AI Search Research · Law Firm Strategy

AI Search Cites What Brands Control: What Yext’s 6.8M-Citation Study Means for Law Firms

A 6.8 million-citation analysis across ChatGPT, Gemini and Perplexity found that 86% of AI citations come from sources brands already manage. Here is what that means for law firm AEO, GEO, AIO, SEO and E-E-A-T strategy — with the caveats that matter for legal marketers.

🎯 Key Takeaways

• 86% of AI citations come from sources brands already control — websites, listings, and reviews/social — per Yext’s analysis of 6.8M citations across ChatGPT, Gemini and Perplexity (Yext Research, October 9, 2025; data collected July 1–August 31, 2025).
• First-party websites generated 44% of citations (2.9M), listings 42% (2.9M), reviews/social 8% (545K), and forums like Reddit just 2% once location and query intent were applied (Yext, October 9, 2025).
• Models source differently: Gemini favored websites (52.1%), OpenAI/ChatGPT leaned on listings (48.7%), and Perplexity diversified across directories like MapQuest and TripAdvisor (Yext, October 9, 2025).
• The study covered four industries — retail, financial services, healthcare and food service — and did not include legal services, so direct extrapolation to law firms requires care (Yext methodology disclosure, October 9, 2025).
• For law firms, the directional implication is clear: invest in first-party site architecture, accurate local listings, structured data, and review signals before chasing Reddit threads or off-site “hacks.”

Yext analyzed 6.8 million AI citations across ChatGPT, Gemini and Perplexity between July and August 2025 and found that 86% came from brand-controlled sources — websites, listings, and reviews. For law firms, that finding reframes AEO, GEO, AIO and SEO around one practical idea: own the sources AI models cite, and citation visibility follows.

🔍 What Yext Actually Found

On October 9, 2025, Yext (NYSE: YEXT) published a study analyzing 6.8 million AI citations collected from 1.6 million queries per model across ChatGPT (OpenAI), Gemini (Google), and Perplexity between July 1 and August 31, 2025. Queries tested four intent quadrants — branded/unbranded crossed with objective/subjective — across four industries: retail, financial services, healthcare and food service. The dataset identified 20,820 unique citation domains overall.

The 86% Headline

Across all industries tested, Yext reported that 86% of AI citations came from sources brands already control or manage — primarily first-party websites, business listings, and brand-managed reviews/social properties. Forums such as Reddit accounted for just 2% of citations once location context and query intent were factored in (Yext Research, October 9, 2025).

The breakdown of citation sources across the full dataset:

How Each Model Sources Differently

One of the more useful findings — and one that complicates any single “AI search” strategy — is that the three models have distinct sourcing preferences. Per Yext (October 9, 2025):

• Gemini (Google): 52.1% of citations came from brand-owned websites — behavior consistent with its grounding in Google Search.
• ChatGPT (OpenAI): 48.7% of citations came from listings, suggesting heavier reliance on directory and third-party data layers.
• Perplexity: Diversified across niche directories — MapQuest and TripAdvisor were called out specifically — with stable behavior across sectors.

This matters because it means optimizing only for ChatGPT (the most-discussed model in 2024–2025) risks invisibility in Gemini, and vice versa. Our ChatGPT optimization guide, Google Gemini optimization guide, and Perplexity AI optimization guide each address the platform-specific signals in more depth.

The “Just 2% Reddit” Caveat

The headline that drew the most social commentary — including the post that prompted this article — was that Reddit and similar forums accounted for only about 2% of citations once location context and query intent were applied. That figure should be read carefully.

⚖️ Translating the Findings: AEO, GEO, AIO, SEO & E-E-A-T

Before reading the Yext data as either a vindication of SEO or a dismissal of AEO, it helps to define each discipline precisely. These acronyms are often used interchangeably; they should not be.

SEO — The Foundation

Search Engine Optimization (SEO) is the discipline of earning visibility in traditional ranked search results — Google and Bing’s blue links, local 3-pack, and organic SERPs. Its inputs (crawlable architecture, page experience, authoritative content, link signals, structured data) overlap heavily with what AI models need to retrieve, parse and cite information.

The Yext study reinforces, rather than displaces, SEO. When 44% of AI citations come from first-party websites, the implication is that the same crawlability, content quality and schema work that drives traditional rankings is also feeding AI answers. Our law firm SEO services and local SEO program are built around this overlap.

AEO — Answer Engine Optimization

Answer Engine Optimization (AEO) targets visibility in direct-answer surfaces: Google’s featured snippets, “People Also Ask,” voice assistants, and the synthesized answers AI tools generate before linking out. AEO tactics emphasize Q&A formatting, clear definitions, concise direct-answer paragraphs, FAQ schema, and entity clarity.

The Yext finding that 86% of citations come from brand-managed sources doesn’t invalidate AEO — it specifies where AEO effort should be applied. AEO best practices like structured Q&A pages and clear topical answers should live primarily on the firm’s own site, not be outsourced to forum posts hoping for citation. For a deeper walkthrough, see our guide on Answer Engine Optimization for law firms in 2026.

