Google AI Overviews Traffic Drop: What a 40% Click Loss Means for Indie Hackers
Google AI Overviews traffic can leave while the ranking stays. You can be the first organic link and still lose the click when the answer is already on the results page.
A randomized study finds that an AI Overview cuts outbound organic clicks by 39.8% when it appears. RevScope’s post on 9 November 2026 carried that result under the headline “Google’s AI Overviews cut clicks to websites by 40%”. The Hacker News thread on 18 November stood at about 677 points and about 526 comments. Counts on that thread move. Part of it was glad to see SEO lose the click. Part of it was people who run a site on those visits.
This was written on 19 November 2026. The click study is a working paper. Founder figures, and one consultant’s client totals, are what those people reported. RevScope sells analytics. Ahrefs and DataForSEO sell SEO data.
What the Google AI Overviews traffic study actually found
Saharsh Agarwal and Ananya Sen posted “The Impact of Google AI Overviews on Publisher Traffic and User Experience: Evidence from a Field Experiment” on SSRN on 3 April 2026. Agarwal is at the Indian School of Business. Sen is at Carnegie Mellon Heinz College. They last revised the paper on 8 July 2026. It is a working paper, and it has not been peer reviewed.
The experiment used a custom Chrome extension and random assignment. PPC Land says 1,065 people in the United States, recruited through Prolific, produced 68,089 unique searches. Those accounts describe a desktop-only sample from January and February 2026.
The abstract states it in one sentence: “Conditional on appearing, AIOs reduce outbound organic clicks by 39.8% and increase zero-click searches by 34.5%, without affecting sponsored clicks or overall search frequency.” AIOs means AI Overviews. The unpaid click to a site fell. Sponsored clicks held, and so did the number of searches.
Engagement on the sites, conditional on the click, was not different. The authors report “no measurable improvements in perceived search quality”. The AI Mode condition suggests “even greater reductions in traffic”.
Search Engine Journal, on the revised paper, says the lost clicks were not lower quality. About 18% of visits ended within 10 seconds in both conditions. An earlier version of the paper had reported 38%.
Do AI Overviews reduce clicks to websites?
When an overview is on the page, the paper’s answer is yes. PPC Land reports 0.37 clicks per search with an AI Overview and 0.62 without one. The probability of a zero-click search was 0.73 with an overview and 0.54 without. Overviews triggered on about 41% of queries. More than 87% of those overviews sat in the top position. PPC Land says that removing the top AI Overview meant an 88% relative increase in outbound clicks.
Informational queries were 71% of the sample, and they drove the result. Estimates for navigational queries and for transactional queries were too noisy to use.
The 39.8% applies when an overview appears. RevScope writes that across all searches the study authors put the loss at about 18.5% of the organic clicks a site gets from Google. That 18.5% is RevScope’s account of the authors. Pivot News summarized the same post.
RevScope sells analytics, and it discloses that most of its datasets come from companies that sell SEO data. DataForSEO, which sells SEO data, is the source of a 20-million-keyword panel in that post. From the panel, RevScope reports the share of results with an AI Overview moving from 5.93% in September 2024 to 61.43% in August 2026. Desktop click-through on the first organic result moved from 17.83% in September 2025 to 7.43% in August 2026, and the mobile rate from 25.17% to 20.80%. Those figures are the panel’s, reported by RevScope, and they are separate from the randomized experiment.
Google disputes aggregate traffic declines. On 6 August 2025, Liz Reid, VP Head of Google Search, wrote: “Overall, total organic click volume from Google Search to websites has been relatively stable year-over-year.” She also says Google is sending “slightly more quality clicks”, with traffic shifting toward forums, videos, and pages that offer “a unique perspective or a thoughtful first-person analysis”. TechCrunch reported the same day that Google denies its AI search features are killing website traffic. Reid’s post publishes no figures. She is describing the total. The experiment is describing outbound clicks when an overview is present, on desktop in the United States in January and February 2026. A site can lose those clicks while that total, in Google’s telling, stays level.
The indie hackers losing traffic
Nicolas Bouliane, who posts on Hacker News as nicbou, has run All About Berlin since 2017 and has lived from the site since 2020. On 26 May 2026 he wrote that the site had lost 75% of its traffic to Google AI Overviews. The post shows 70% struck through and replaced with 75%. On 18 November, nicbou wrote: “My traffic is down over 80%.” Both figures are self-reported, from his own site on two dates. The study’s 39.8% is a different measurement.
