The AI Citation Paradox: How Google’s AI Overviews Weaponize Self-Promotional Content Against B2B Brands

In the rapidly evolving landscape of search engine optimization (SEO), B2B brands have long relied on a reliable playbook to capture high-intent search traffic: the self-promotional "best [category] software" listicle. By publishing comprehensive, well-structured comparison guides that conveniently rank their own product as the number-one solution, software-as-a-service (SaaS) companies successfully drove organic traffic and qualified leads for over a decade.

However, a groundbreaking study by prominent SEO strategist Lily Ray reveals that Google’s generative AI search feature, AI Overviews (formerly known as Search Generative Experience, or SGE), has turned this playbook on its head.

According to Ray’s analysis of B2B software search queries, Google’s AI Overviews cited self-promotional listicles as sources of information but actively excluded the brands hosting those pages from their recommended product lists in 69% of analyzed cases. Instead of driving traffic to the content creator, Google used the brand’s own content to train its AI response, only to recommend the brand’s direct competitors to the user.

1. Main Facts: The Paradox of AI Search Citations

The core finding of Ray’s research introduces a troubling paradigm shift for digital marketers: a citation in an AI Overview is no longer synonymous with a recommendation.

For years, ranking at the top of Google’s search engine results pages (SERPs) meant capturing the lion’s share of user clicks. In the age of AI-driven search, however, Google acts as an aggregator and synthesizer of information. When a user searches for a query like "best CRM software for startups," Google’s AI model scans the web, extracts data from high-ranking pages, and presents a consolidated answer.

[Traditional Search] 
User Query -> Search Results -> Brand's Listicle -> Click to Brand Website (Conversion)

[AI Overview Search]
User Query -> AI Synthesizes Brand's Listicle -> AI Recommends Competitors -> No Click to Brand Website

Ray’s research highlights a stark reality:

• The Citation Trap: Google’s algorithms recognize the structured, informative nature of brand-created comparison pages, using them to understand market categories and product features.
• The Recommendation Gap: Despite citing the brand’s page as an information source, the AI model frequently filters out the host brand from the actual recommendations, labeling the self-ranking as biased or lacking objective consensus.
• Competitor Hijacking: In more than two-thirds of the studied queries, the competitors listed on a brand’s own self-serving listicle were the ones highlighted and recommended in the final AI Overview box.

This dynamic creates an environment where B2B brands are essentially funding and writing the research that Google uses to divert potential customers to their competitors.

2. Chronology: The Rise and Fall of the Self-Promotional Listicle

To understand how B2B marketing reached this inflection point, it is necessary to trace the evolution of search engine algorithms and the emergence of Google’s generative AI initiatives.

Timeline of Search Evolution & Self-Promotional Listicles:

2015–2022: The Golden Age of SaaS Listicles
  └─ Brands publish "Best of" lists ranking themselves #1 to capture high-intent B2B search traffic.

May 2023: Google Announces SGE (Search Generative Experience)
  └─ Generative AI answers begin testing, relying heavily on web scraping and synthesis.

March 2024: The Core Update & Helpful Content System Integration
  └─ Google penalizes low-effort, self-serving content; organic visibility for self-ranked pages drops by 30-50%.

May 2024: Google Launches AI Overviews Globally
  └─ AI Overviews roll out to the public, fundamentally altering SERP layouts and click-through dynamics.

June 2024: Lily Ray Publishes Breakthrough Study
  └─ Empirical data proves Google cites self-promotional listicles in 69% of cases while excluding the creators from recommendations.

The Golden Age of SaaS Listicles (2015–2022)

During this period, search engines rewarded comprehensive, long-form content. B2B SaaS companies discovered that "best of" keywords carried the highest conversion rates. By creating unbiased-looking comparison charts that ultimately favored their own software, brands could capture searchers at the bottom of the marketing funnel.

The Rise of Generative AI and SGE (May 2023)

At its annual I/O conference in May 2023, Google introduced Search Generative Experience (SGE). This marked the beginning of Google’s transition from a search engine that points users to external websites to an "answer engine" that synthesizes information directly on the SERP.

The March 2024 Core Update

In early 2024, Google executed a massive Core Update designed to eradicate spam and unhelpful, SEO-first content. During this update, many B2B brands that relied heavily on self-ranked listicles saw their organic search visibility drop by 30% to 50%. Google’s systems began actively identifying and devaluing pages where the publisher ranked their own product as the undisputed winner without objective, third-party validation.

The Launch of AI Overviews and Lily Ray’s Study (Spring 2024)

By May 2024, SGE was officially rebranded as AI Overviews and rolled out to hundreds of millions of users. Recognizing the profound impact this would have on B2B search, SEO researcher Lily Ray initiated a multi-month tracking study starting in mid-April to measure how AI Overviews handled B2B software queries. Her findings, published in June 2024, confirmed the industry’s worst fears: self-promotion was actively hurting brand visibility in AI-generated answers.

3. Supporting Data: Deconstructing Lily Ray’s Findings

To establish empirical proof of this shift, Lily Ray conducted a rigorous data collection process using Ahrefs Brand Radar.

