Google's AI Overviews are killing the open web: what we're seeing in 2026

Google AI Overviews now appear on the majority of informational queries in 2026, and the data behind that fact is hard to spin. Pew Research found in March 2025 that when an AI Overview is present, users click on traditional search results about half as often as they do without one — and click on the citation links inside the overview itself only about one percent of the time. Two years after the May 2024 rollout, Google AI Overviews are the default search experience, not a feature. The traffic damage is no longer theoretical: it shows up in earnings calls, in lawsuits, and in our own Search Console logs. This post is our read on where AIO stands right now, who got hit, and what comes next. We cover the broader picture in our analysis of AI search in 2026; here, we zoom in on AIO.

TL;DR

• Google AI Overviews appear on a majority of informational queries as of 2026, up from around 14 percent at the May 2024 launch.
• Click-through to traditional search results when AIO is present is roughly half what it is without one — about 8 percent vs about 15 percent per Pew (March 2025). Clicks on the AIO's own citation links land closer to one percent.
• Affiliate sites, mid-tail publishers, and how-to content have taken the hardest hits. Brand-loyal and paywalled outlets are surviving better.
• Chegg sued Google in 2025 over AIO traffic impact. More litigation is likely.
• Publishers are pivoting toward direct traffic, newsletters, paywalls, and brand-search defense. The economics of the open commercial web have changed.

What AI Overviews actually are

An AI Overview is the AI-generated summary block that appears at the top of a Google search results page, above the traditional ten blue links. Google built it on Gemini and rolled it out broadly in the U.S. in May 2024. The block synthesizes information from multiple sources and surfaces a direct answer to the query, with source citations linked off to the right or below the summary.

In practical terms: where you used to see ten link results and click one, you now see a paragraph of text that answers your question without requiring a click. This is not the same as a featured snippet, which pulled a single passage from a single source. An AI Overview is a generated answer composed from multiple inputs, and the reader does not see whose work fed it unless they actively look.

When AIO became the default

The timeline matters because it explains why the impact in 2026 is worse than what most people read in 2024 and 2025.

May 2024: Google rolled out AI Overviews to all U.S. users at I/O. Early measurements placed AIO presence around 14 percent of informational queries (BrightEdge, May 2024). The launch was rocky, with widely shared failures (the glue-on-pizza answer, the eat-a-rock guidance) that made it look like a beta nobody wanted.

Through 2025: Google expanded AIO to more query categories, languages, and countries. Pew's March 2025 study captured a midpoint: AIO present on a majority of informational searches, with the click-through pattern described above.

2026: AIO is the default surface. It triggers on most informational queries, many commercial-investigation queries, and a growing share of how-to and comparison searches. Even transactional and navigational queries are starting to surface AI-generated context blocks.

This is not a feature publishers can avoid by targeting different query types. The surface has eaten the categories where the open web made its living.

What the data shows in 2026

The numbers we trust come from a handful of sources. Pew Research is the cleanest study on click behavior. Similarweb and Sistrix track publisher organic traffic at scale. Ahrefs and Authoritas have published query-cluster analyses showing where AIO triggers and what it does to the URLs underneath.

The headline findings from Pew Research (March 2025): when an AI Overview was present, users clicked on a traditional search result in roughly 8 percent of visits, compared to about 15 percent when AIO was absent — roughly half. The click-through rate on the AIO's own citation links was about 1 percent. AIO was already present on a majority of informational queries in early 2025, and the trend line since has only steepened.

Similarweb's 2025 publisher tracking, picked up across Press Gazette and Digiday, showed organic traffic declines in the 30 to 60 percent range on AIO-triggering query clusters for affiliate-heavy and how-to publishers.

Named publishers that have publicly discussed the impact:

Publisher	What's been reported
Forbes	Publicly discussed AIO-related traffic impact, particularly on commercial how-to and product queries
Vox	Editorial leadership has acknowledged search-referral declines tied to AI summaries
The Atlantic	Has publicly discussed AIO-related traffic impact
Mail Online / Daily Mail	Frequently cited in trade press as a high-profile case of AIO-driven traffic loss
Conde Nast (Wired)	Parent company cut staff in 2025 to 2026, citing search-traffic decline among other factors
HuffPost	Reported significant declines on informational and how-to coverage
Chegg	Sued Google in 2025 specifically over AIO traffic harm

We do not have a single audited industry-wide percentage for total publisher traffic loss to AIO, and we are skeptical of anyone who claims one. The honest version: declines are significant, they vary by category, and the median publisher in affiliate and how-to verticals is meaningfully worse off than they were eighteen months ago.

