[The Engines]
How Can Brands Control Their Visibility in AI Search Results?
A new analysis of 6.8 million AI citations found that 86% come from sources brands already own, not forums. Here is what changed, who it affects, and what to do next.
The short answer
A Yext analysis of 6.8 million AI citations across ChatGPT, Gemini and Perplexity, collected between July and August 2025 and published October 9, 2025, found that 86% of citations come from sources brands already own or manage, not from Reddit or other forums. Websites accounted for 44% of citations and business listings 42%, while forums accounted for just 2%. The takeaway: brand visibility in AI answers is mostly an owned-content problem, not a third-party popularity contest.
What Did The New Yext Research Actually Find?
Yext's data team pulled 6.8 million AI citations from 1.6 million queries per model across ChatGPT, Gemini and Perplexity, collected between July 1 and August 31, 2025, and published the results on October 9, 2025. The study tested branded and unbranded, objective and subjective prompts across four industries: retail, financial services, healthcare and food service, and traced citations back to 20,820 unique domains.
The headline number is 86%: that is the share of citations that came from sources a brand directly owns or manages, not from independent third parties. Broken down by type, first-party websites accounted for 44% of citations, business listings for 42%, reviews and social platforms for 8%, and forums, the category that includes Reddit, for just 2%.
That last figure is the one worth sitting with. Forums have dominated the AI search conversation for two years running. In this dataset they are the smallest bucket by a wide margin.
Who Does This Change Affect?
Two groups should read this closely. B2B SaaS and tech growth teams trying to win category and comparison prompts (think 'best [category] tool') now have direct evidence that their own site content, not a scattering of third-party reviews, is the highest-leverage lever they control. Local, multi-location and service businesses get an even sharper signal: business listings tied with websites as the second-largest citation source, and in healthcare, listings alone accounted for 52.6% of citations.
Anyone who built a 2025 AI visibility plan around seeding Reddit threads or chasing forum mentions is optimizing for 2% of the pie. Anyone who has been treating their own website and their listings as an afterthought, while a competitor keeps both current, is losing the other 86% by default.
| AI engine | Primary citation lean | Share |
|---|---|---|
| Gemini | First-party websites | 52.1% |
| ChatGPT (OpenAI) | Business listings | 48.7% |
| Perplexity | Diversified across sources (e.g. MapQuest, TripAdvisor) | No single dominant category |
What Did The Industry Assume Before This Data, And What Changed?
Before this study, the working assumption in a lot of AI search commentary, fueled by widely circulated reports of Reddit's outsized presence in Google's AI Overviews, was that user-generated forums were the dominant fuel for AI answers. That assumption shaped a wave of advice telling brands to go seed discussions on Reddit and Quora rather than invest in their own content.
Here is the dated before-and-after. Before: a forum-first narrative with no comprehensive cross-engine citation count behind it. After, per Yext's July 1 to August 31, 2025 data collection published October 9, 2025: forums supply just 2% of citations, while brand-owned websites and listings together supply 86%. The gap between the story and the data is the whole story.
The pattern holds with variation by industry. In retail, 47.6% of citations trace to brand websites. In financial services, 48.2% come from authoritative local pages. In healthcare, 52.6% come from listings such as WebMD and Vitals. Food service is the outlier: reviews and social sources supply 13.3% of its citations, the highest share of any vertical, which tracks with how people actually decide where to eat.
What to do about it now that the data is dated and public: stop treating forum presence as your primary AI visibility lever and start treating your own website and listings as the primary one, then check whether your industry skews toward one or the other before you allocate effort.
How Do The Three AI Engines Differ In What They Cite?
The 86% figure is an average across engines, and the engines do not behave identically. Gemini leans hardest on first-party websites, at 52.1% of its citations. ChatGPT (OpenAI) leans harder on listings, at 48.7%. Perplexity spreads its citations more widely across sources like MapQuest and TripAdvisor rather than concentrating on either category.
This is why a strategy built around a single engine is a fragile strategy. A brand optimized purely for Gemini's website preference could still be invisible in ChatGPT's listing-heavy answers, and vice versa. Real AI search visibility means being cited across ChatGPT, Perplexity, Gemini and AI Overviews, not just the one engine that happens to be easiest to test.
