How Google's AI Overviews Use Facebook, Instagram, and TikTok: A 300-Million-Search Analysis

Many marketing teams assume a large social media following automatically guarantees a mention in AI answers. But an analysis of more than 300 million US searches by BrightEdge co-founder and CTO Lemuel Park suggests that's not quite true.

According to the analysis, Google's AI-powered search summaries (AI Overviews) cited Facebook as a source 19.5 million times, Instagram roughly 877,000 times, and TikTok 78,000 times. In other words, in one out of every 15 US searches, Facebook or Instagram content gets pulled into a Google AI answer.

Crucially, citation has nothing to do with follower count. Google pulled an Instagram post to answer "mobile payment app" (roughly 18.5 million searches a month); a baseball team's Facebook post answered "where do the Brewers play the Reds"; and a TikTok video answered a "Yankees vs. Dodgers" query. Park noted an account with just 3,000 followers got cited more often than a creator with millions of followers — simply because that particular post answered the question precisely.

Different platforms earn trust for different query types: Facebook for quick, local, and community questions; Instagram for culture and shopping topics; TikTok for trends and "how-to" queries; Reddit for personal-experience and problem-solving questions; and YouTube leads on topics requiring step-by-step instructions.

The gap widens at the purchase-decision stage: Google AI pulls from Instagram at the bottom of the funnel — where-to-buy, price, or discount questions — nearly 90% of the time. Facebook shows up mostly post-purchase, for troubleshooting, returns, or usage questions (about 23% of such cases — more than double Instagram's share). Interestingly, Google mostly treats social platforms as a location-and-availability signal (11-14% of cases, for queries like "nearby" or "open now"), while ChatGPT leans more toward discounts and pricing (20-24%), often naming a specific product — a graphics card model or a sneaker release, for example.

Another key finding: when Google AI cites Facebook or Instagram in a shopping query, about 85% of the time it names a major retailer or marketplace, while product-manufacturer brands account for just 3-4% of such mentions, and nearly 75% of cited brands are mentioned only once.

Park's advice is clear: lead with the number, turn stories into case studies with hard figures and a clear methodology, drop "recap"-style copy and self-congratulatory language, and publish content in open, text form rather than locking it inside video or private groups — because text remains the one format AI systems can actually read.