What if Google AI Overviews answered every query? Traffic maths

Today an AI summary sits on about one Google search in five. Put it on all of them and a typical Indian brand site loses close to half its organic clicks.

The domed main reading room of the Library of Congress with rows of desks
Photo: Carol M. Highsmith, Public domain, via Wikimedia Commons

If Google AI Overviews answered every query, clicks from Google to websites would roughly halve on the searches that are informational today, and fall less on product and brand searches. Pew found people clicked a result on 8% of visits with an AI summary against 15% without one, and Ahrefs measured a 58% drop in clicks for the top-ranked page where an Overview appears. For a typical Indian brand site that earns most organic visits from informational content, my model puts the loss at about 45% of organic clicks, which would cost several lakh rupees a month to replace with paid search.

Key takeaways
  • 18% of Google searches carried an AI summary in Pew's March 2025 sample, and 60% of who/what/when/why questions did.
  • Clicks on a normal result fall from 15% to 8% of visits when a summary appears; links inside the summary get about 1%.
  • Ahrefs measures a 58% drop in position-one CTR where an Overview shows.
  • A content-led Indian brand would lose about 45% of organic clicks; a product-and-brand site about 30%.
  • Replacing 22,500 lost clicks a month at ₹25 a click costs about ₹5.6 lakh a month.

    Calculated in this post from the weighted-loss model below: 50,000 example clicks × 45% = 22,500, and 22,500 × ₹25 (an assumed cost per click) ≈ ₹5.6 lakh. The 58% informational loss is Ahrefs' figure.

This is the tier-one what-if that is already half true. Google says AI Overviews reached 2 billion monthly users by July 2025. AI Mode launched in English in India in 2025, added Hindi in September and seven more Indian languages in October.

The question is not whether it spreads, but what happens to your traffic at 100%. So here is the maths, keyword type by keyword type.

The story

In September 2025, Penske Media, the family-owned company behind Rolling Stone, Billboard and Variety, sued Google in federal court in Washington, D.C. It was the first time a major US publisher took Google to court over AI Overviews.

The complaint put numbers on the fear. About 20% of Google searches that linked to Penske's sites already showed an AI Overview, a share it expected to rise. Its affiliate revenue, the commission it earns when a reader clicks through and buys, had fallen by more than a third from its peak at the end of 2024 as search traffic declined.

Google's spokesperson Jose Castaneda answered that Overviews give users a better experience and send traffic to a wider range of sites, and called the claims meritless. The case is about one fifth of searches. This post asks what happens at five fifths.

Claims Journal (Reuters): Rolling Stone, Billboard owner Penske sues Google over AI Overviews (15 Sep 2025)
18%of US Google searches showed an AI summary in March 2025 (Pew)
8% vs 15%visits with a click on a result, with vs without a summary
−58%position-one CTR where an AI Overview appears (Ahrefs, Dec 2025)
₹16,581 crspent on paid search in India in 2025 (dentsu)

How many searches show an AI Overview today?

About 18% of Google searches showed an AI summary in Pew Research Center's March 2025 sample, but the share depends heavily on how the query is phrased. Pew tracked 900 US adults' real browsing, covering 68,879 Google searches, of which 12,593 produced a summary.

Query type (Pew, March 2025)Searches with an AI summary
One or two words8%
Full sentence with a noun and a verb36%
Ten words or more53%
Starts with who, what, when or why60%
All searches18%

So the Overview is already the default for questions. What the "every query" scenario adds is the short, commercial and branded searches:

  • Short generic searches such as "running shoes", where Pew saw summaries on only 8% of one- or two-word queries;
  • Commercial searches such as "term insurance", where people are comparing before they buy;
  • Brand searches, your own name, typed by people who already want you.

Those are where most Indian businesses make money, which is why the what-if matters.

How much do AI Overviews cut clicks?

Where an AI summary appears, clicks to ordinary results fall by roughly half. Pew found people clicked a traditional result on 8% of visits to a results page with a summary and 15% without.

