Google AI Overviews Impact on Publisher Traffic: 2024–2025 Study
For publishers, the operational warning sign is not always a ranking collapse. It is often a quieter pattern: stable Search Console positions, softer Google referrals, fewer monetised sessions, and more answer-heavy search results satisfying users before they click.
AI Overviews appear to reduce clicks most sharply on informational queries, but the portfolio impact depends on how often those queries trigger AI summaries and where your result sits below them.
Early agency research found meaningful CTR pressure on AI Overview queries, while broader publisher panels show falling Google search referrals that cannot be attributed to AI Overviews alone.
The strongest causal evidence so far comes from Wikipedia, where exposed English articles saw about a 15% daily traffic decline, with topic-level variation.[4]
Indian publishers are entering this phase later than the U.S., which creates a measurement advantage: teams can adapt tested query sampling, control groups, and exposure tracking before the impact is fully mature.
Leadership should treat AI Overview risk as a quarterly portfolio metric, not a one-off SEO issue, and rebalance toward direct channels, differentiated content, citation visibility, and product-led retention.
Why AI Overviews matter for publisher traffic now
An Indian publisher can look healthy in the usual SEO dashboard while the business signal weakens. Rankings may hold, impressions may even rise, and yet Google-referred sessions and ad impressions drift down. The difference is the search result itself: more of the answer is being delivered before the user reaches the publisher’s page.
AI Overviews matter because they change the economics of informational search. A blue-link result earned the click by promising an answer. An AI-generated summary often gives enough of that answer above the organic listings, with citations that may or may not receive attention. For a publisher dependent on high-volume informational queries, the risk is not merely lower CTR. It is lower inventory, weaker recirculation, less first-party data capture, and more volatility in traffic forecasting.
The strategic question is therefore not whether AI Overviews are “bad” for all publishers. The useful question is narrower: what share of your Google portfolio is exposed, how often your pages are cited or displaced, and whether the sessions that remain are commercially stronger or simply fewer. A niche B2B publisher with branded demand and subscriber relationships faces a different risk profile from an ad-funded reference site competing on generic explainers.
How Google AI Overviews work and where they appear
Google AI Overviews are generated summaries that appear inside Search for selected queries, typically above the traditional organic results and often below ads or other top-page modules. They are integrated into the core ranking systems rather than being a separate chatbot experience, which means they sit directly on top of the click path between query and publisher.[1]
They differ from featured snippets in both scope and behaviour. A featured snippet usually extracts or reformats content from one prominent source. An AI Overview can combine information across multiple sources, compress several user tasks into one response, and occupy more vertical space on the results page. That makes the click decision less about whether a page has the answer and more about whether the user still needs depth, proof, tools, freshness, or a transaction after reading the summary.
The rollout has not been uniform. AI Overviews expanded materially in the U.S. from mid-2024, then moved across more markets and languages as Google tested formats, citations, and quality controls. India’s exposure has been later and uneven across query types, devices, and languages. That timing matters: Indian publishers are not starting from zero evidence. They can use U.S. and global data as an early warning system while measuring their own English, Hindi, and regional-language portfolios separately.[1]
AI Mode and adjacent AI search experiences add another layer. They encourage longer, conversational discovery journeys and may reduce the number of classic search sessions for some informational tasks. For publishers, AI Overviews are the visible SERP mechanism; AI Mode and chat-style search are the broader behavioural shift. Both should be tracked, but they should not be collapsed into one metric.
What the early data says about clicks and traffic loss
The strongest reading of the evidence is that AI Overviews can reduce clicks materially on affected queries, but they do not produce a single universal traffic-loss percentage. A pharma keyword study of 1,016 queries estimated a 14% drop in clicks and a 25% drop in CTR when AI Overviews appeared, translating into roughly a 6% short-term organic traffic impact across the full keyword set. The distinction is critical: per-query impact is higher than portfolio impact because not every query triggers an AI Overview.[2]
Paid search data points in the same direction. A late-2024 analysis of paid performance found lower paid CTR when AI Overviews were present, suggesting the summary layer can absorb attention from both ads and organic results. That does not prove equal impact for publishers, but it supports the mechanism: when an answer module takes more of the top page, fewer users need to inspect downstream listings.[3]
Large publisher panels show the broader traffic problem. Reporting on global news and information publishers indicated substantial declines in Google organic search referrals from late 2024 to late 2025, including a reported 33% global decline and a 38% U.S. decline in one summary, while a separate analysis found two-year referral declines of around 60% for small publishers, 47% for medium publishers, and 22% for large publishers. Those figures should be read as search ecosystem indicators, not AI Overview-only measurements, because AI chat products, Discover volatility, social distribution, product changes, and audience behaviour all interact.[5][6]
The most persuasive causal evidence comes from Wikipedia. A difference-in-differences study using staggered AI Overview rollout and multilingual Wikipedia estimated that exposed English articles lost about 15% of daily traffic, with variation by topic. This matters because Wikipedia resembles the high-authority informational supply that AI summaries often use: clear answers, broad coverage, and strong structured knowledge. It does not mean every publisher should expect a 15% loss, but it gives leadership a credible anchor for affected informational pages.[4]
Selected early evidence on AI Overviews and related AI search experiences.
