Traffic Source Analysis

Traffic Source Analysis in Traditional SEO vs. Generative Engine Optimization (GEO) represents the systematic examination and categorization of visitor origins to digital properties, serving as a critical diagnostic tool for understanding user acquisition patterns across both conventional search engines and AI-powered answer engines ¹. The primary purpose of this analysis is to quantify the shifting balance between conventional organic search traffic and emerging generative AI referrals, enabling organizations to allocate resources strategically across both optimization paradigms ¹⁴. This matters profoundly because generative engines fundamentally alter user behavior by providing synthesized answers directly within the search interface, potentially reducing click-through rates to source websites while simultaneously creating new attribution challenges and traffic patterns that traditional analytics frameworks struggle to capture ⁴⁶.

Overview

The emergence of Traffic Source Analysis in the SEO/GEO context reflects a fundamental transformation in how users discover and consume information online. Historically, traffic source analysis focused primarily on distinguishing between organic search traffic from conventional search engines like Google and Bing, direct visits, referral traffic, social media, email campaigns, and paid search ¹. The fundamental principle involved tracking UTM parameters, referrer headers, and session data to attribute user sessions to specific acquisition channels.

However, the introduction of AI-powered generative engines—including ChatGPT, Google's Search Generative Experience (SGE), and Bing Chat—has created a paradigm shift where large language models synthesize information from multiple sources to generate comprehensive answers, often without requiring users to click through to original sources ¹⁶. This evolution addresses the fundamental challenge of understanding "zero-click searches" in a new context—not just featured snippets that prevent clicks, but AI-generated responses that aggregate information without clear attribution ⁴. Research indicates that generative engines can reduce website visibility by up to 18-64% for certain query types, fundamentally altering traffic patterns and requiring new measurement frameworks that account for both visible traffic and invisible influence on AI training data and response generation ¹⁴.

The practice has evolved from simple channel attribution to understanding the complex interplay between AI intermediation and direct website access, necessitating sophisticated tracking mechanisms that can capture generative engine referrals alongside traditional organic search traffic ⁵⁶.

Key Concepts

Traditional Organic Search Traffic

Traditional organic search traffic represents the baseline component of traffic source analysis, measured through platforms like Google Search Console, Bing Webmaster Tools, and analytics platforms ³. This encompasses traffic from conventional search engines where users click on blue links in search engine results pages (SERPs) to visit websites directly.

Example: A healthcare website publishes an article about diabetes management that ranks #3 in Google for the query "how to manage type 2 diabetes." Through Google Search Console, the site tracks 1,200 monthly clicks from this ranking position, with users arriving directly at the article page. The analytics show an average session duration of 3 minutes and a 15% conversion rate to newsletter signups, providing clear attribution from the organic search channel to measurable outcomes.

Generative Engine Referral Traffic

Generative engine referral traffic represents an emerging component requiring new tracking mechanisms, including direct referrals from ChatGPT's browsing feature, Bing Chat citations, Google SGE link clicks, and other AI-powered search interfaces ⁵⁶. Unlike traditional referrals, these sources may provide limited referrer data, making attribution challenging.

Example: The same healthcare website notices an unusual traffic pattern where 300 monthly visitors arrive with referrer strings containing "chat.openai.com" or appear as direct traffic but exhibit behavior patterns consistent with AI-referred users (immediately scrolling to specific sections, spending time on technical details). By implementing custom tracking parameters and analyzing user behavior, the site identifies these as ChatGPT referrals where users asked the AI about diabetes management and clicked through to verify or explore deeper information.

Citation Visibility

Citation visibility refers to how frequently a source is referenced in AI responses, regardless of whether those citations generate direct traffic ¹. This metric represents a new form of brand exposure and authority that may influence user behavior without creating immediately trackable referral patterns.

Example: A financial advisory firm systematically queries ChatGPT, Claude, and Bing Chat with 50 investment-related questions relevant to their expertise. They discover their content is cited in 12 of these responses (24% citation rate), compared to their main competitor's 8 citations (16%). However, their analytics show only 45 direct referrals from AI platforms monthly, suggesting significant "dark funnel" influence where users encounter their brand through AI but visit later through other channels like direct navigation or branded search.