They clicked a link inside the summary on just 1% of visits, and ended their browsing session on 26% of summary pages against 16% of others. The summary does not redirect attention; it ends the search.

100 visits to a Google results page (Pew, 2025)No AI summary15click a result16 end the session on this pageWith an AI summary88 click a result, 1 clicks inside the summary26 end the session on this pageClicks to ordinary results fall from 15 to 8: a cut of about 47%
Pew's browsing panel in one picture. The accent bar is the traffic a website actually gets.

Ahrefs measured the same thing from the publisher side, using Google Search Console data for 300,000 keywords. In December 2023, before Overviews, the top result on keywords that later got an Overview earned a desktop click-through rate of 7.3%. By December 2025 it was 1.6%.

Ahrefs compared that with informational keywords that never got an Overview, whose CTR also fell, from 7.6% to 3.9%, and attributed the extra fall to the Overview: about 58% at position one.

Change in click-through rate when an AI Overview appears, by ranking position
See the data as a table
CTR change (%)
Position 1-58
Position 2-50.8
Position 3-46.4
Position 4-38.8
Position 5-32.6
Position 6-30.5
Position 7-29.7
Position 8-28.8
Position 9-29.7
Position 10-19.4

Source: Ahrefs, Update: AI Overviews reduce clicks by 58% (Feb 2026) · Desktop CTR, December 2025 vs a forecast based on non-Overview informational keywords. Vendor research; correlation, not a controlled experiment.

The long research reading room of the New York Public Library
A reading room is where people go for depth. Those are the visits AI answers take least of, which is why the Wikipedia effect is small.Photo: Diliff, CC BY 2.5, via Wikimedia Commons

Is the click loss really that big?

Not everywhere: the most careful causal study so far finds a much smaller effect on Wikipedia, about 5%, which tells you the loss depends on what kind of page you run. The big numbers come from comparing results pages; the small one comes from measuring a whole site's traffic before and after the rollout.

Two kinds of evidence, two different questions

StudyWhat it comparesEffect found
Pew (2025)Results pages with and without a summaryResult clicks 15% → 8% of visits
Ahrefs (2026)Position-one CTR on Overview vs non-Overview keywords−58%
Khosravi and Yoganarasimhan (2026)Whole-site Wikipedia traffic, staggered country rolloutAbout −5%
Impact of AI Search Summaries on Website Traffic: Evidence from Google AI Overviews and WikipediaMehrzad Khosravi and Hema Yoganarasimhan · arXiv 2602.18455 · 2026

The authors used the staggered country-by-country rollout of AI Overviews as a natural experiment, comparing external search traffic to English Wikipedia articles with traffic to the same articles in German and French, where Overviews arrived later. In a difference-in-differences design, default Overview availability reduced English Wikipedia search traffic by about 5% (5.45% and 4.82% in their two main estimates in the latest version). The limit: Wikipedia is an unusual site. It is often the source the Overview itself draws on, its readers often want depth, and the effect on a brand blog or a product page may be very different.

Read the paper (free)

The two kinds of evidence fit together once you think in pages, not searches. A page that answers a simple factual question ("what is the GST rate on gold") loses most of its clicks, because the Overview answers it. A page people go to for depth, tools, prices or trust loses less.

A page that only answers a question is now competing with the answer itself.

Most brand blogs in India are full of the first kind. I go deeper into which sources Overviews pick in how AI Overviews and ChatGPT choose sources.

Open drawer of a library card catalogue full of index cards
A card catalogue sent every reader to a shelf. Classic search did the same with ten blue links; an AI answer removes the walk.Photo: TBurmeister (WMF), CC BY-SA 4.0, via Wikimedia Commons

What would it do to an Indian brand's traffic?

With these example inputs (an Indian brand site with 50,000 organic clicks a month, mostly from informational content), answering every query with an Overview would cut organic clicks by about 45%, to roughly 27,500. Here is the model, with the assumptions marked.

