpot AI Search Grader is sufficient for initial diagnosis. Dedicated tools like Otterly.ai (from $25/month) or Rankscale (from $20/month) make sense when AI visibility becomes a strategic KPI. Caveat: AI responses are non-deterministic, so tool data should be used as trend indicators, not absolute truth.

Is traditional SEO dead?

No. SparkToro/Datos/Semrush data shows traditional search usage remained stable in 2025. LLM adoption is additive, not substitutive—for now. Organic traffic from search still represents the overwhelming majority, while AI traffic is 1.08%. The winning strategy is dual: keep SEO strong and build GEO visibility to capture 60-70% of overall visibility.

Conclusion: dual optimization is not optional

2026 is not the year SEO dies. It's the year it splits. On one side, traditional search continues generating most traffic. On the other, a new channel—AI responses—grows exponentially and attracts users who convert better, stay longer, and engage more with the sites they visit.

The good news for serious SEOs: GEO foundations are the same as well-done SEO. Structured content, concrete data, real expertise, clear answers. The difference is in details: answer-first format, self-contained snippets, advanced schema markup, third-party brand mentions, multi-platform presence.

The bad news for those who thought they'd use AI to produce content at scale: Google proved—with 800+ deindexed sites and the December 2025 update— that quantity without quality doesn't pay. AI is an extraordinary tool in the creation process, not a replacement for human value.

To dive deeper on the relationship between AI and SEO, read the complete AI and SEO 2026 guide.

For a complete overview of SEO in 2026, from technical to strategy, start with the SEO 2026 guide.

Structured data is a multiplier for GEO: learn how to implement them in the structured data and Schema.org guide.

To implement GEO keyword research for your project, contact me for personalized consulting.