Multi-Touch Attribution Modeling

Multi-touch attribution modeling in the SEO/GEO context accounts for users who first encounter information through AI interfaces before later visiting websites directly or through traditional search ¹. This approach assigns fractional credit across touchpoints, accounting for the influence of AI visibility even when direct referral tracking is incomplete.

Example: An e-commerce retailer selling outdoor equipment implements advanced attribution tracking and discovers that 22% of customers who eventually purchase hiking boots follow this journey: first encounter the brand through a ChatGPT recommendation when asking "best waterproof hiking boots for beginners," then search for the brand name directly three days later, and finally convert through a Google organic search for "brand name hiking boots sale" two weeks after initial exposure. Traditional last-click attribution would credit only the final organic search, missing the AI's initiating role.

Zero-Click AI Answers

Zero-click AI answers represent queries where generative engines provide synthesized responses that fully satisfy user intent without requiring clicks to source websites ⁴. This phenomenon extends beyond traditional featured snippets to comprehensive AI-generated explanations that aggregate information from multiple sources.

Example: A cooking website that previously received 5,000 monthly visits from the query "how long to bake chicken breast at 375 degrees" sees traffic from this query drop by 68% after Google SGE begins providing a direct answer: "Bake chicken breast at 375°F for 20-30 minutes, until internal temperature reaches 165°F." Users receive complete information without clicking, though the site's content may have informed the AI's training data or real-time retrieval without attribution.

Intent-Based Traffic Segmentation

Intent-based traffic segmentation categorizes traffic sources by user intent rather than just channel, recognizing that the same user might employ different search methods for different needs ¹. This framework distinguishes between quick fact-checking (predominantly AI), research and comparison (mixed traditional and AI), and ready to purchase (still predominantly traditional search).

Example: A software company analyzes their traffic and discovers distinct patterns: informational queries like "what is customer relationship management" increasingly come from AI referrals (65% of traffic for these terms), while comparison queries like "Salesforce vs HubSpot features" show mixed sources (40% traditional organic, 35% AI referrals, 25% direct), and transactional queries like "buy CRM software small business" remain dominated by traditional organic search (78%). This insight leads them to create AI-optimized educational content while maintaining traditional SEO focus for commercial pages.

Competitive Citation Analysis

Competitive citation analysis involves systematically querying AI engines with relevant topics and documenting which sources receive citations, creating "share of voice" metrics for AI responses analogous to traditional SERP visibility tracking ¹. This methodology reveals whether optimization efforts are improving AI visibility even when direct traffic attribution is unclear.

Example: A legal services firm creates a monitoring system that queries ChatGPT, Claude, Perplexity, and Bing Chat weekly with 30 questions about estate planning. They track that their content receives citations in 18% of responses, while their top three competitors receive 25%, 14%, and 9% respectively. After implementing GEO optimizations—adding structured data, improving content authoritativeness, and creating more comprehensive guides—their citation rate increases to 28% over three months, even though direct AI referral traffic only increases modestly from 120 to 180 monthly visits.

Applications in Digital Marketing Strategy

Content Strategy Development and Optimization

Traffic source analysis directly informs content strategy by revealing which content types generate traditional organic traffic versus AI citations ¹. Organizations use this data to decide whether to optimize existing content for better GEO performance, create parallel content designed specifically for AI synthesis, or maintain focus on traditional SEO approaches.

A technology publication analyzes six months of traffic data and discovers that their in-depth tutorial content (5,000+ word guides) generates strong traditional organic traffic (averaging 2,500 visits per article monthly) but receives minimal AI citations. Conversely, their concise definition articles (800-1,200 words) generate lower direct traffic (averaging 400 visits monthly) but appear in AI responses 3x more frequently. Based on this insight, they develop a dual content strategy: maintaining comprehensive tutorials optimized for traditional search while creating a new series of authoritative, citation-worthy definition pages specifically structured for AI extraction, using clear headings, bulleted key points, and Schema.org markup ³.

Resource Allocation and Budget Planning

Organizations observing shifting traffic patterns use source analysis to rebalance investments between conventional SEO tactics and GEO strategies ¹⁴. This application becomes critical when traditional organic traffic declines while generative engine referrals increase, requiring sophisticated modeling to estimate the full value of AI visibility.