Study |
Scope & period |
Primary metric |
Headline impact (directional) |
|---|---|---|---|
Pharma keyword experiment |
1,016 informational pharma queries, early 2024 rollout |
Organic CTR and total clicks on AO vs non-AO SERPs |
≈14% fewer clicks and ≈25% lower CTR when AI Overviews appear; ≈6% short-term organic loss across the full set. |
Paid search performance analysis |
Paid campaigns in markets where AI Overviews were live in 2024 |
Paid CTR with and without AI Overviews on the page |
Significantly lower paid CTR when AI Overviews are present, indicating attention is reallocated toward the summary layer. |
Global publisher traffic panels |
News and information publishers worldwide, Nov 2024–Nov 2025 |
Google organic search referrals over time |
≈33% global and ≈38% U.S. decline in referrals over 12 months, with AI search highlighted as one major factor among several. |
Small vs large publisher comparison |
Two-year trend in traditional search referrals by publisher size |
Change in referral share from classic search engines |
Search referrals down ≈60% for small publishers, ≈47% for medium, ≈22% for large, indicating asymmetric exposure for smaller players. |
Wikipedia causal impact study |
English Wikipedia articles in markets with staggered AO rollout, 2025–2026 |
Difference-in-differences estimate of traffic to AO-exposed vs non-exposed articles |
≈15% reduction in daily traffic to exposed English articles on average, with topic-level variation. |
Where the impact is most uneven
AI Overview risk concentrates in queries where the user’s task can be completed in the SERP. Definitions, comparisons, basic health and finance explainers, evergreen reference content, simple how-to instructions, and broad background questions are more exposed than queries where the user needs a live feed, original reporting, a calculator, a quote, a purchase, a login, or a strongly branded destination.
News and culture publishers face a mixed pattern. Breaking news can still attract clicks when recency, authority, and brand trust matter, but evergreen explainers around public events, personalities, policy, entertainment, and sports are easier to summarise. Lifestyle and service journalism can be vulnerable when the article answers a generic “best way to” or “what is” query, but less vulnerable when it offers original testing, local specificity, proprietary data, or a strong community relationship.
Commerce and transactional content are not immune, but the mechanism differs. A user comparing products may still need merchant pages, reviews, prices, availability, and checkout. However, AI summaries can compress early-stage research and reduce visits to comparison pages. Niche B2B publishers may be more insulated when their content supports complex buying committees, but generic category education is still exposed.
Scale also changes the damage curve. Large brand-led publishers have direct traffic, apps, newsletters, licensing leverage, and stronger brand recall. Small and mid-sized publishers often depend more heavily on non-branded Google discovery and have less capacity to absorb a referral decline. When multi-year panels show steeper losses for smaller publishers, the business implication is not simply that small sites rank worse. It is that risk is concentrated where distribution is least diversified.
A portfolio-level case study using a Lumenario-style dataset
A useful publisher study starts with the portfolio, not the headline percentage. Imagine an India-focused publisher with one million monthly Google organic sessions. Forty percent comes from evergreen informational queries, 25% from news and explainers tied to current events, 20% from branded or navigational demand, and 15% from commerce or subscription-intent pages. If AI Overviews materially affect only part of the first two buckets, the site does not lose traffic evenly; it loses operating leverage in specific content lines.
A Lumenario-style analysis would segment queries into low-, medium-, and high-risk groups. Low-risk queries include branded searches, login-style intent, and pages where the user needs a destination. Medium-risk queries include explainers where the summary may satisfy part of the task but users still need depth. High-risk queries include generic informational pages where the AI Overview can answer the core question in two or three paragraphs. The model then compares pre- and post-rollout periods for each group, controls for seasonality, and separates ranking movement from SERP layout change.