Calculated in this post from the weighted-loss model below (60% / 25% / 15% traffic mix, with losses of 58% from Ahrefs and my 35% and 10% estimates): 0.60 × 0.58 + 0.25 × 0.35 + 0.15 × 0.10 ≈ 0.45, and 50,000 × (1 − 0.45) ≈ 27,500. The traffic mix is an assumption, not a measured average for Indian sites.

The weighted-loss model

  1. Traffic mix. 60% of clicks from informational keywords, 25% from commercial ones ("best term plan for 30 year old"), 15% from brand searches. This is my assumption for a content-led brand; check your own Search Console.
  2. Informational loss. 58%, Ahrefs' position-one figure, since most of your informational clicks come from top positions.
  3. Commercial loss. 35%, my estimate, close to Ahrefs' positions four and five. People comparing products still click, but fewer.
  4. Brand loss. 10%, my estimate. Someone searching your name usually wants your site.
  5. Weighted loss. \( 0.60\times0.58 + 0.25\times0.35 + 0.15\times0.10 = 0.348 + 0.088 + 0.015 \approx 0.45 \).
  6. Business impact. 22,500 fewer clicks a month. At a 1.5% lead rate that is about 340 fewer leads; replacing the clicks with paid search at ₹25 a click (an assumption, CPCs vary widely) would cost about ₹5.6 lakh a month.

    Calculated in this post from the 22,500 lost clicks above (the 45% weighted loss, built on Ahrefs' 58%), with these example inputs: a 1.5% lead rate (22,500 × 0.015 ≈ 340) and ₹25 a click (22,500 × 25 ≈ ₹5.6 lakh). Both rates are assumptions, not sourced figures.

\[ L = \sum_i w_i\,d_i = 0.60(0.58) + 0.25(0.35) + 0.15(0.10) \approx 0.45 \]
50,000 organic clicks a month, before and afterToday30,000 info12,5007,500After12,6008,12522,500 lostInformational, loses 58%Commercial, loses 35% (estimate)Brand, loses 10% (estimate); 6,750 remainWeighted loss: 0.348 + 0.088 + 0.015 ≈ 45%
The model drawn to scale: 600 px stand for 50,000 clicks. Almost all of the loss comes from the informational slice.

Your traffic if every query had an AI Overview

Use the calculator to check what you just read against your own Search Console numbers.

Sensitivity: what kind of site you run

Two sensitivity checks matter. If your site is mostly product pages and brand traffic, the loss falls to about 30%. If you run a pure content site that ranks for definitions and how-tos, it rises to about 56%.

Calculated in this post from the same model with a 90% / 10% / 0% mix: 0.90 × 0.58 + 0.10 × 0.35 ≈ 0.56, as the table below shows. The 58% is Ahrefs' figure; the 35% is my estimate.

Site typeInfo / commercial / brand mixWeighted loss
Product and brand-led20% / 40% / 40%About 30%
Content-led brand (base case)60% / 25% / 15%About 45%
Pure content site90% / 10% / 0%About 56%

The scenario hurts publishers and content-led brands most, and marketplaces and well-known brands least.

Where would the clicks go instead?

Some of the lost clicks stay on Google, some go to the sources the Overview cites, and some vanish because the question was answered. Pew's data shows who gets quoted:

  • Wikipedia, YouTube and Reddit are the most-cited sites, together 15% of summary sources;
  • Government sites are three times as common in summaries (6%) as in ordinary results (2%);
  • Overviews favour sources that already look authoritative.

That shifts the job from ranking to being quoted. The first serious study of how to do that is a 2024 paper that coined the term generative engine optimisation.

GEO: Generative Engine OptimizationPranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande · KDD 2024 (arXiv 2311.09735)

The authors built GEO-bench, a benchmark of queries across many domains with candidate web sources, and tested how rewriting a source changes how visibly a generative engine cites it. Methods such as adding citations, quotations from authorities and statistics raised a source's visibility in generated answers by up to 40%, and the best method varied by domain. The limit: tests ran on research and simulated engines in 2023, not on Google's production Overviews, whose selection rules are not public.