GEO — Generative Engine Optimization

Generative Engine Optimization (GEO) is the discipline of optimizing content so it is more likely to be selected, quoted, and cited inside generative AI responses (ChatGPT, Gemini, Perplexity, Claude, Copilot). The term was formalized in peer-reviewed research presented at the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’24), Barcelona, Spain, August 25–29, 2024, where Aggarwal et al. demonstrated that specific content modifications — including the strategic use of quotations, statistics and citations — could improve generative engine visibility by up to 40% on tested benchmarks (Aggarwal et al., 2024, DOI: 10.1145/3637528.3671900).

The Yext data is fully compatible with GEO theory. If 86% of generative-engine citations are coming from sources brands already control, then the GEO levers — content design, citation density, schema, structured Q&A — are most effective when applied to the firm’s own pages. Our GEO for law firms pillar and GEO audit service operationalize this.

AIO — AI Optimization (and AI Overviews)

AI Optimization (AIO) is the umbrella term increasingly used to describe optimization across all AI-driven discovery surfaces — including Google’s AI Overviews (formerly Search Generative Experience), Microsoft Copilot answers, and embedded AI search inside other platforms. AIO subsumes AEO and GEO for many practitioners, though usage is not yet standardized.

For attorneys, the practical AIO question is: when a prospect asks “best personal injury lawyer near me” inside an AI surface, does the firm appear in the synthesized answer, and is the cited source one the firm controls? The Yext findings suggest that, more often than not, the cited source will be a website, listing, or review the firm could have managed. Our AI search optimization framework and LLM optimization for law firms guide cover this in detail.

E-E-A-T — The Underlying Quality Signal

E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) is Google’s quality framework, originally articulated in its Search Quality Rater Guidelines. It is not a ranking algorithm directly, but a set of signals that human raters use to evaluate content quality — signals that are increasingly relevant to how AI systems evaluate which sources to cite, particularly in YMYL (Your Money or Your Life) categories like legal services.

The Yext findings are consistent with an E-E-A-T-aware retrieval logic: AI models appear to weight first-party authoritative content and structured directory data heavily — both signals that align with what Google’s framework values. For law firms, where YMYL standards apply with full force, E-E-A-T is not optional. See our E-E-A-T safety guardrails for law firm marketing for the operational checklist.

🏛️ What This Means for Law Firms Specifically

The Yext study is the strongest evidence yet that brand-controlled assets drive AI visibility, but legal marketing has features that should adjust how the findings are applied.

1. Legal services were not in the dataset. Yext tested retail, financial services, healthcare and food service. Healthcare is the closest analog to legal in some respects — both are high-trust, regulated, locally-delivered services with directory ecosystems. In healthcare, Yext found that 52.6% of citations came from third-party directories like WebMD and Vitals, more than from first-party sites. For law firms, the analogous directories include Justia, Avvo, FindLaw, Martindale-Hubbell and Super Lawyers. The directional implication is that both first-party site quality and legal directory presence likely matter for AI citation visibility in this vertical.

2. Legal queries are often complex and high-consideration. Yext acknowledged that its query set tested four intent quadrants — branded/unbranded crossed with objective/subjective — but did not isolate B2B or long-form legal-style queries. A query like “what should I do after a rear-end accident in Los Angeles” generates a different citation pattern than “best Italian restaurant near me.” Practitioner observations suggest that AI models do pull from forum and Q&A content for some legal queries, particularly when the asker is in pre-consult research mode (Yext methodology disclosure, October 9, 2025).

3. Local intent is dominant in legal search. The Yext finding that “AI generates answers based on a person’s real-world location and context, not a generic brand view” maps directly onto legal client acquisition. Most consumer legal queries are local. This makes AI-powered local optimization more important for law firms than it might be for other verticals.

4. Compliance and review constraints affect what firms can publish. Bar association advertising rules limit how testimonials, results, and comparative claims can be presented — even when those signals are exactly what AI models tend to surface. Law firm AEO and GEO strategy must work within those constraints. A general-industry citation playbook will need adjustment before it is deployed for an attorney.

📋 A Citation-First Playbook for Law Firms

The most defensible reading of the Yext data, combined with the GEO research from KDD ’24, is that law firms should invest first in the four asset classes AI models cite most often — and apply AEO/GEO best practices on top of those assets, not separately from them.

1. First-Party Site Architecture

Because first-party websites generated the largest single share of citations across all tested industries (44% per Yext, October 9, 2025), the firm’s own site is the highest-leverage AI visibility asset. Practical steps include practice-area pillar pages with clear topical scope, location pages tied to a hub-and-spoke architecture, FAQ pages with FAQPage schema, and direct-answer paragraphs (30–50 words) at the top of every important page. Our 10 SEO rules rewritten for the AI search era walks through the specifics.

2. Local Listings & NAP Consistency

Listings accounted for 42% of citations in the Yext dataset — nearly tied with websites — and OpenAI’s ChatGPT leaned on listings most heavily (48.7%). For law firms, that translates to disciplined Google Business Profile management, accurate NAP (name, address, phone) data across legal directories (Justia, Avvo, FindLaw, Martindale-Hubbell, Super Lawyers), and structured local landing pages for every physical office. AI-powered local optimization is where most of this work happens.

