Meta never announced Facebook Answers. Saul Westbrook of Peripheral found the directory in June, and by October it was pulling more Google traffic each month than most publishers see in a year.
Meta Turned Old Facebook Posts Into Millions of Google Pages
Every Facebook Answers page is an AI summary on a single topic, stitched together from public posts, group threads and comments. The original posts are linked inside the page, some of them going back to 2010.
The pages appeared right after Meta added AI Mode to Facebook search in June (Gigazine on the launch). Westbrook believes they are static, indexable copies of AI Mode answers. Meta has not commented.
From 130K to 40M Visits in Five Months
The traffic barely moved until August. Then the number of ranking keywords exploded from 4 million to 53 million, and visits followed.
| Month | Worldwide visits | US visits |
|---|---|---|
| May 2026 | 129,514 | 72,617 |
| Jul 2026 | 1,032,527 | 682,664 |
| Aug 2026 | 12,167,857 | 6,781,442 |
| Sep 2026 | 36,388,769 | 23,104,796 |
| Oct 2026 | 40,590,906 | 21,058,474 |
Source: Semrush estimates, via Peripheral.
These are not long-tail scraps either. Peripheral found Facebook Answers at position 1 for an LAS airport query, position 2 for an Amex travel query (368K monthly searches) and position 6 for "Labcorp."
Google Cites These Pages. ChatGPT Can't Even See Them.
Here is the part that should bother every publisher. Westbrook found that Meta blocks LLM crawlers from Facebook Answers, so ChatGPT, Perplexity and Claude cannot read them. Google can. Its AI Overviews cite the pages regularly, and Glenn Gabe saw them in AI Mode and Gemini too (Search Engine Roundtable).
Meta is feeding its users' posts to Google's AI and to nobody else's.
Any Other Site Would Probably Be Penalized for This
Google's spam policies ban scaled content abuse, meaning many pages made mainly to manipulate rankings, however they are produced. They also ban doorway pages built to rank for similar queries. Westbrook counted more than 180 English pages on padel tennis alone.
Still, the pages came through the September spam update (September 24 to October 8, per Search Engine Roundtable), and October traffic grew. The likely difference: Facebook's domain trust and the fact that every answer is built on, and links to, its own first-party content.
Meanwhile Reddit, the old king of user content in Google, lost more than a quarter of its market value on October 30 as CEO Steve Huffman blamed Google's AI search for falling traffic (Coin Alert News).
What SEOs Should Do About Facebook Answers
Our take: this does not prove mass AI content is safe. It proves Facebook can get away with it. Do not copy the tactic on your own site.
What is worth doing:
- Search
site:facebook.com/fb-answers "your brand"to see what Meta's AI says about you. - Check whether the fb-answers subfolder ranks on your informational keywords in Semrush or Ahrefs.
- Keep a real presence on Facebook. Page posts, group posts and comments all get used as sources.
- Do not plan around these pages. A single Google action could wipe them out.
Barry Schwartz of Search Engine Roundtable doubts the pages will stay indexed now that the story is out. If they do, the message is clear: Google's spam rules depend less on how content is made than on who makes it.
Credit: original research and data by Saul Westbrook, Peripheral.