--- title: "Facebook Answers: 6M Meta AI pages in Google" description: "Facebook Answers are Meta AI summary pages, and Google has indexed about 6 million of them. What the data shows, what the spam policy says and what not to copy." date: 2026-11-06 category: SEO tags: ["SEO", "Facebook Answers", "Scaled content abuse", "AI content", "Indexing"] url: https://uper.pl/en/blog/facebook-answers-google-ai-pages/ --- # Facebook Answers in Google 6 million AI-generated pages. Is it scaled content abuse? **Facebook Answers are AI-written summary pages that Meta publishes under facebook.com/fb-answers/. According to Search Engine Roundtable, Google has about 6 million of these URLs in its index, and some of them rank near the top.** Each page is a short summary of a topic, followed by the public Facebook posts the summary was built from. Glenn Gabe calls it "scaled content abuse" outright. Neither Google nor Meta has commented. ## How did Facebook Answers come to light? Saul Westbrook of the Peripheral Creative agency described the directory in a post dated 4 November 2026. He writes that in June 2026 he noticed Meta had "quietly indexed millions of AI-generated Facebook pages" and that they were showing up often on Google. On 5 November Barry Schwartz picked the story up on Search Engine Roundtable, quoting Glenn Gabe's posts on X as well. ### What a page looks like Westbrook shows a screenshot of facebook.com/fb-answers/using-american-express-for-travel/. At the top sits the AI summary, and below it the posts by people and pages used as its sources. At the bottom there is a "More like this" box linking to other answers, and a question asking whether the result was good or bad. Westbrook says the pages appeared shortly after Meta launched its own AI Mode, a conversational search engine built on content from Facebook and Instagram. He thinks they are "probably" static versions of that tool's answers, and that the rating question may be there to collect feedback on their quality. That is his guess: as of 3 November, Meta had not answered his request for comment. He sees no sign of editorial oversight. Topics overlap, and he found more than 180 English-language pages on padel tennis alone. ## How many pages and how much traffic? Two different numbers are going around in this story, and they shouldn't be mixed up: - **about 6 million** is the number of /fb-answers/ pages in Google's index, according to Schwartz and Gabe, - **over 39 million** is the number of pages that, by Semrush's estimate, rank in Google in the US for at least one keyword. Westbrook is the source of this figure. Nobody has explained the gap yet. Both numbers come from tools and SEO observations, not from Google's data. Westbrook also published Semrush's monthly organic traffic estimates: | Month 2026 | Worldwide | US | UK | |---|---|---|---| | April | 0 | 0 | 0 | | May | 129,514 | 72,617 | 5,445 | | June | 114,728 | 89,438 | 2,588 | | July | 1,032,527 | 682,664 | 50,786 | | August | 12,167,857 | 6,781,442 | 673,471 | | September | 36,388,769 | 23,104,796 | 2,000,039 | | October | 40,590,906 | 21,058,474 | 2,936,009 | Between July and August, the number of keywords with a Facebook Answers page ranking for them rose worldwide from 4 million to 53 million. In October worldwide traffic grew to about 40.6 million, while US traffic fell from 23.1 million to 21.1 million. These are Semrush model estimates, not Facebook's analytics, so I read them as an order of magnitude. Example US rankings Westbrook lists: | Keyword | Search volume | Position | Estimated traffic | |---|---|---|---| | Amex Travel | 368K | 2 | 23.9K | | Champions League Schedule | 1M | 9 | 22K | | LAS Airport Location | 22.2K | 1 | 17.8K | | Labcorp | 2.7M | 6 | 16.4K | The sources disagree on dates. Westbrook noticed the pages in June, Gabe says Facebook started populating the directory in July, and Semrush records the first visits as early as May. Nobody knows when the first pages went up, but the table shows they started ranking at scale in July. ### Visibility in AI results Gabe writes that the directory ranks for millions of queries in the regular results, in AI Overviews and in AI Mode, and his screenshots show Gemini citations too. Westbrook confirms that AI Overviews cite these pages often. He adds that Facebook blocks LLM crawlers from them, and he couldn't get any model to read the pages. ## Is it scaled content abuse under Google's rules? Google's spam policies define scaled content abuse as many pages generated for the primary purpose of manipulating search rankings and not helping users. It is typically about large amounts of unoriginal content that provides little to no value, no matter how it's created. The examples in the documentation include: - using generative AI tools or other similar tools to generate many pages without adding value for users, - scraping feeds, search results or other