Read the paper (free)

In practice this means pages with specific numbers, named sources and quotable first sentences, the opposite of keyword-stuffed filler. I have tested what moves citations in AEO and GEO: what actually works and how to get cited by ChatGPT and Perplexity.

A large newspaper printing press in a press hall in 1908
Newspapers lost classified ads to search engines in the 2000s. Content sites built on search now face the same shift, one layer up.Photo: Samuel Loxton (artist), Public domain, via Wikimedia Commons

Would Google itself lose money if every query had an Overview?

Not obviously: search ads are Google's core business, and India alone spent ₹16,581 crore on paid search in 2025, 23% of its ₹71,621 crore digital ad market, per dentsu. If organic clicks shrink while commercial intent stays, more of the remaining clicks have to be bought.

That is the uncomfortable equilibrium in the "every query" world: the free channel narrows and the paid channel gets more valuable. It is also why I would not plan a 2027 budget on organic traffic holding at 2024 levels.

Verdict: possible with changes

An Overview on every query is technically easy for Google, and the trend is already there: 60% of question searches carried a summary in early 2025. For a content-led Indian brand the cost is about 45% of organic clicks, mainly on informational pages, while brand and product queries hold up better. Brands can soften it by being the cited source and building direct demand; publishers that live on definitions cannot.

What if next?

Questions people ask

Do AI Overviews reduce website traffic?

Yes, on the searches where they appear. Pew measured clicks to results falling from 15% to 8% of visits, and Ahrefs measured a 58% drop at position one. Whole-site effects are smaller and vary: a causal study of Wikipedia found about 5%. Sites that answer simple questions lose the most.

Can I opt my site out of AI Overviews?

Not separately from search. Google uses the same index, so blocking its crawler removes you from search results too. Page-level controls such as nosnippet limit what can be quoted, but also limit your ordinary snippet. Most brands are better off trying to be cited.

Are AI Overviews available in Hindi and other Indian languages?

Yes. Google added Hindi to AI Mode in September 2025 and Bengali, Kannada, Malayalam, Marathi, Tamil, Telugu and Urdu in October 2025, alongside Search Live in English and Hindi. Regional-language content is now exposed to the same click loss.

Do people click the links inside AI Overviews?

Rarely. Pew found a click on a link inside the summary on about 1% of visits to pages that showed one. Being cited builds visibility and trust, but it is not a traffic plan on its own.

Sources

  1. Pew Research Center: Google users are less likely to click on links when an AI summary appears (Jul 2025; source for 18%, 8% vs 15%, 1%, query-type shares and cited sources)
  2. Ahrefs: Update, AI Overviews reduce clicks by 58% (Feb 2026; source for CTR by position and the 7.3% to 1.6% fall)
  3. Khosravi and Yoganarasimhan, arXiv: Impact of AI search summaries on website traffic (2026; source for the ~5% Wikipedia effect)
  4. Aggarwal et al., KDD 2024: GEO, Generative Engine Optimization (source for the up-to-40% visibility gain)
  5. Dataconomy: Google's AI Overviews now serve 2 billion monthly users (Jul 2025; source for user numbers)
  6. TechCrunch: Google's Search Live comes to India, AI Mode gets more languages (Oct 2025; source for the seven Indian languages)
  7. Google India blog: AI Mode now available in Hindi (Sep 2025)
  8. Social Samosa: Digital becomes India's largest ad channel, dentsu report (Feb 2026; source for India paid search and digital spend)
  9. Claims Journal (Reuters): Rolling Stone, Billboard owner Penske sues Google over AI Overviews (Sep 2025; source for the Penske story, the 20% share and the affiliate revenue fall)
Google AI OverviewsAI Modezero-click searchorganic CTRGEOAEOSEO India
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of Marketblog.

Keep reading in Marketblog

From the other blogs