SparkToro's zero-click search research, from 2019 to 2023, already showed roughly 65 percent of Google searches ended without a click. AIO did not invent zero-click. It accelerated a trend that had been building for years.

Who got hit hardest

The damage is uneven, and the pattern is consistent enough to call.

Affiliate and review sites. "Best X for Y" queries are now answered directly in the overview. Mid-tail review sites have lost their job at the search level: the user gets the recommendation without clicking.

How-to and tutorial content. Recipe sites, fitness sites, DIY, coding tutorials. AIO summarizes steps from multiple sources into a single answer block. The original creator gets a citation if they're lucky.

Mid-tail publishers without brand recall. If a user does not know your name, they have no reason to click past the AI answer. Outlets built purely through SEO are exposed in a way brand-driven outlets are not.

Ad-supported news without a paywall. Page-view economics break when CTR collapses. The publishers laying off staff in 2025 to 2026 are almost all in this bucket.

Comparison shopping. Product comparison queries now resolve in the overview block, often with price and feature data scraped from review sites. Comparison publishers get the worst of both worlds: their content trains the answer, and they lose the click.

Doing OK by comparison: paywalled outlets with strong brand search (NYT, WSJ, FT), creator-driven publications with direct subscription bases, and outlets whose readers arrive through email or social rather than search. Their traffic is down too, but their revenue is not tied to the CTR AIO is compressing.

What we're seeing in our own search data

We run ToolDirectory.AI as an editorial review site, so we have a direct view from inside this. We are not going to publish exact numbers, but we will tell you the shape of the data.

Our category and "best of" pages saw meaningful CTR compression starting in mid-2024 and accelerating through 2025. The queries where we lost the most traffic are exactly the ones our methodology was built to win: "best AI tool for X," "X vs Y AI," comparison and recommendation queries. Those now trigger AI Overviews almost universally.

What still works: branded queries, long-tail comparisons AIO does not yet handle well, and pages linked from email and social rather than ranked into. Our curated collections and editorial blog are holding up because their referral mix was never search-pure.

What we are not seeing: any recovery on the queries we lost. This is the part that does not get said clearly enough in 2026. Publishers who assumed AIO impact would normalize as Google refined the product have been waiting for a rebound that has not arrived.

Google's defense, and where it falls short

Google's public position is consistent and worth taking seriously before disagreeing with it.

The argument: AIO drives higher-quality clicks because users who do click through are more informed and more likely to engage. Total search volume is up, so even with lower per-query CTR the absolute click count to publishers is allegedly not falling as dramatically as critics claim. AIO surfaces a broader range of sources than the top three blue links did, so smaller publishers may actually get more visibility.

Where this falls short empirically: the "higher-quality clicks" claim has not been demonstrated with public data publishers can verify, and reports from large publishers do not show a per-click engagement uplift that compensates for the CTR collapse. The "total search volume is up" argument is true and largely beside the point — if overall searches grow ten percent but per-search CTR is cut in half, publishers still lose net referrals. And the "broader source surfacing" point cuts both ways: AIO can cite a long-tail blog that would never have ranked top three, but citation is not a click. The whole design of the interface is that users do not need to click to get value.

We respect that Google is genuinely trying to balance user experience against publisher economics. We do not believe the current balance is sustainable for the open web, and the empirical record from 2024 to 2026 supports that read.

What publishers are actually doing

The responses we are watching:

Lawsuits. Chegg's 2025 antitrust filing against Google is the marquee case but unlikely to be the last. The DOJ's 2024 ruling that Google is a search monopolist is a separate case, but the legal context matters: publishers now have a recent monopolist finding to point at when arguing harm.

Direct traffic investment. Newsletter operations, app builds, social-first content. Outlets are spending against channels they own. Substack, beehiiv, and the broader newsletter stack are direct beneficiaries.

Paywall acceleration. Outlets that hesitated to paywall are moving faster. If your ad-supported model depended on search referral CTR, that model is broken, and subscription is the only structural fix.

Brand-search defense. Publishers are investing in being the named entity users search for, rather than the page that ranks for a topic. This is harder, slower, and more expensive than SEO. It is also durable in a way ranking-based traffic is not.