HN user jgalt212 wrote: “Our content-based market efforts (i.e. SEO clicks) are down more than 80% over the last 2 years.” That is self-reported.
Tomislav Glogovac, an SEO consultant, posted on 1 April 2026 about nine SaaS clients. In his aggregate, top-of-funnel organic traffic was down 34%. Bottom-of-funnel pages (demos, pricing, comparisons, and branded queries) were down about 6%. Organic pipeline was down about 11%. Definitional queries were down 70% to 80%. Situational and contested queries were up 5% to 20%. Those percentages are his, from nine clients, which is a different sample from the 1,065 randomized participants.
The ranking that holds is the hard case. Whether indie hacking is dead is partly an argument about search posts that were supposed to compound. Whether the work still pays depends on a customer after that click.
Does traffic from AI search convert better?
HN user robeym wrote: “our search traffic took a dip, but our sign up rate and conversions are significantly higher.” That line is self-reported, and it does not include the rates.
RevScope, citing Ahrefs, says AI assistants sent 0.28% of traffic across 74,752 sites in March 2026, about 3.5 million visits against 345 million from Google. Ahrefs sells SEO software. On Ahrefs’ own site, RevScope says AI visitors were 0.5% of visitors and 12.1% of signups. That pair is one site, reported by a company that sells analytics.
The revised paper, via Search Engine Journal, does not class the missing clicks as low quality: about 18% of visits ended within 10 seconds in both conditions. Reid’s “slightly more quality clicks” has no table under it. robeym’s rate and nicbou’s drop both sit on the same thread. A higher rate on a smaller top of funnel can still be a smaller business. These sources do not show either side’s revenue.
Is SEO dead in 2026?
HN user pcarolan wrote: “SEO killed the web, AI killed SEO, and so it goes…”
HN user wraptile wrote: “As someone who did a fair amount of SEO in the past - good.”
HN user vicentwu wrote: “Google needs to find a new way to distribute its revenue to the sites it has crawled”.
HN user aleaph wrote that if search engines do not drive traffic, “what’s the point of allowing them to index your site.”
pcarolan and wraptile welcome the fall of SEO. vicentwu wants the crawled sites paid. aleaph asks why indexing should continue if the visit does not come back. The study counts clicks. It does not pick among those endings.
Glogovac’s clients are where the channel splits, on his numbers. Definitional queries fell 70% to 80%. Situational and contested queries rose 5% to 20%. The study’s loss was driven by informational queries. His definitional drop sits near that result, from a different dataset. Awesome-list backlinks, and the AI-written pull requests aimed at them, are a separate argument about an old way of collecting mentions. Whether you need an audience to indie hack is the neighbouring question once a ranking is a weak promise of a visit.
How do you get cited in Google AI Overviews?
RevScope’s description of what people are calling GEO is short. It “looks mostly like SEO plus two additions”: “being present on the third-party sites AI answers draw from”, and “writing pages that are easy to quote”. That reading comes from a company that sells analytics.
On 13 September 2026, 13jawad described a site whose traffic slid while the rankings held. He moved the direct answer into the first 100 to 150 words. He tracks 15 queries each month across AI Overviews, ChatGPT, and Perplexity. He says Google’s May 2026 guidance, as he reads it, was that optimizing for AI Overviews is still fundamentally SEO. That is his paraphrase. These sources do not include a Google page from May 2026 that says it.
The extension study tested the click when an overview is present. It did not test a lead answer, a third-party mention, or a log of 15 queries. A page that is easy to quote can still be a page the reader does not open. Bouliane’s struck-through percentage is one report of that. Glogovac’s clients showed the other cut: definitional queries fell, and situational and contested queries rose. Indie hacking in 2026 is the wider year these graphs sit in. What indie hackers now budget for AI coding tokens is a separate cost, the bill for building the page.