An online education platform notices that organic traffic from informational queries has declined 32% year-over-year, while AI referral traffic has increased from negligible to 8% of total traffic. However, users from AI referrals show 40% higher course enrollment rates than traditional organic visitors. They calculate that despite lower volume, AI referral traffic generates $45,000 monthly revenue compared to $180,000 from traditional organic (which has declined from $265,000). Based on this analysis, they reallocate 25% of their link-building budget ($15,000 monthly) to GEO initiatives including content restructuring, expert author profiles, and citation-worthy research publication, while maintaining core technical SEO investments.

User Experience and Conversion Path Optimization

Traffic source analysis reveals behavioral differences between users arriving from traditional search versus generative engines, necessitating different landing page strategies and conversion paths ¹. Users from AI referrals often arrive seeking deeper information or verification after receiving preliminary answers.

A B2B software company discovers through behavior analysis that traditional organic search visitors typically enter on product pages and navigate to pricing (conversion rate: 3.2%), while AI referral visitors enter on educational content, spend 65% more time on site, visit an average of 4.8 pages including case studies and documentation, and convert at 5.7%. They optimize for this pattern by adding prominent "See Implementation Examples" and "Read Customer Stories" calls-to-action on educational pages where AI traffic enters, creating a guided journey that acknowledges users' existing baseline knowledge from AI interactions. This optimization increases AI referral conversion rates to 7.1%.

Competitive Intelligence and Market Positioning

Comparative traffic source analysis reveals which competitors dominate AI citations versus traditional rankings, exposing market dynamics and strategic vulnerabilities ¹. Organizations use this intelligence to identify differentiation opportunities and inform competitive response strategies.

A financial services firm conducts comprehensive competitive analysis across both traditional SEO and GEO dimensions. They discover that while Competitor A dominates traditional rankings (appearing in top 3 positions for 65% of target keywords), Competitor B has established stronger authority signals that AI engines prefer (receiving citations in 42% of AI responses versus Competitor A's 28% and their own 19%). Further analysis reveals Competitor B has published original research studies, maintains detailed author credentials, and uses extensive structured data. The firm responds by launching a quarterly research initiative, enhancing author bios with professional certifications, and implementing comprehensive Schema markup ³, resulting in citation rate increases to 31% within six months.

Best Practices

Implement Comprehensive Multi-Platform Tracking

Organizations should deploy tracking mechanisms that capture both traditional organic search and generative engine referrals across multiple platforms, including Google Analytics 4, Search Console, and emerging GEO analytics tools ¹⁵. The rationale is that generative engines often provide minimal referrer information, requiring custom tracking configurations to properly attribute traffic.

Implementation Example: A media company implements a comprehensive tracking system that includes: (1) Google Analytics 4 with custom dimensions configured to capture AI referrer patterns, including regular expression filters that identify chat.openai.com, bing.com/chat, and other AI platforms; (2) server-side tracking that captures referrer data missed by client-side analytics due to ad blockers; (3) UTM parameters embedded in all externally shared content; and (4) custom event tracking that flags sessions exhibiting AI-referral behavior patterns (specific scroll depths, time-on-page thresholds, section-specific engagement). This system reveals that 14% of traffic previously categorized as "direct" actually originates from AI platforms, changing their understanding of channel performance and informing a $50,000 reallocation toward GEO initiatives.

Establish Baseline Metrics and Monitor Trends Over Time

Practitioners should establish pre-GEO traffic patterns as control groups, then track deviations as generative engines gain market share, segmenting historical data by query intent categories ¹⁴. This approach enables organizations to distinguish between normal traffic fluctuations and fundamental shifts caused by AI adoption.

Implementation Example: An e-commerce retailer creates an automated dashboard that segments their 18 months of historical traffic data (pre-dating significant AI search adoption) by query intent: informational (how-to guides, product education), navigational (brand searches), and transactional (product purchases, comparisons). They establish baseline conversion rates and traffic volumes for each segment, then implement weekly monitoring that flags deviations exceeding 15% from baseline patterns. Within three months, they identify that informational query traffic has declined 28% while transactional query traffic remains stable, leading them to shift content investment from general educational content toward detailed product comparisons and buying guides that maintain traditional search value.

Develop Proxy Metrics for GEO Performance

Since direct traffic attribution from AI platforms remains incomplete, organizations should develop proxy metrics including citation frequency, brand mention volume in AI responses, and share of voice across multiple AI platforms ¹. These indicators reveal GEO progress even when referral traffic is difficult to track.

