--- title: "Is my content business at risk from Google’s AI summaries?" description: "It depends on your content. Factual pages lose visits to Google’s AI summaries; opinion and experience content gained, until AI Mode eroded the gain." canonical: "https://underneath.agency/resources/content-business-risk-from-ai-search" published: 2026-10-08 updated: 2026-10-10 publisher: "Underneath (https://underneath.agency/agent)" entity: "https://underneath.agency/.well-known/entity.json" --- Guide · AI search # Is my content business at risk from Google’s AI summaries? It depends on what your content is. Research on Wikipedia and Reddit suggests that factual content, which a summary can restate, loses visits, while opinion, advice and experience content can gain. Google’s conversational AI Mode erodes even those gains, so no content business should assume it is safe. ## The short version 1. English Wikipedia, a factual reference site, lost about 5% of its search traffic after AI Overviews became Google’s US default in 2024 (Khosravi and Yoganarasimhan). 2. Reddit communities that AI Overviews can cite gained 12.0% more daily comments, and experience-based communities gained 2.3 to 2.8 times more than fact-based ones (Zhang, Cui and Zhang). 3. After AI Mode rolled out globally in August 2025, that extra gain for experience-based communities fell 59% in commenters and disappeared in comments. 4. 50.63% of the pages AI Overviews cite carry display ads, so many of the sites supplying the answers depend on the clicks the answers absorb (Xu and colleagues, 2026). 5. Forcing every search into AI Mode cut the share of people clicking through to news sites by 12.5 points and to Reddit by 21.2 points, in a 2026 US experiment. ## Which kinds of content are most at risk? Factual content that a summary can restate in full is most at risk of losing visits. [Zhang, Cui and Zhang](https://arxiv.org/abs/2605.16428) split online content into two kinds. Fact-based content, such as reference entries or a quick technical answer, can be judged from a summary alone. Experience-based content, such as opinions, advice and personal stories, has to be read in full to be worth anything. For the first kind, an AI summary does the job and replaces the visit. For the second, the summary works more like a trailer that sends people to the full discussion. In the authors’ words, factual-content platforms “face substitution risk, as AI can directly satisfy the information need.” | Content type | Examples | What an AI summary tends to do | |---|---|---| | Fact-based | Definitions, reference facts, quick how-to answers | Answers the question, so fewer people click | | Experience-based | Reviews, advice threads, personal experience, opinion | Points people to the full content, so more may visit | Most content mixes both. Of the 105,012 Reddit communities the study classified, an AI model labeled 56.6% fact-based and 43.4% experience-based. ## What happened to a factual reference site? Wikipedia lost about 5% of its English search traffic once AI Overviews became the US default. [Khosravi and Yoganarasimhan](https://arxiv.org/abs/2602.18455) compared English Wikipedia articles with the same articles in German and French through 2024. English search traffic fell by about 5.45% relative to German and 4.82% relative to French. That is roughly 100.27 million fewer search visits a month. Wider evidence on [whether AI Overviews reduce clicks](https://underneath.agency/resources/do-ai-overviews-reduce-clicks) points the same way. That figure covers only the first page a visitor lands on. Article-to-article links make up about 40% of Wikipedia’s traffic, so a lost search visit may also cost the pages a reader would have clicked next. Wikipedia runs no ads. For an ad-supported publisher losing the same traffic, the authors estimate roughly $0.901M to $3.090M of ad revenue a month. Press reports collected by [Xu, Iqbal and Montgomery](https://arxiv.org/abs/2605.14021) describe harder cases, such as a travel blog that shut down after its traffic fell 90%. These are single stories, not measured studies, but they show the tail of the risk. ## What happened to discussion and experience content? It gained: Reddit communities that AI Overviews could cite saw more people join their discussions. Google keeps adult communities out of AI Overviews but still lists them in normal results. That gave Zhang and colleagues a clean comparison across 105,012 Reddit communities from January 2024 to July 2025. After AI Overviews expanded internationally in August 2024, citable communities gained 12.0% more daily comments and 12.4% more daily commenters. The gain came mostly from new people taking part, not existing members posting more. It was 2.3 times larger for comments, and 2.8 times larger for commenters, in experience-based communities than in fact-based ones. Smaller communities gained most in relative terms: in the second-smallest size group, comments rose 56% and commenters 77%. Reddit is a special case. Khosravi and Yoganarasimhan point out that Reddit is cited unusually often in AI Overviews following its data-sharing agreement with Google. [Our study of Reddit citations in Google’s AI](https://underneath.agency/research/ai-reddit-citations-study) looks at which threads get cited. A business blog with the same kind of content may not see the same lift. ## Does AI Mode change the picture? Yes: Google’s conversational AI Mode erased much of the gain experience content had won. AI Mode is Google’s chat-style search, where people can ask follow-up questions instead of clicking through. Zhang and colleagues extended their data through December 2025, after AI Mode rolled out globally in August 2025. The extra gain for experience-based communities fell 59% in commenters, and for comments it disappeared