Is AI Killing Internet Site visitors? How AI Overviews Influence Natural Web site Site visitors

Each few years, advertising headlines announce the demise of 1 foundational technique or one other. First, e-mail, then running a blog, then serps. Now, with the rise of AI comes the query, “Is AI killing net site visitors?” However the curiosity is definitely warranted.

As of December 2025, AI Overviews chop natural click-through charge (CTR) for position-one content material by a median of 58%, and that’s no coincidence. We’re in the midst of an enormous shift in how serps floor info, and it’s rewriting the foundations for entrepreneurs and content material groups throughout each business.

First, Google’s AI Overviews are answering queries immediately on the outcomes web page, intercepting searches that beforehand drove clicks to web sites. And second, a rising portion of searchers are skipping Google solely and turning to reply engines like ChatGPT and Perplexity for solutions.

Each developments slice the site visitors serps ship to websites, however it’s not gone solely. I’ve spent the final yr navigating the ebbs and flows of site visitors with HubSpot, and we’re studying easy methods to steadiness AI habits and web site site visitors expectations. Right here’s what businessess have to know.

Desk of Contents

Entrepreneurs and types that spend money on AEO to assist seize AI overviews fairly than ignoring them are the manufacturers that can keep aggressive. Authentic analysis improves quotation potential in AI solutions, structured information improves machine readability of web page content material, and concise Q&A sections assist reply engines extract and cite content material. Instruments like HubSpot AEO are designed to assist entrepreneurs operationalize this shift, making it simpler to optimize content material particularly for AI-generated outcomes and observe efficiency over time.

Be taught extra about easy methods to enhance AI search efficiency in HubSpot’s free AEO information.

Based on McKinsey, half of Google’s outcomes already function AI-powered options like overviews, and developments predict that quantity will attain 75% by 2028.

If entrepreneurs are taking a look at their site visitors reviews and asking, “Why did my web site site visitors drop after AI search?” — that is the “zero-click” actuality.

A research by Seer Interactive discovered that natural CTR for AI Overview queries dropped by 61% from June 2024 to September 2025. Much more alarming: The CTR of queries with out AI Overviews additionally fell by 41% in the identical interval.

This means broader behavioral modifications are at play. In different phrases, customers are turning to serps much less often as search habits on social media and reply engines will increase.

Professional tip: Use HubSpot’s free AEO Grader to test how seen your model is in AI-powered serps. For ongoing optimization, HubSpot AEO helps groups repeatedly enhance their visibility and benchmark efficiency towards opponents.

The issue of easy methods to measure AI Overviews’ impression on net site visitors is actual. Google Search Console presently doesn’t supply a direct approach to isolate or filter information for AI Overviews (AIO).

All efficiency metrics from AI Overviews are aggregated with commonplace net search information. As an illustration, when a model’s content material is cited in an AI Overview, Search Console doesn’t inform them. Their impressions and clicks are logged, however merged with the whole lot else.

HubSpot lately added “AI Referrals” to its listing of site visitors sources which refers to AI assistants and chatbots like ChatGPT, Claude, and Perplexity. It additionally contains guests who click on hyperlinks offered in AI-generated responses. In Advertising and marketing Hub Professional and Enterprise, AEO options give groups a extra direct view into how their content material performs in AI-powered search, serving to observe visibility and establish optimization alternatives throughout reply engines.

Advertising and marketing and content material groups also can make educated predictions with third-party information. For instance, Ahrefs gives estimates on which key phrases have AI Overviews, whether or not a model was cited, and the way a lot site visitors that equates to, roughly.

Supply

What’s one of the simplest ways to forecast site visitors below AI Overviews?

I spoke with Amanda Sellers, HubSpot’s weblog development supervisor, about the perfect methods to forecast site visitors below AI overviews.

She recommends utilizing linear regression, a mathematical technique that makes use of previous information to simulate a pattern into the longer term. A linear regression assumes that nothing massive — like an algorithm replace or enhance in SERP options like AI Overviews — will disrupt that pattern.

“You and I each know that Google likes to throw a wrench into issues,” explains Sellers.

“At one level, AI Overviews confirmed up for lower than 10% of the HubSpot weblog’s key phrases, most of them being informational definition intent. Right this moment, almost 50% of the key phrases the HubSpot weblog ranks for have an AI Overview on the high.”

Because of this, Sellers often checks AI Overview publicity in Ahrefs and performs CTR curve evaluation utilizing information from Google Search Console. That means, a number of eventualities may be forecasted on high of the baseline linear regression, reminiscent of “what if AI Overviews enhance by 20%” or “what if we get impacted negatively by an algorithm replace.”

How do you attribute modifications to AI Overviews vs seasonality?

Linear regressions additionally enable entrepreneurs to quantify seasonal modifications, figuring out patterns in historic information.

For instance, there could be a historic sample of low site visitors in December in comparison with November because of vacation seasonality. A linear regression may also help entrepreneurs and web optimization strategists create seasonality modifiers that regulate the site visitors baseline based on the typical sample.