Implementation Example: A healthcare information provider creates a systematic monitoring program that queries ChatGPT, Claude, Perplexity, Bing Chat, and Google SGE with 100 health-related questions monthly, documenting citation frequency, citation context (primary source vs. supporting reference), and competitive presence. They track these metrics alongside traditional traffic data, discovering that citation rate increases from 15% to 34% over six months correlate with a 12% increase in branded search volume and a 23% increase in direct traffic, even though identifiable AI referral traffic only accounts for 3% of total visits. This validates their GEO investment by revealing indirect traffic impacts that traditional attribution misses.

Implement Structured Data and Entity Optimization

Organizations should enhance content with Schema.org markup and entity-based optimization to improve both traditional search visibility and AI citation likelihood ³⁶. Structured data helps AI systems extract and attribute information more accurately while maintaining traditional SEO benefits.

Implementation Example: A recipe website implements comprehensive Schema markup across 2,000 recipes, including Recipe schema with detailed nutrition information, author credentials with Person schema, and organization information with Organization schema. They also create entity-optimized content that clearly defines ingredients, techniques, and equipment with consistent terminology. After implementation, they track a 15% increase in traditional organic traffic (due to enhanced SERP features), a 28% increase in AI citation frequency (AI systems more easily extract structured information), and a 40% increase in identifiable AI referral traffic. The structured approach benefits both optimization paradigms simultaneously, maximizing return on content investment.

Implementation Considerations

Tool Selection and Analytics Configuration

Traditional analytics platforms like Google Analytics 4 weren't designed for the GEO era and struggle to properly categorize some generative engine referrals ¹⁵. Organizations must supplement standard analytics with specialized tools and custom configurations to capture the full picture of traffic sources.

Practitioners should implement custom channel groupings in Google Analytics 4 that distinguish between conventional Bing organic traffic and Bing Chat referrals, or between standard Google search and SGE-originated visits. This requires creating regular expression patterns that identify AI platform referrers, configuring custom dimensions to capture additional context, and implementing server-side tracking to capture data that client-side analytics miss due to ad blockers or privacy settings. Organizations should also evaluate emerging GEO monitoring platforms that track AI citation frequency, custom data pipelines that parse server logs for AI bot activity, and API integrations that pull data from multiple sources into unified dashboards. Regular audits of channel classification rules ensure new AI referrer patterns are properly categorized rather than lumped into "direct" or "other" categories.

Audience-Specific Customization and Segmentation

Different user segments adopt AI search tools at varying rates based on demographics, technical sophistication, and use cases ¹. Organizations must segment their traffic analysis by audience characteristics to understand adoption patterns and optimize accordingly.

A professional services firm discovers through cohort analysis that their younger audience (ages 25-34) generates 22% of sessions from AI referrals, while their older audience (ages 55+) generates only 3% from AI sources. They also identify that mobile users show higher AI adoption (18% of mobile traffic) compared to desktop users (9% of desktop traffic). Based on these insights, they create audience-specific optimization strategies: maintaining traditional SEO focus for content targeting older professionals while implementing aggressive GEO optimization for content targeting younger audiences, and ensuring mobile-optimized content is particularly well-structured for AI extraction. This segmented approach prevents over-investing in GEO for audiences still primarily using traditional search while capturing opportunities with early AI adopters.

Organizational Maturity and Phased Implementation

Organizations at different maturity levels require different approaches to traffic source analysis in the SEO/GEO context ¹. Starting with pilot programs and incremental testing allows organizations to build capabilities while managing risk.

A mid-sized retailer with limited analytics resources begins with a phased approach: Phase 1 (months 1-2) involves implementing basic AI referrer tracking through Google Analytics 4 custom dimensions and establishing baseline metrics for 20 high-priority product categories. Phase 2 (months 3-4) adds competitive citation monitoring for their top 10 product categories, querying AI platforms weekly to track share of voice. Phase 3 (months 5-6) implements GEO optimizations on a subset of content (100 product pages) while maintaining control groups, measuring impact on both traditional organic and AI referral traffic. This incremental approach allows them to demonstrate value, build internal expertise, and refine methodologies before scaling to their full 10,000-product catalog, avoiding the risk and resource requirements of immediate comprehensive implementation.