Methodology

Ray tracked 100 high-value B2B "best [category] software" queries (e.g., "best email marketing software," "best project management tool"). Data was collected and analyzed across three distinct checkpoints in 2024:

• Checkpoint 1: April 15
• Checkpoint 2: May 15
• Checkpoint 3: June 8

The analysis focused on two primary metrics:

1. Did Google’s AI Overview cite the brand’s self-promotional listicle?
2. Did Google’s AI Overview include that brand in its final recommended list?

The 69% Exclusion Rate

The most striking metric from the report was the 69% exclusion rate. In nearly seven out of ten instances where a brand’s self-promotional listicle was used as a foundational source for the AI Overview, the brand itself was omitted from the AI’s final recommendations.

The Rise of Third-Party Authority and UGC

Ray’s data also revealed what Google was choosing to recommend instead of the self-promoting brands. The AI Overviews heavily favored:

• Independent Review Platforms: Sites like G2, Capterra, and TrustRadius.
• User-Generated Content (UGC): A massive, unprecedented spike in citations from Reddit and Quora.
• Established Category Leaders: Dominant, household-name brands with massive digital footprints and strong backlink profiles.

For instance, if an up-and-coming project management software company published a listicle titled "Best Project Management Software of 2024," Google’s AI Overview might cite that listicle for definitions or feature breakdowns, but recommend industry giants like Asana, Monday.com, or Trello based on sentiment analyzed from Reddit threads and G2 reviews.

4. Search Engine Philosophy and Quality Guidelines

While Google rarely comments on specific third-party SEO studies, the behavior observed in Ray’s research aligns perfectly with Google’s public documentation, patent filings, and Search Quality Rater Guidelines.

The Core of E-E-A-T: Trustworthiness

Google’s Search Quality Rater Guidelines place heavy emphasis on E-E-A-T (Experience, Expertise, Authoritativeness, and Trustworthiness). Of these, Trustworthiness is considered the most critical.

Google’s guidelines explicitly state that content should be created with a high degree of objectivity. When a company writes a review of its own product and ranks itself above competitors, it violates the core tenet of independent, unbiased reporting. Google’s algorithms are trained to recognize this conflict of interest.

How Large Language Models (LLMs) Process Bias

At a technical level, the Large Language Models powering AI Overviews are trained to identify consensus and entity relationships across the web. If a brand claims "We are the best CRM," but the broader web consensus (found in forums, social media, news outlets, and review sites) does not support that claim, the LLM identifies the brand’s self-ranking as an anomaly or biased data.

Consequently, the LLM will extract the structural data from the brand’s page (because it is well-formatted and easy to parse) but filter out the brand’s self-recommendation to provide a more objective, helpful response to the end-user.

5. Implications: Strategic Pivots for B2B Marketers

The revelation that self-promotional listicles are being weaponized by AI algorithms requires an immediate, fundamental shift in B2B digital marketing strategies. Marketers must move away from artificial self-ranking and transition toward holistic digital PR and brand-building.

Old SEO Playbook vs. New AI Search Playbook:

[OLD PLAYBOOK]
- Write self-serving listicles ranking your brand #1.
- Focus on keyword density and on-page optimization.
- Buy links to boost specific product landing pages.
- Measure success by organic keyword rankings.

[NEW AI PLAYBOOK]
- Earn unbiased, third-party mentions on G2, Capterra, and Reddit.
- Build genuine brand authority to influence LLM training data.
- Publish truly objective comparison guides (even if you don't rank #1).
- Measure success by brand share of voice in AI-generated answers.

1. Shift from On-Page Optimization to Digital PR

If AI Overviews rely on web-wide consensus to make recommendations, B2B brands must focus on earning genuine, third-party mentions. This means investing heavily in digital PR, secure media coverage, and analyst relations (such as Gartner and Forrester). The goal is to ensure that when an AI model scans the web for opinions on a category, your brand is consistently mentioned by independent, authoritative sources.

2. Optimize for User-Generated Content (UGC)

With Reddit and Quora citations soaring in AI Overviews, community management has become a critical pillar of SEO. Brands must actively monitor and participate in relevant discussions on these platforms. Ensuring that real users are discussing, recommending, and reviewing your product on Reddit can directly influence whether your brand appears in an AI Overview.

3. Embrace True Objectivity in Content Marketing

If your brand continues to produce comparison content, it must be genuinely helpful and objective. Instead of automatically ranking your tool as #1, provide an honest analysis of where your software excels and where it may fall short compared to competitors. Ironically, by presenting a balanced view, your content is more likely to be deemed trustworthy by Google’s algorithms, potentially keeping your brand in the consideration set.

4. Focus on Brand-Level Authority

Ray’s research showed that "stronger brands"—those with established category leadership, extensive third-party mentions, and clean, authoritative backlink profiles—were highly resilient against AI exclusion. Building a recognizable, trusted brand remains the ultimate defense against algorithmic shifts. Direct search volume (users searching specifically for your brand name) is a powerful signal to AI models that your company is a trusted market leader.

Conclusion: The Death of the Self-Serving SEO Hack

The findings of Lily Ray’s study signal the definitive end of the "self-serving SEO hack." For years, search engines could be manipulated by well-structured but fundamentally biased content. Google’s AI Overviews have closed this loophole by separating the source of information from the recommendation itself.

For B2B marketers, the path forward is clear. To win in the age of AI search, brands must stop trying to convince search engines that they are the best through self-published declarations. Instead, they must do the harder, more impactful work of building a product and a brand reputation that the rest of the web cannot help but recommend.