Short answer: In the largest brand-level studies published so far, how often a business is mentioned, especially on YouTube, lines up with AI visibility far more closely than how many backlinks it has. That doesn’t make links worthless. It means the signal AI engines lean on looks more like reputation than like PageRank, and social platforms are where most of that reputation is now written down.
What the numbers actually say
Ahrefs analyzed 75,000 brands and measured which brand-level factors correlate with how often a brand appears in Google AI Overviews. Branded web mentions came out on top at 0.664, while backlink count sat at just 0.218. A follow-up across ChatGPT, Google AI Mode, and AI Overviews found YouTube mentions (a brand named in a video’s title, transcript, or description) correlated at roughly 0.737, the strongest of any factor measured.
Two cautions belong right next to those numbers. First, these are correlations, not proof that mentions cause AI visibility. Second, the sample skews toward established brands. Still, the gap between 0.737 and 0.218 is too large to ignore.
Why AI engines read the web this way
Large language models don’t crawl a link graph the way classic Google ranking did. They synthesize answers from passages that describe an entity clearly, often, and in context. A YouTube transcript that says “we called a local plumber for a water heater replacement in Asheville” is exactly that kind of passage. A footer link is not.
That’s also why the most-cited domains in AI answers are social and community platforms. A Peec AI analysis of 30 million sources found Reddit was the most-cited domain overall, followed by YouTube, LinkedIn, and Wikipedia. Across several data firms, YouTube’s share of social-platform citations roughly doubled between August and December 2025, from 18.9% to 39.2%.
The volatility problem
Citation patterns move fast, which is the main reason to spread your presence rather than bet on one platform. Semrush tracked more than 230,000 prompts over 13 weeks and saw ChatGPT’s Reddit citation rate drop from close to 60% to around 10% in about six weeks. A strategy built only on Reddit threads would have lost most of its AI exposure in a month and a half.
Mentions spread across YouTube, community discussions, reviews, local news, and directories are more durable because no single engine change wipes them all out.
What this looks like for a local business
National brand studies only go so far, so here is what PushLeads sees in its own client tracking of local service businesses:
- Across 39 client accounts, YouTube was cited 6,102 times in AI answers about those businesses, more than Reddit (3,786) and Facebook (1,920) combined. None of those videos were made by the clients themselves.
- Generic, AI-narrated “what is SEO” style videos earned zero citations in a check of 493 videos. Videos built around one specific customer question, embedded on a matching page with video schema, started getting cited within two weeks.
- The fleet held only about 1.5% of the available AI citation slots for its tracked questions, which means most local AI answers are still up for grabs.
And the local bar is high. SOCi’s 2026 study of more than 350,000 locations found 35.9% appear in Google’s map pack but only 1.2% are recommended by ChatGPT. Mentions and reviews are what close that gap; proximity alone won’t.
A mention-first checklist
- Make your business easy to identify. One name format, one phone format with area code, and one service-area description everywhere you’re listed. Mentions only help if engines can tie them to the same entity.
- Turn customer questions into YouTube videos. Put the exact question in the title, answer it in the first 15 seconds, say your business name and town out loud (it lands in the transcript), add chapters, and upload corrected captions.
- Embed each video on the matching page of your site with VideoObject schema, so the transcript, the page, and the brand reinforce each other.
- Earn conversation, don’t post ads. Answer real questions in local subreddits and Facebook groups as yourself. Ask happy customers to mention the service and their town in reviews.
- Collect off-site mentions on purpose. Local news, chamber features, supplier “find a pro” pages, podcasts, and sponsorships all add to the signal.
- Measure mentions, not just rankings. Keep a fixed list of 30 to 40 customer questions and check monthly which businesses ChatGPT, Perplexity, Gemini, and Google’s AI features name.
Backlinks aren’t dead, they’re just not the whole story
Links still help pages rank, and pages that rank well are frequently cited in Google’s AI Overviews. But for being named in an AI answer, the data points toward reputation signals that live mostly on social and community platforms. The businesses that win treat social content, SEO, and AI visibility as one system.
For the full picture, including how AI answers affect clicks, the local map pack vs. AI recommendation gap, and a four-stage action plan, read PushLeads’ complete guide: The Visibility Flywheel: How Social Media, AI Answers, and Google Rankings Now Feed Each Other.