--- title: "What do lost clicks to AI answers mean for pipeline and revenue?" description: "Fewer search clicks mean fewer prospects entering your funnel, and AI answers appear most on the research and comparison searches where buying starts." canonical: "https://underneath.agency/resources/ai-answers-pipeline-revenue" 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 # What do lost clicks to AI answers mean for pipeline and revenue? They mean fewer prospects at the top of your funnel, and AI answers show up most on the research-stage searches where buying starts. Studies of Google’s AI features find drops from about 5% of Wikipedia’s search visits to nearly half of result clicks on pages with an AI summary. No study has yet traced those lost clicks through to a company’s leads, pipeline or revenue. ## The short version 1. Researchers estimate Google’s AI Overviews cut search visits to English Wikipedia by 5.45%, about 100.27 million visits a month (Khosravi and Yoganarasimhan, 2026). 2. In a Pew Research Center panel of 900 US adults, people clicked a result on 8% of pages with an AI Overview and 15% without. 3. In a 2025 test, Google showed AI Overviews on 92% of product question searches but only 17.4% of plain product searches, so research-stage traffic is most exposed. 4. In a 2026 US experiment with 1,100 people, making AI Mode the default cut clicks to outside websites by 18.8 percentage points. 5. Not all content loses: Reddit discussion communities gained 12.0% more comments after AI Overviews launched, though AI Mode later erased much of the gain. ## How do lost search clicks reach your pipeline? Every search visit you lose is one fewer person who can enter your funnel. That is the plain logic set out by [Khosravi and Yoganarasimhan](https://arxiv.org/abs/2602.18455), who measured what Google’s AI Overviews did to Wikipedia. Many sites, they note, depend on search visits for advertising, subscriptions, lead generation and commerce. For them, “a decline in search traffic reduces both immediate monetization opportunities and the number of users entering the top of the conversion funnel.” Losses can compound: on Wikipedia, roughly 40% of all traffic comes from clicks between articles. If a search visit never happens, the pages that visitor would have read next are lost too. The authors say the total effect could therefore be larger than they measured. A related question is [whether AI Overviews end browsing sessions](https://underneath.agency/resources/ai-overviews-end-browsing-sessions) altogether. For a business site, the equivalent is the visitor who lands on a guide and then reads a product, pricing or demo page. No study has measured that chain for company websites. ## How large are the click losses so far? Between a few percent and about half, depending on the site, the measure and the AI feature. | Study | Who and when | What they found | |---|---|---| | Khosravi and Yoganarasimhan | English Wikipedia, May to December 2024 | Search visits fell 5.45% against German editions and 4.82% against French | | Chapekis and colleagues, Pew Research Center | 900 US adults, March 2025 | A result was clicked on 8% of AI Overview pages against 15% of others | | Wang and colleagues | 1,100 US users, March 2026 | Hiding AI Overviews raised outside clicks by 8.8 points; forcing AI Mode cut them by 18.8 | The Wikipedia figure is an average over all searches, including many that showed no AI Overview. The authors say the effect on searches that did show one is likely larger. Projected over a year, their estimate comes to about 1.20 billion fewer search visits. We weigh all the evidence in [how much AI Overviews cut clicks](https://underneath.agency/resources/do-ai-overviews-reduce-clicks). The [Pew study](https://arxiv.org/abs/2608.04831) is observational, so it shows an association, not proof of cause. The [field experiment by Wang and colleagues](https://arxiv.org/abs/2608.18352) is the strongest causal evidence. It ran for one week, though, and its sample skewed young: 75% were under 45. ## Which searches put your pipeline most at risk? Research-stage searches: comparisons and questions draw AI answers far more often than plain product searches. [Grossman and colleagues](https://arxiv.org/abs/2604.27790) ran product searches through Google in December 2025. An AI Overview, the AI summary at the top of Google’s results, appeared on 17.4% of plain product searches. It appeared on 88.2% of product comparisons and 92% of product questions, which were written by AI from real searches. The authors conclude that AI search matters more in the “consideration or research stages than the final purchase stage.” [Our own study](https://underneath.agency/research/ai-overviews-frequency-study) of US commercial keywords found the same tilt. 60.8% of 800 commercial keywords showed an AI Overview, rising to 98.1% of keywords phrased as questions. Searches that showed a map of local businesses were the exception: their predicted chance was 30.7%, against 78.6% for searches without one. For a business, the risky searches are the early ones, such as “how do I choose a…” or “X vs Y”. Those are often where a buyer first met your brand. ## Can you put a dollar figure on the loss? Not reliably yet: the published dollar figures are illustrations, not measured losses. Khosravi and Yoganarasimhan priced their Wikipedia estimate as if every lost visit were a page carrying ads. That gave $0.901 million to $3.090 million a month. Wikipedia carries no ads, so this shows scale, not money anyone lost. [Xu and colleagues](https://arxiv.org/abs/2605.14021) looked at the pages AI Overviews cite, across 55,393 trending searches in March and April 2026. 