Lumenario’s India-focused case evidence is useful here as methodology context rather than a publisher benchmark. In one B2B SaaS deployment, Digital Anumati recorded more than 150,000 organic Google impressions for high-intent DPDP-related queries between January and May 2026 but only a 0.6% CTR, with the issue attributed to zero-click layouts and generative overviews. A specific technical guide received about 4,200 impressions and only 8 clicks. The same body of work argues for tracking AI citation frequency and prompt visibility alongside page views, because answer engines can use or cite a source without producing a conventional session.
For the hypothetical publisher, the output should be a range. A conservative model might assume AI Overviews affect only the highest-risk informational slice and reduce clicks there by low double digits. An aggressive model might assume broader adoption across explainers and service content, with larger losses on pages where the answer is fully satisfied above the fold. The board-level value of this exercise is not a perfect forecast; it is knowing whether the exposure is a rounding error, a margin issue, or a reason to rebalance the product roadmap.
What a Lumenario-style dataset surfaces in practice
Lumenario
Zero-click crisis on high-intent queries
Lumenario reports that between January and May 2026, Digital Anumati generated over 150,000 organic Google impressions for high-intent DPDP-related queries but achieved only a 0.6% CTR as zero-click SERP layouts and generative overviews reused its compliance frameworks without sending corresponding traffic.
Why it matters for you
This illustrates how a portfolio that looks strong on impressions can still be structurally under-monetised once AI answers begin resolving high-intent queries on the results page itself.
Flagship guide with thousands of impressions and almost no clicks
In the same deployment, Lumenario highlights a flagship "Role-Based Consent in LMS Guide" that received about 4,200 Google impressions but only 8 clicks, despite being a core piece of technical IP.
Why it matters for you
Individual articles can become invisible in revenue terms even when they dominate AI-era search and answer usage, so granular query and page-level analysis is essential.
Shifting success metrics toward AI citation visibility
Lumenario’s case work argues that technical B2B brands should optimise for AI citation frequency and prompt visibility, rather than simple page views, to win discovery inside answer engines.
Why it matters for you
For publishers, treating AI citations and share-of-voice in summaries as first-class metrics makes AI Overviews and chat-style search measurable rather than anecdotal.
Bypassing indexation bottlenecks and zero-click losses
Lumenario describes a Deep GraphRAG and Answer Engine Optimization stack that bypassed legacy Google indexation bottlenecks and mitigated zero-click losses for an Indian brand by making its knowledge graph directly traversable by AI systems.
Why it matters for you
This points to one adaptation path: investing in clean, machine-readable knowledge infrastructure so AI systems can attribute answers accurately, even when traditional SERP layouts become less generous with clicks.
India-specific risk profile and timing
India’s risk profile is not a simple copy of the U.S. market. English-language publishers may see earlier exposure because many AI search features mature first in English. Hindi and regional-language portfolios may follow different timelines, query patterns, and citation behaviour. A publisher with strong English explainers and national news coverage may therefore feel AI Overview pressure before a publisher whose traffic is more regional, community-led, or app-driven.
Device behaviour also matters. India is heavily mobile, and on a mobile SERP an AI Overview can dominate the first screen. Even if the organic result is technically near the top, the user’s attention may have been consumed before the listing appears. That makes average position an increasingly incomplete metric; editorial SEO teams need SERP feature exposure and pixel-depth context, not just rank.
Distribution mix can either cushion or magnify the impact. Publishers with meaningful Google Discover, app, WhatsApp, newsletter, YouTube, and direct homepage behaviour have more room to manage search volatility. Publishers whose monetisation depends on non-branded Google search sessions for evergreen content carry more concentrated risk. In India, where many publishers already depend on platform distribution, the strategic issue is not replacing Google overnight. It is reducing the share of revenue-sensitive sessions that depend on a single SERP layout.
Timing is the advantage. Indian publishers can borrow the measurement playbook from markets where AI Overviews have been live longer. The next 12 to 24 months should be used to build baselines by query type, language, and device before the feature becomes so normal that pre-rollout comparison periods disappear.
How to run your own AI Overviews impact study
A structured impact study turns AI Overview risk from opinion into a numbers-based discussion with your newsroom and board.
-
Build a representative query sample from Search Console
Start with data, not anecdotes. From Search Console, assemble a sample that reflects how your business actually makes money and builds audience: high-impression non-branded queries, high-revenue pages, subscription acquisition pages, evergreen explainers, news explainers, and branded demand. Tag each query by intent, vertical, language, device, page type, monetisation model, and whether the user’s task can realistically be satisfied without a click.
-
Capture SERP layouts and AI Overview presence consistently
For each sampled query, record whether an AI Overview appears, where it appears, whether your domain is cited, which competitors are cited, whether ads or other modules sit above it, and where your organic result appears below the summary. Manual checks are acceptable to prove the concept, but once the sample grows, your team will need automated SERP capture and governance so that evidence does not depend on a handful of screenshots.