entirely. A summary that can answer follow-up questions starts to replace the discussion itself. A field experiment by [Wang and colleagues](https://arxiv.org/abs/2608.18352) points the same way. When 1,100 US volunteers had every Google search sent to AI Mode for a week in March 2026, the share clicking through to news sites fell 12.5 points, to Reddit 21.2 points and to Wikipedia 9.9 points. Our article on [AI Mode as Google’s default](https://underneath.agency/resources/ai-mode-default-traffic-loss) covers that experiment in detail. ## Are big publishers or small sites more exposed? The research disagrees, so treat any confident answer with caution. [Grossman and colleagues](https://arxiv.org/abs/2604.27790) found that AI Overviews lean less on the most popular websites than Google’s normal results. Counting only the first source for each search, sites in the 1,000 most popular domains held the top spot for 52.7% of searches in normal results and 40.0% in AI Overviews. The authors conclude that generative search may help niche sites at the expense of large ones. [Aral, Li and Zuo](https://arxiv.org/abs/2602.13415), using a different set of searches across many countries, found the opposite. In their data, AI search pointed to the most-visited websites more often, and to the long tail of smaller sites less often, than normal search. What both agree on is exposure. Xu and colleagues found that 50.63% of the pages AI Overviews cite run display ads. A large share of the content that feeds these summaries belongs to businesses that earn money from the visits the summaries can absorb. ## What should you do about it? Sort your content by whether a summary can replace it, and move investment toward what it cannot. 1. Audit your traffic pages. Mark which ones mainly state facts or answer a quick question, and which offer experience, judgment, tools or discussion. 2. Treat the factual pages as the most exposed. On Wikipedia’s evidence, [being cited did not stop the loss](https://underneath.agency/resources/ai-overview-citations-vs-lost-clicks). 3. Build content that has to be consumed in full: firsthand reviews, original data, community discussion and tools people use rather than read. 4. Watch AI Mode closely. It removed much of the gain experience content had won from AI Overviews. 5. Reduce dependence on ad-supported page views where you can, since the answer now often sits above your page. If you want help assessing which of your pages are exposed, see our [generative engine optimization service](https://underneath.agency/services/generative-engine-optimization). ## What does the research not tell us yet? The research covers two very large platforms, not a typical content business. - Wikipedia is a nonprofit reference site; Reddit is a forum with a data deal with Google. Neither looks like a niche publisher or a company blog. - The Reddit study measures comments and commenters, not page views or revenue. - Content types were labeled by an AI model, which adds some error. - Video, audio and other formats are untested; the Reddit authors flag this themselves. - Studies disagree on whether large or small sites lose more. - The data after AI Mode’s launch covers only a few months, so the long-run effect is unknown. ## Frequently asked questions ### Will Google’s AI Overviews kill my blog traffic? Not necessarily, but factual pages are at risk. English Wikipedia lost about 5% of its search traffic after AI Overviews launched, while opinion and advice communities on Reddit gained engagement. ### Does AI search help forums and online communities? It did at first. Reddit communities that AI Overviews could cite gained 12.0% more daily comments, but much of that gain faded after AI Mode rolled out. ### Which topics see the most AI Overviews? In one 2026 study of general-knowledge questions, business (84%), history (81%) and education (74%) questions produced AI Overviews most often, while sports, technology and travel (32% to 34%) did least, according to [Huang and colleagues](https://arxiv.org/abs/2603.16138). ### Are news publishers losing clicks to Google’s AI? The evidence says yes for AI Mode. In a 2026 US experiment, forcing every search into AI Mode cut the share of people clicking through to news sites by 12.5 percentage points. ## Sources - Zhang, Cui and Zhang (2026), [The Impact of AI Search on the Online Content Ecosystem: Evidence from Google and Reddit](https://arxiv.org/abs/2605.16428), arXiv:2605.16428. - Khosravi and Yoganarasimhan (2026), [Impact of AI Search Summaries on Website Traffic: Evidence from Google AI Overviews and Wikipedia](https://arxiv.org/abs/2602.18455), arXiv:2602.18455. - Xu, Iqbal and Montgomery (2026), [Measuring Google AI Overviews: Activation, Source Quality, Claim Fidelity, and Publisher Impact](https://arxiv.org/abs/2605.14021), arXiv:2605.14021. - Wang, Gleason, Bart, Wilson and Metaxa (2026), [AI in Search Reduces Publisher Referrals Without Improving User Experience: Experimental Evidence](https://arxiv.org/abs/2608.18352), arXiv:2608.18352. - Grossman, Liu, Chen, Smith, Borcea and Chen (2026), [How Generative AI Disrupts Search: An Empirical Study of Google Search, Gemini, and AI Overviews](https://arxiv.org/abs/2604.27790), arXiv:2604.27790. - Aral, Li and Zuo (2026), [The Rise of AI Search: Implications for Information Markets and Human Judgement at Scale](https://arxiv.org/abs/2602.13415), arXiv:2602.13415. - Huang and colleagues (2026), [Answer Bubbles: Information Exposure in AI-Mediated Search](https://arxiv.org/abs/2603.16138), arXiv:2603.16138. --- This is the Markdown twin of https://underneath.agency/resources/content-business-risk-from-ai-search. The HTML page is canonical. Publisher: Underneath, https://underneath.agency/agent. Site index: https://underneath.agency/llms.txt.