50.63% of the cited pages displayed ads. They add that sites earning through subscriptions or direct conversions lose in the same way when the answer arrives without a click. To estimate your own exposure, multiply the search visits at risk by your visit-to-lead rate and your lead value. Use your own numbers; no study supplies them. ## Does every kind of content lose clicks? No: discussion and advice content gained engagement from AI Overviews, at least until AI Mode arrived. [Zhang, Cui and Zhang](https://arxiv.org/abs/2605.16428) tracked 105,012 Reddit communities through Google’s rollout. Communities that AI Overviews could cite saw daily comments rise by 12.0% and commenters by 12.4%, compared with communities Google bars from AI Overviews. The gain was largest for experience content, such as opinions, advice and personal stories. There the effect on comments was 2.3 times larger than for factual communities. The authors’ explanation: a summary can replace a fact, but it points people toward experiences they want to read in full. That advantage faded once Google launched AI Mode, its conversational search. For experience communities, the extra gain in commenters fell by 59%, and for comments it disappeared. These are Reddit comments, not visits or sales, so treat this as a pattern, not a forecast for your site. ## What if AI Mode becomes Google’s default? Then the losses would likely grow: forcing AI Mode cut outside clicks by 18.8 points in a one-week experiment. In that experiment, the share of people clicking through fell for news sites, Reddit and Wikipedia alike. People also ran 0.92 fewer search sessions a day, and 11.2 percentage points more of them tried a rival search engine. In [our comparison of Google’s two AI surfaces](https://underneath.agency/research/ai-mode-vs-ai-overviews-study), AI Mode answered all 400 searches we tested. We look at the traffic risk in more detail in [our article on AI Mode as Google’s default](https://underneath.agency/resources/ai-mode-default-traffic-loss). ## What should you do about it? Re-plan your funnel for fewer research-stage visits, and work to be named inside the AI answer itself. Practical steps: 1. Find your exposed traffic. List the question and comparison searches that feed your pipeline, and check which already show an AI Overview. 2. Size the risk in your own numbers. Lost visits times conversion rate times lead value gives a range to plan around. 3. Aim to be cited, not just ranked. In [our citation study](https://underneath.agency/research/ai-overview-citations-study), 28.7% of AI Overview citations were page-one results, so ranking alone does not get you into the answer. 4. Invest in what a summary cannot replace. First-hand experience, advice and opinion drew more engagement in the Reddit study than plain facts did. 5. Track influence beyond clicks. Branded searches, direct visits and what new leads say about how they found you all capture effects that referral data misses. If you want help working out where AI answers touch your pipeline, see our [generative engine optimization service](https://underneath.agency/services/generative-engine-optimization). ## What does the research not tell us yet? Nobody has yet measured the effect on business pipeline or revenue directly. The main gaps: - Pipeline effects. No study follows lost AI clicks through to leads, deals or revenue for B2B or consumer companies. - Dollar values. The only dollar figures are illustrations built on assumed ad rates. - Click quality. Google argues that AI features improve the quality of clicks that remain, but the studies here did not test that. - AI assistants. Most evidence covers Google’s AI features; how ChatGPT and other assistants change visits to business sites is barely measured. - Time. The studies cover short windows between 2024 and 2026, and these products change monthly. ## Frequently asked questions ### Do AI Overviews reduce website traffic? Yes, on the best evidence. In a 2026 field experiment with 1,100 US users, hiding AI Overviews raised clicks to outside websites by 8.8 percentage points. ### How much revenue do websites lose to AI search? No study has measured it. The closest is an illustration for Wikipedia of $0.901 million to $3.090 million a month in ad value, for a site that carries no ads. ### Which marketing content is most at risk from AI answers? Factual answers to research questions. Google showed AI Overviews on 88.2% of product comparison searches in one test, while experience-based discussion content gained engagement. ### Are the clicks that remain worth more? Possibly, but it is unproven. Google says AI features improve click quality, and none of the studies reviewed here has tested that claim. ## Sources - Khosravi, M. and Yoganarasimhan, H. (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. - Chapekis, A., Lieb, A., Shah, S. and Smith, A. (2026), [Investigating Click Behaviors On Google Search Result Pages That Produce an AI Overview](https://arxiv.org/abs/2608.04831), arXiv:2608.04831. - Wang, S. T., Gleason, J., Bart, Y., Wilson, C. and Metaxa, D. (2026), [AI in Search Reduces Publisher Referrals Without Improving User Experience: Experimental Evidence](https://arxiv.org/abs/2608.18352), arXiv:2608.18352. - Grossman, R., Liu, S., Chen, M. K., Smith, M., Borcea, C. and Chen, Y. (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. - Xu, H., Iqbal, U. and Montgomery, J. M. (2026), [Measuring Google AI Overviews: Activation, Source Quality, Claim Fidelity, and Publisher Impact](https://arxiv.org/abs/2605.14021), arXiv:2605.14021. - Zhang, P., Cui, R. and Zhang, D. J. (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. - Underneath (2026), [When does Google show an AI Overview? 1,248 US searches](https://underneath.agency/research/ai-overviews-frequency-study) - Underneath (2026), [AI Mode vs AI Overviews: how different are the sources?](https://underneath.agency/research/ai-mode-vs-ai-overviews-study) - Underneath (2026), [Do AI Overviews cite the pages that rank? 4,051 citations](https://underneath.agency/research/ai-overview-citations-study) --- This is the Markdown twin of https://underneath.agency/resources/ai-answers-pipeline-revenue. The HTML page is canonical. Publisher: Underneath, https://underneath.agency/agent. Site index: https://underneath.agency/llms.txt.