Privacy Compliance and First-Party Data Strategy

Data privacy regulations including GDPR and CCPA limit tracking capabilities, particularly for users who opt out of cookies, while the deprecation of third-party cookies further constrains attribution ¹. Organizations must balance comprehensive tracking with privacy compliance and prepare for a cookieless future.

Best practices include implementing consent management platforms that respect user preferences while maximizing permissible data collection, focusing on aggregate trend analysis rather than individual user tracking where possible, and being transparent about data usage in privacy policies. Organizations should build robust first-party data strategies that don't rely on third-party cookies, including authenticated user experiences that enable tracking across sessions with user consent, server-side tracking that captures referrer data without client-side cookies, and probabilistic modeling that infers traffic patterns from aggregate data. A publishing company implements a registration system offering free access to premium content in exchange for account creation, enabling them to track 65% of their audience across sessions with explicit consent while using aggregate analysis for non-authenticated traffic, maintaining comprehensive traffic source insights while respecting privacy regulations.

Common Challenges and Solutions

Challenge: Attribution Complexity and Dark Funnel Tracking

Generative engines often provide minimal referrer information, causing users to arrive as "direct" traffic after copying URLs from AI responses, or users consume AI-synthesized content before visiting websites through other channels days or weeks later ¹⁵. This "dark funnel" where AI exposure influences later conversion behavior without clear attribution trails makes it difficult to measure the true impact of GEO efforts and justify resource allocation.

Solution:

Implement multi-layered attribution strategies that combine direct tracking with proxy metrics and behavioral analysis. Deploy UTM parameters in all external content and use link shorteners with tracking for citations in AI responses when possible. Configure first-party tracking that can identify users who previously interacted with AI-generated content containing your brand, even when direct referral data is missing. Implement brand lift studies that survey customers about how they discovered your organization, specifically asking about AI tool usage. Use statistical modeling to correlate increases in AI citation frequency with changes in branded search volume, direct traffic, and conversions, establishing probabilistic attribution even when deterministic tracking fails. A SaaS company implements this comprehensive approach and discovers that while only 4% of traffic shows direct AI referrals, their brand lift surveys reveal 18% of new customers first encountered their brand through AI tools, validating their GEO investment despite incomplete direct attribution.

Challenge: Lack of Industry Benchmarks and Standards

Unlike traditional SEO where industry benchmarks for click-through rates, rankings, and traffic exist, GEO performance standards remain undefined ¹⁴. Organizations struggle to determine whether their AI citation rates, referral traffic volumes, or conversion rates from AI sources represent strong or weak performance, making it difficult to set realistic goals and evaluate success.

Solution:

Create internal benchmarks by tracking your own performance over time, establishing baseline metrics and monitoring trends rather than comparing to external standards that don't yet exist. Participate in industry surveys and studies that aggregate anonymized data, contributing your metrics while gaining access to comparative insights. Join professional communities and peer groups where practitioners share GEO performance data in confidential settings. Focus on directional trends and relative performance shifts rather than absolute numbers—a 25% increase in AI citation frequency over six months represents progress regardless of whether the absolute rate is "good" by undefined external standards. Segment benchmarks by content type, query intent, and industry vertical, recognizing that informational health content may achieve 40% AI citation rates while transactional e-commerce content may only reach 8%, both representing strong performance in their respective contexts. A marketing agency creates an internal benchmarking system tracking 15 clients across different industries, identifying that B2B software clients average 22% AI citation rates while consumer retail clients average 9%, providing context-appropriate performance expectations.

Challenge: Tool Limitations and Platform Fragmentation

Traditional analytics platforms struggle to properly categorize generative engine referrals, while the proliferation of AI platforms (ChatGPT, Claude, Perplexy, Bing Chat, Google SGE, and emerging competitors) creates fragmentation that makes comprehensive tracking difficult ¹⁵⁶. Each platform may provide different referrer data, update at different frequencies, and require different monitoring approaches.