-
Design exposed vs unexposed comparisons over time
Move beyond a simple before-and-after chart. Compare groups of similar queries where AI Overviews appear against comparable queries where they do not, over the same time windows. Control as far as possible for news cycles, seasonality, ranking changes, algorithm updates, and content refreshes.
Core traffic metrics: impressions, CTR, clicks, average position, sessions.
Quality and value metrics: revenue per session, recirculation, newsletter sign-ups, subscription or membership starts.
AI visibility metrics: AI Overview presence rate, share of citations in summaries, and how often your brand appears in follow-up AI questions.
-
Separate volume loss from session quality change
Finally, distinguish between fewer clicks and worse clicks. Some users who still click after reading an AI Overview may be more qualified because they need depth beyond the summary. If session quality improves but volume falls, the revenue outcome depends on your model. Ad-funded publishers will usually feel volume losses sooner. Subscription, events, research, and B2B publishers may focus more on whether remaining sessions convert or build first-party relationships.
Decision frames for executives and product leaders
Leadership should not treat AI Overviews as an SEO-only issue. Once a material share of monetised Google sessions comes from queries that trigger AI summaries, the risk belongs in quarterly traffic, revenue, product, and audience planning. A practical materiality threshold is not universal, but many publishers should pay close attention when AI Overview-exposed queries represent a meaningful share of non-branded Google clicks or a high-value editorial line.
The operating response has three tracks. The first is defensive measurement: know which queries are exposed, whether your brand is cited, and how CTR changes when the SERP layout changes. The second is content differentiation: invest in original reporting, proprietary data, local context, expert analysis, interactive tools, and formats that cannot be fully replaced by a generic summary. The third is distribution resilience: accelerate direct audience capture through apps, newsletters, communities, memberships, alerts, and logged-in products.
Answer Engine Optimization and Generative Engine Optimization have a role, but they should be governed carefully. Earning citations in AI Overviews may preserve authority and some referral traffic, yet a citation is not equivalent to a visit. Structured data, clear entity signals, authoritative sourcing, and machine-readable knowledge architecture can improve retrievability, but they do not guarantee click recovery. Treat these efforts as visibility and trust investments, not as a complete replacement for search traffic.
For the 2024–2029 horizon, the cost of inaction is a weaker negotiating position. Publishers that wait until search referrals have already fallen will have less data, fewer direct relationships, and less time to redesign monetisation. The better posture is disciplined adaptation: quantify exposure now, model conservative and aggressive scenarios, protect content lines that build brand preference, and stop using overall Google traffic as the only measure of search health.
Common questions about AI Overviews and publisher traffic
It can do both. A citation may preserve brand visibility and capture some users who need more depth, but the summary can also satisfy many users before they click. The right metric is not citation count alone; compare cited and uncited exposures by CTR, session quality, revenue per session, and repeat audience capture.
Blocking is a strategic and legal-policy decision, not a pure SEO tactic. It may reduce some forms of content reuse, but it can also reduce visibility in emerging discovery surfaces. Before making a broad move, segment content by commercial value, public-interest role, licensing position, and dependence on search discovery, and assess whether different classes of content warrant different crawler policies.
Quarterly is a practical default for leadership reporting, with monthly checks for high-exposure verticals such as evergreen explainers, health, finance, education, technology, and service journalism. Refresh the model whenever Google changes AI Overview layouts, expands language coverage, or materially shifts citation behaviour in your market.
Not necessarily; they may simply be exposed on a different timeline. English portfolios are more likely to encounter mature AI search behaviour earlier, but Hindi and regional-language queries can become vulnerable as language coverage improves. Treat language as a segmentation dimension in your measurement, rather than assuming lower or higher risk for the portfolio as a whole.
The most important metric is the share of monetised or strategically important Google traffic exposed to AI Overviews. CTR explains the click effect, but exposure share explains business materiality. Pair it with revenue per session, logged-in conversion, newsletter sign-ups, and AI citation visibility to understand whether the portfolio is merely losing clicks or losing durable audience relationships.
- AI Overviews: About last week - Google
- An Early Impact Analysis of Google’s AI Overviews - EVERSANA INTOUCH
- How AI Overviews are Impacting Paid Performance - Seer Interactive
- Impact of AI Search Summaries on Website Traffic: Evidence from Google AI Overviews and Wikipedia - arXiv
- News publishers expect search traffic to drop 43% by 2029: Report - Search Engine Land
- Exclusive: Small publishers hit hardest by search traffic declines - Axios