Quiet acceptance. Some sites are closing down. Mid-tail affiliate operations that cannot pivot to a brand or subscription model are exiting the market. The pattern echoes what we have documented on the tools side in our AI Graveyard: a category gets reshaped by a platform shift, and the under-differentiated middle gets squeezed first.

What this means for the open web

We are a content business. We are not neutral on this. Our editorial work — the Top 100 list, our category pages, our reviews — exists in the same ecosystem AIO is reshaping.

The open commercial web was built on a deal. Publishers create content, Google indexes it and sends users back via the SERP, publishers monetize the visit. The deal worked because the index was a directory, not an answer. AIO turns the index into an answer, and the deal breaks.

For AI tool buyers: be skeptical of any "best AI tool" answer you read in an AI Overview. The summary is composed from sources that may be outdated, SEO-optimized rather than editorially honest, or missing the alternative that would actually fit your need better. Editorial review sites like ours exist to make judgment calls AIO cannot. Click through to a human review when the stakes are real.

There is also a structural irony worth naming. The companies making AIO possible are also the search alternatives. Perplexity and ChatGPT have built genuine traction as Google replacements. And Gemini, Google's own conversational assistant, is in many ways AIO unbundled from the SERP. The companies inheriting Google's search dominance may, in the medium term, be the ones least dependent on the open web for their training data and answers.

That is not a comfortable future for anyone who publishes for a living.

Frequently asked questions

What are Google AI Overviews? AI Overviews are AI-generated summary blocks that appear at the top of Google search results, above the traditional blue links. Built on Gemini, they synthesize information from multiple sources and present a direct answer to the query, with source citations. Google rolled them out broadly in the U.S. in May 2024.

How much have AI Overviews reduced publisher traffic? Pew Research (March 2025) found that when an AI Overview was present, users clicked on a traditional search result in roughly 8 percent of visits, compared to about 15 percent when no AIO was present — roughly half. The click-through rate on AIO citation links themselves was about 1 percent. Publisher-level traffic declines vary widely by category, with affiliate and how-to sites reporting drops in the 30 to 60 percent range on AIO-triggering queries.

Did AI Overviews cause zero-click search? No, but they accelerated it. SparkToro's research showed roughly 65 percent of Google searches already ended without a click before AIO existed. AIO concentrated and worsened a pre-existing trend rather than inventing one.

Is Google being sued over AI Overviews? Yes. Chegg filed an antitrust suit against Google in 2025, citing AIO's impact on its traffic. The case is distinct from the DOJ's 2024 ruling that Google is a search monopolist, but the legal context overlaps.

Which publishers have publicly reported AIO traffic impact? Forbes, Vox, The Atlantic, Mail Online (Daily Mail), HuffPost, and Chegg have all discussed AIO-related traffic declines in public statements, trade press, or legal filings. Conde Nast (Wired's parent) cited search-traffic decline as a contributing factor in 2025 to 2026 staff cuts.

How is Google defending AI Overviews? Google argues AIO produces higher-quality clicks, that total search volume growth offsets per-query CTR declines, and that AIO surfaces a broader range of sources than the top three blue links. Publisher data has not validated the higher-quality-click claim, and the volume argument does not compensate for the magnitude of CTR drops.

What can publishers do about AI Overviews? The credible responses are direct traffic investment (newsletters, apps, social), paywall acceleration, brand-search defense, and pivoting away from SEO-pure business models. Lawsuits are happening but will take years to resolve. Sites that cannot pivot are exiting.

Will AI Overviews get better for publishers over time? We have not seen evidence of recovery between 2024 and 2026. The structural incentive for Google is to keep users on the SERP, which works against any pro-publisher recalibration.

What we're betting on

The honest version: editorial review and human judgment remain the highest-value layer in a world where AI summaries handle the rest. AIO can tell you the consensus answer. It cannot tell you which tool we actually recommend after testing, which ones got worse this quarter, or which alternative nobody is talking about yet.

That is what we are building toward. If you want our work:

• The Top 100 AI tools list, our editorial ranking, updated continuously.
• Curated collections by use case and category.
• Our methodology, so you can see how we make the calls.
• The blog for ongoing analysis, including the state of AI search in 2026 companion piece.

We are not pretending AIO is going away. We are betting that what it cannot do still matters.

— The ToolDirectory.AI editorial team