Solution:

Build a hybrid analytics infrastructure that combines traditional platforms with custom solutions and emerging specialized tools. Supplement Google Analytics 4 with server-side tracking that captures referrer data missed by client-side implementations. Develop custom data pipelines using tools like Google BigQuery or AWS that aggregate data from multiple sources—web analytics, server logs, Search Console, and GEO monitoring platforms—into unified dashboards. Implement regular expression libraries that identify and properly categorize referrers from all major AI platforms, updating these patterns monthly as new platforms emerge and existing ones evolve. Consider emerging specialized GEO analytics tools that automate citation tracking across multiple AI platforms, while maintaining realistic expectations about coverage gaps. Create manual monitoring protocols for strategic content, where team members systematically query AI platforms to supplement automated tracking. A media company builds a custom data pipeline that combines Google Analytics 4, server log analysis, and a specialized GEO monitoring tool, achieving 85% coverage of AI referral traffic compared to 40% coverage from Google Analytics alone, while accepting that some attribution gaps will remain until the industry develops more mature solutions.

Challenge: Volatile AI Engine Behavior and Model Updates

AI platforms frequently update their underlying models, retrieval mechanisms, and citation logic, causing citation rates and referral traffic to fluctuate unpredictably ¹⁶. A source that receives frequent citations one month may see dramatic declines after a model update, making it difficult to distinguish between optimization failures and platform changes beyond your control.

Solution:

Maintain long-term trend perspectives rather than reacting to short-term fluctuations, using rolling 90-day averages to smooth volatility. Implement change detection systems that monitor multiple AI platforms simultaneously—if citation rates decline across all platforms, it likely indicates a problem with your content, but if declines occur on only one platform, it suggests a platform-specific change. Subscribe to official announcements and developer communities for major AI platforms to stay informed about model updates and feature changes. Create content resilience by optimizing for fundamental quality signals (authoritativeness, accuracy, comprehensiveness, clear structure) that should maintain value across model updates rather than gaming specific platform behaviors. Diversify across multiple AI platforms rather than over-optimizing for a single engine, similar to how traditional SEO best practices recommend not over-optimizing exclusively for Google. Document correlation between known platform updates and traffic changes to build institutional knowledge about typical impact patterns. A financial services firm implements a monitoring dashboard that tracks citation rates across five AI platforms with annotations for known model updates, revealing that their content maintains stable aggregate performance despite individual platform volatility, validating their focus on fundamental quality over platform-specific optimization.

Challenge: Balancing SEO and GEO Resource Allocation

Organizations face difficult decisions about how to balance investments between traditional SEO (which still drives the majority of traffic for most sites) and GEO (which represents future opportunity but with uncertain ROI) ¹⁴. Over-investing in GEO may sacrifice current traffic and revenue, while under-investing may leave organizations unprepared for continued AI adoption.

Solution:

Implement portfolio-based resource allocation that maintains diversified investments across both paradigms rather than making binary choices. Start with pilot programs targeting specific content categories where GEO shows strongest potential (typically informational content in industries where AI adoption is high), measuring impact before scaling. Use incremental testing where GEO optimizations are applied to subsets of content while control groups maintain traditional optimization only, enabling direct comparison of results. Develop dual-purpose optimizations that benefit both SEO and GEO simultaneously—comprehensive, authoritative content with clear structure, expert authorship, and Schema.org markup ³ improves traditional rankings while increasing AI citation likelihood. Create decision frameworks based on content type and business goals: prioritize traditional SEO for high-commercial-intent transactional content that still drives direct conversions, while investing more heavily in GEO for informational content where AI adoption is reducing traditional click-through rates. Monitor the ratio of traditional organic to AI referral traffic monthly, gradually shifting resource allocation as the balance evolves, but maintaining minimum viable investment in traditional SEO given its continued dominance. A publishing company allocates 70% of optimization resources to traditional SEO, 20% to dual-purpose optimizations, and 10% to GEO-specific initiatives, reviewing this allocation quarterly based on traffic source trends and adjusting gradually rather than making dramatic shifts based on short-term data.

References

1. Semrush. (2024). Generative Engine Optimization. https://www.semrush.com/blog/generative-engine-optimization/
2. Ahrefs. (2024). SEO Statistics. https://ahrefs.com/blog/seo-statistics/
3. Google Developers. (2025). Introduction to Structured Data. https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data
4. Search Engine Land. (2024). AI Overviews Impact on Organic Search Traffic. https://www.searchengineland.com/ai-overviews-impact-organic-search-traffic-446128
5. Bing Webmasters. (2024). Bing Chat and Your Website FAQ. https://www.bing.com/webmasters/help/bing-chat-and-your-website-faq-8c7e6c4e
6. Google Blog. (2024). Generative AI in Search. https://blog.google/products/search/generative-ai-search/