content to generate many pages, where little value is provided to users, - stitching or combining content from different web pages without adding value. Westbrook also brings up doorway abuse, meaning pages created to rank for specific, similar search queries, and points out that many Facebook Answers pages are very similar to each other. The padel pages are a good example. At the top of the policies Google states that they apply to all web search results, including those from Google's own properties. The documentation makes no exception for big platforms. At the same time, the test is purpose and value to users, not the number of pages or the fact that AI wrote them. Westbrook allows that Google may see these pages as offering genuine value, because they draw on Facebook's own posts, and notes that they survived the September spam update. Gabe has no doubt it's scaled content abuse. Schwartz suspects that now the story is public, the pages won't stay indexed for long. These are opinions, and Google hasn't announced any decision. ### The content quality problem Westbrook found pages with advice on investing in cryptocurrency, medical advice and tips on how to gamble on a budget, built mostly from posts and comments by individuals rather than institutions. One page told him the next day's temperature in London would be 34-35, in November. Among the sources he found some of his own group posts from 2024, and a 2010 post from a group with 6 members. He found no way to pull his post out of a page that is already published. I write more about how Google treats content made with AI in [AI vs. human content: how Google tells them apart](/en/blog/ai-vs-human-content/) and [AI content generation and SEO](/en/blog/ai-content-generation-and-seo/). ## What does it mean for your site? ### Facebook has what your domain doesn't In its site reputation policy, Google says it generally presumes that individual pages, including new ones, match the overall quality of the other pages on the domain. facebook.com has an enormous history and its own supply of content in users' posts. I think that's why the new directory started ranking so fast, although Google hasn't confirmed it. A small site that put out millions of similar AI summary pages in a few months would have neither that history nor its own data. ### Don't copy the pattern Westbrook concludes that if Facebook isn't penalised, consistent sourcing and first-party data can let you scale content without a penalty. I disagree. One case involving a huge platform, with no word from Google, says nothing about how a small site would be judged. Westbrook himself advises against building a strategy on these pages, because they are new and may not keep ranking. If you write with AI, what counts is what the text adds for the reader: your own data, experience, checked facts. ### Check your keywords and your brand If your pages answer informational questions, check your main keywords by hand and see whether the /fb-answers/ directory shows up in the results or in AI Overviews. Westbrook also suggests the query `site:facebook.com/fb-answers "your brand"`. It lists the indexed pages that mention your company, built from other people's posts, which is a simple way to see what Meta's AI says about your brand in Google. In my view this case also shows how differently Google filters pages on a big domain and a small one. I cover that filter in [why Google doesn't index every crawled page](/en/blog/google-indexing-semantic-selectivity/). ## Limitations and unknowns - **No word from Google or Meta.** None of the sources I read include a comment from either company. - **The numbers are estimates.** The 6 million is an index observation reported by SER and Gabe; the 39 million pages and the traffic data are Semrush estimates. Westbrook writes about "over 100 million" visits since launch, while his own table for May to October adds up to about 90.4 million. - **I didn't check the pages myself.** When I tried to fetch facebook.com/fb-answers/, it returned only an empty page shell, so my description of the pages relies on Westbrook's screenshots and account. - **Instagram.** Schwartz mentions rumours that Instagram is doing something similar. Nobody has confirmed this. ## Summary Facebook Answers is a directory of AI summary pages built from public posts, which within a few months reached about 6 million URLs in Google's index and, by Semrush's estimate, around 40 million visits a month. Google's spam policy describes very similar practices and makes no exceptions for big sites, but Google hasn't announced any action or taken a position. For the owner of an ordinary site this is not a traffic recipe. It is better to watch it and check whether these pages are taking your place in the results. AI vs. human content: how Google tells them apart for details.' } ]} heading="Frequently Asked Questions" id="faq" /> ## Sources