- Gemini: 52.1% of citations come from first-party websites
- ChatGPT (OpenAI): 48.7% of citations come from business listings
- Perplexity: citations spread across a wider mix, including sources like MapQuest and TripAdvisor
How Should Brands Respond?
The response is not exotic. It is the unglamorous, compounding work of keeping owned sources accurate, structured and current, at the volume and pace AI engines expect. Yext's own framing backs this: brands influence AI visibility by keeping website content accurate, structured and crawlable, which is a description of disciplined content operations, not a trick.
Four concrete moves follow directly from the data. First, audit your first-party website for the specific questions your buyers ask AI, and make sure the answers exist in clear, citable form, not buried in a PDF or a sales deck. Second, get your business listings and location data consistent everywhere, since listings tied websites as the largest citation category overall and led outright in healthcare. Third, if your industry skews toward reviews and social, as food service does, treat review volume and freshness as a visibility metric, not just a reputation metric. Fourth, measure the outcome, not the activity: track your citation share, the percentage of relevant AI answers in your category that cite you, across all four engines, not just Google.
This is the same mechanism behind the idea of Citation Engineering: engineering authoritative, current, well-structured content at a scale and cadence that AI engines trust, across every engine that matters. The Yext data does not describe a new trick to learn. It describes the size of the opportunity sitting in content you already control.
Key takeaways
- 86% of AI citations across ChatGPT, Gemini and Perplexity come from sources brands already own or manage, per a Yext analysis of 6.8 million citations (Yext, 2025-10-09).
- Forums, including Reddit, account for just 2% of citations, far below the forum-first narrative that shaped a lot of 2025 AI search advice.
- Websites (44%) and business listings (42%) are the two biggest citation sources, and which one leads depends on your industry.
- Healthcare leans hardest on listings (52.6%), while food service leans hardest on reviews and social (13.3%), the highest of any vertical tested.
- Gemini, ChatGPT and Perplexity do not cite the same way: Gemini favors websites (52.1%), ChatGPT favors listings (48.7%), and Perplexity spreads across a wider mix.
- The practical response is unglamorous: keep first-party content accurate and structured, keep listings consistent, and track citation share across every engine, not just one.
Omnicite Editorial. "AI Search Citations: How Brands Control Visibility" The Citation Report, Omnicite. https://omnicite.co/blog/how-can-brands-control-their-visibility-in-ai-se/
Sources
86% of AI citations come from sources brands own or manage, based on 6.8 million citations analyzed across ChatGPT, Gemini and Perplexity Yext, Inc., 2025-10-09
Citation source breakdown: websites 44%, listings 42%, reviews and social 8%, forums 2%; data collected July 1 to August 31, 2025 Search Engine Land, 2025-10-09
Frequently asked questions
What did the Yext AI citation study actually measure?
It measured where 6.8 million AI citations, drawn from 1.6 million queries per model across ChatGPT, Gemini and Perplexity, actually came from, collected between July 1 and August 31, 2025 and published October 9, 2025. It covered branded and unbranded, objective and subjective prompts across retail, financial services, healthcare and food service.
Does Reddit still matter for AI search visibility?
Not nearly as much as the 2025 narrative suggested. Forums, the category that includes Reddit, accounted for just 2% of citations in this dataset, far behind websites (44%) and business listings (42%).
Which AI engine cites brand websites the most?
Gemini, at 52.1% of its citations coming from first-party websites. ChatGPT leans more on business listings, at 48.7%, and Perplexity spreads its citations across a wider mix of sources.
Is the brand-controlled pattern the same in every industry?
No. It skews by vertical: healthcare draws 52.6% of citations from listings, financial services draws 48.2% from authoritative local pages, retail draws 47.6% from brand websites, and food service is the outlier, with reviews and social supplying 13.3% of citations, its highest share of any category tested.
How can a brand actually increase its share of AI citations?
Keep first-party website content accurate, structured and crawlable, keep business listings consistent across every directory, and track citation share, the percentage of relevant AI answers that cite you, across ChatGPT, Perplexity, Gemini and AI Overviews rather than just one engine.
Does this mean more content automatically means more citations?
No. The data points to accuracy, structure and freshness of owned sources, not raw volume. A smaller set of clear, current, well-structured pages that directly answer buyer questions is worth more than a large volume of stale or unstructured content.