A short timeline
| When | What is on the record |
|---|---|
| 6 August 2025 | Liz Reid: total organic click volume from Google has been relatively stable year over year. No figures in the post. |
| 3 April 2026 | Agarwal and Sen post the working paper on SSRN. Last revised 8 July 2026. |
| 26 May 2026 | Bouliane: All About Berlin has lost 75% of its traffic to AI Overviews. 70% is struck through. |
| 9 November 2026 | RevScope publishes "Google's AI Overviews cut clicks to websites by 40%". |
| 18 November 2026 | Hacker News thread: about 677 points and about 526 comments. |
Reid's post is on the Google blog, covered the same day by TechCrunch. The paper is on SSRN. The sample is in PPC Land, and the revision is in Search Engine Journal. Bouliane's post is Death by AI. His November line is on Hacker News. RevScope's post is the 9 November piece, summarized by Pivot News. The thread is Hacker News. Glogovac's client note, dated 1 April 2026, is on his site. 13jawad's post, dated 13 September 2026, is on Indie Hackers.
What do you think? How much of your traffic still comes from Google, and what are you changing?
FAQ
Do AI Overviews reduce clicks to websites?
When an AI Overview appears, the working paper by Saharsh Agarwal and Ananya Sen says outbound organic clicks fall by 39.8% and zero-click searches rise by 34.5%, with no change in sponsored clicks or in how often people search. PPC Land's account of that experiment, 1,065 people in the United States and 68,089 unique searches, reports 0.37 clicks per search with an overview and 0.62 without, and a zero-click probability of 0.73 versus 0.54. RevScope writes that across all searches the authors put the loss at about 18.5% of the organic clicks a site gets from Google. Liz Reid, VP Head of Google Search, wrote on 6 August 2025 that total organic click volume to websites has been relatively stable year over year. The paper is a working paper, not peer reviewed, on desktop searches in the United States in January and February 2026.
Is SEO dead in 2026?
The Hacker News thread does not settle it. pcarolan wrote that SEO killed the web and AI killed SEO. wraptile, who had done a fair amount of SEO, called the outcome good. vicentwu said Google needs a new way to pay the sites it crawls. aleaph asked why a site should stay open to indexing if search sends no traffic. Tomislav Glogovac, an SEO consultant, reported nine SaaS clients on 1 April 2026. In his aggregate, definitional queries were down 70% to 80%, situational and contested queries were up 5% to 20%, top-of-funnel organic traffic was down 34%, and bottom-of-funnel pages were down about 6%. Those client figures are self-reported. They describe a split inside search, which is a different claim from the channel being gone.
Does traffic from AI search convert better?
HN user robeym wrote that search traffic dipped while the signup rate and conversions were significantly higher. That line is self-reported, and it does not include the rates. RevScope, citing Ahrefs, says AI assistants sent 0.28% of traffic across 74,752 sites in March 2026, about 3.5 million visits against 345 million from Google. On Ahrefs' own site, AI visitors were 0.5% of visitors and 12.1% of signups. Ahrefs sells SEO software, and that signup share is one site. Search Engine Journal's account of the revised paper says the lost clicks were not lower quality: about 18% of visits ended within 10 seconds in both conditions, where an earlier version had said 38%. Engagement after a click was not different. Google says it sends slightly more quality clicks and published no figures for that claim.
How do you get cited in Google AI Overviews?
RevScope says the work people call GEO looks mostly like SEO, plus being present on the third-party sites AI answers draw from, and writing pages that are easy to quote. RevScope sells analytics. On 13 September 2026, Indie Hackers user 13jawad wrote that traffic slid while rankings held. He moved the direct answer into the first 100 to 150 words, and he tracks 15 queries a month across AI Overviews, ChatGPT, and Perplexity. He says Google's May 2026 guidance, in his paraphrase, was that optimizing for AI Overviews is still fundamentally SEO. The randomized study measured clicks when an overview appears. It did not test these tactics.
Did Google say AI search is reducing website traffic?
On 6 August 2025 Liz Reid wrote that total organic click volume from Google Search to websites has been relatively stable year over year. She also said Google is sending slightly more quality clicks, with traffic shifting toward forums, videos, and pages that offer a unique perspective or a thoughtful first-person analysis. TechCrunch reported the same day that Google denies AI search features are killing website traffic. The post publishes no figures. The later working paper finds a large drop in outbound organic clicks when an overview is present. It does not publish Google's aggregate totals either.