3. Reviews & Reputation Signals

Reviews and brand-managed social content drove 8% of citations across the Yext dataset — modest in aggregate, but concentrated in subjective and branded queries (and 13.3% in food service specifically). For law firms, this means review velocity, response quality, and reputation across Google, Avvo and Yelp matter — bounded, of course, by state bar advertising rules on testimonials and endorsements. Sustained review programs that comply with ABA Model Rule 7.1 and state-specific variants should be a standing component of any AI visibility plan.

4. Schema & Structured Data

None of the above works at full effect without machine-readable structure. Schema.org markup — LegalService, LocalBusiness, Attorney, FAQPage, Article, and BreadcrumbList at minimum — is how AI retrieval systems disambiguate entities, locations, and topics. Google’s own Search Central documentation outlines the canonical entity types and validation tools. Skipping schema is the most common reason law firm sites underperform on AI citation tests despite strong content.

5. Citation-Worthy Content Design (the GEO Layer)

On top of these four asset classes, the Aggarwal et al. (KDD ’24) research demonstrates that content design choices — inline citation density, statistic anchoring, quotation use, and authoritative source linking — measurably increase the probability that a generative engine selects a page as a citation source. This is the GEO layer. It applies on top of strong SEO and AEO foundations, not as a replacement.

📊 Example Measurement Framework

Yext’s own research noted that 64% of marketing leaders said they were unsure how to measure success in AI search (Yext, October 9, 2025). A practical framework for law firms looks like this:

Our GEO audit for law firms productizes this baseline process, and the citation revolution analysis covers the broader strategic shift.

❓ Frequently Asked Questions

Find Out Where Your Firm Stands in AI Search

InterCore Technologies has built AI-driven marketing systems for law firms since 2002 — 23+ years before “GEO” had a name. Our team will baseline your firm’s citation profile across ChatGPT, Gemini, Perplexity, Google AI Overviews and Copilot, and build a citation-first plan around the assets you control.

Request Your Free AI Visibility Audit →

📚 References

1. Yext, Inc. (October 9, 2025). “Yext Research: 86% of AI Citations Come from Brand-Managed Sources, Clarifying How Marketers Can Compete in the AI Search Era.” Press release and methodology disclosure. Study analyzed 6.8 million AI citations from 1.6 million queries per model across ChatGPT (OpenAI), Gemini (Google), and Perplexity between July 1 and August 31, 2025. URL: https://www.yext.com/about/news-media/ai-citations-release
2. Yext, Inc. (October 9, 2025). “AI Doesn’t Rank, It Cites. And 86% of Its Sources Are Brand-Managed.” Yext Blog. URL: https://www.yext.com/blog/2025/10/ai-citations-86-percent-of-sources-are-brand-managed
3. Yext Research. “AI Citations: User Locations & Query Context.” Full research report. URL: https://www.yext.com/research/article/ai-citations-user-locations-query-context
4. Goodwin, D. (October 9, 2025). “AI search relies on brand-controlled sources, not Reddit: Report.” Search Engine Land. URL: https://searchengineland.com/ai-search-citations-brand-controlled-sources-463166
5. Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K., & Deshpande, A. (2024). “GEO: Generative Engine Optimization.” In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’24), Barcelona, Spain, August 25–29, 2024, pp. 5–16. DOI: 10.1145/3637528.3671900
6. Google Search Central. “Introduction to structured data markup in Google Search.” Developer documentation. URL: https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data
7. Google. “Search Quality Evaluator Guidelines” (E-E-A-T framework). Most recent public version. URL: https://services.google.com/fh/files/misc/hsw-sqrg.pdf
8. Wills, B. (2025). LinkedIn post on AEO/GEO/SEO and Yext research interpretation. URL: https://www.linkedin.com/in/benwills/
9. Yext. “How ChatGPT, Perplexity, Gemini, and Claude Actually Decide What to Cite.” Follow-on Yext Research analysis of 17.2M citations. URL: https://www.yext.com/blog/how-chatgpt-perplexity-gemini-claude-decide-what-to-cite

Conclusion

The Yext study does not declare AEO and GEO dead — it specifies where those disciplines pay off. When 86% of AI citations across 6.8 million data points come from brand-controlled surfaces, the strategic conclusion is not “do less AEO” but “make sure the assets you already own are doing the work AEO, GEO, AIO and SEO are designed to do.” For law firms, that means strong first-party site architecture, disciplined listings management, compliant review programs, and rigorous schema — all layered with the citation-density and content-design tactics that the KDD ’24 GEO research demonstrated can move the needle.

The harder question is measurement. Yext’s own research found that 64% of marketing leaders are unsure how to gauge AI search success. A firm that baselines its citation profile now, defines a target query set, and re-tests monthly will be ahead of competitors who are still arguing about whether AEO is “real.” For a practical starting point, explore our GEO services for law firms, the InterCore legal marketing hub, or schedule a conversation with our team via the contact page.

The citation economy is changing fast, but the underlying lesson from Yext is reassuring for firms that have done the foundational work: control the surface, control the citation.