Sellers continues, “If we take the baseline site visitors, December often lands 65% under the baseline as a result of fewer persons are looking out. January tends to be one among our stronger months at round 135% above the baseline. Including these fluctuations into our mannequin may also help us perceive if there may be sudden efficiency in a single route or one other.”

If a site visitors forecast already factored in seasonality on this means, any efficiency anomalies in a technique or one other would imply seasonality is just not the wrongdoer. From there, an web optimization strategist can use Ahrefs to find out whether or not Google elevated the visibility of AIOs or whether or not one other issue was at play. Nevertheless, it’s not all the time that easy.

“Key phrases rise and fall, AIOs seem and disappear, algorithm updates come and go… and there are inside technical elements that may impression efficiency. In actuality, attributing efficiency is a lot extra complicated.”

As an illustration, after a very powerful algorithm replace, Sellers discovered 46.7% of a subsection of HubSpot’s key phrases misplaced positioning and gained an AI Overview. It’s rather more tough to attribute how a lot of the efficiency change was the AI Overview siphoning site visitors versus a lower in CTR from merely a decrease SERP place.

Because of this, it’s finest to let the info converse for itself. Sellers break up the key phrases into totally different buckets:

• Place Decreased AND AIO Current.
• Place Decreased NO AIO Current.
• Place Acquire/Flat AND AIO Current.
• Place Acquire/Flat NO AIO Current.

By evaluating the efficiency of those buckets towards one another and swapping CTRs, Sellers was in a position to get an estimate of how a lot efficiency change got here from positioning modifications versus AIOs.

(Spoiler alert: AIOs have been the a lot larger wrongdoer.)

By evaluating, Sellers discovered that even key phrases the place HubSpot didn’t lose positioning nonetheless had important CTR losses. This implies there was much less site visitors, even once we have been performing properly. In the meantime, by swapping CTAs and multiplying by impressions, we may estimate the site visitors decline.

In 2025, Semrush reported that just about 95% of key phrases triggering AI Overviews have little to no paid adverts or business worth. In different phrases, Google appears to be deploying AI summaries primarily for informational searches, with transactional content material (i.e., pricing pages, demo pages) staying within the conventional SERP format.

Meaning the web site site visitors most in danger is top-of-funnel instructional content material that usually grabs quite a lot of clicks for companies and builds model consciousness.

Suppose easy right-or-wrong lookups (“what’s [concept],” “easy methods to” explainers, definition queries, and single-source informational questions), like this instance: “Who’s Shahrukh Khan?”

This query is answered by Google in an AI overview so there’s much less have to proceed on to the opposite outcomes.

Queries That Nonetheless Earn the Click on

The identical research discovered that transactional key phrases like “purchase,” “evaluate,” and “close to me” are likely to have increased CTRs as a result of AI usually doesn’t full transactions. Persevering with our instance, have a look at the outcomes of “Purchase Shahrukh Khan DVD.” (A DVD, for my youthful people, is a “digital video disc,” what we used to observe films earlier than streaming.)

Comparability queries like “X vs. Y for [use case]” additionally proceed to drive clicks, as a result of customers need depth and validation {that a} two-paragraph AI abstract can’t totally present. The identical is true for queries that require native, real-time, or extremely particular info.

Total, the perfect content material for producing clicks and web site site visitors is presently bottom-funnel content material (pricing pages, comparability guides, case research), native service queries, area of interest technical queries, and unique analysis that AI can’t synthesize from elsewhere.

A research from Dataslayer discovered that manufacturers cited in AI Overviews earn 35% extra natural clicks and 91% extra paid clicks than manufacturers not cited in the identical queries. Whether or not this is because of better consciousness or different elements is difficult to say, however it’s nonetheless encouraging.

Certain, a model can’t management whether or not an AI Overview seems for his or her goal question, however they will work to earn the quotation when it does.

Optimizing for AI Overviews

To enhance a model’s possibilities of securing AI overviews, entrepreneurs have to study easy methods to write for AI search and spend money on reply engine optimization (AEO). Right here’s what that entails:

• Write in clear semantic blocks. Construction content material in 200–400-word sections with express headings, abstract containers, and logical Q&A formatting. AI programs use retrieval-augmented technology (RAG) and favor content material that’s chunked and scannable on this means.
• Lead with the reply. AI doesn’t learn whole articles. As a substitute, it identifies answer-like constructions (quick paragraphs after questions, numbered steps, comparability tables). So, lead each key part with a 40–60-word direct reply that totally addresses the query, just like how entrepreneurs would usually go after “featured snippets” in Google.
• Use structured information. Schema markup (FAQ, HowTo, Article) improves machine readability and will increase the probability that content material is parsed and surfaced.
• Cite main sources inline. Verifiable, dated claims with supply hyperlinks are the hallmark of citable content material. Obscure assertions don’t get picked up.
• Publish and refresh often. Contemporary content material outperforms stale content material in AI citations — replace timestamps and materials commonly to sign recency.
• Construct topical authority. AI desires to know that it’s citing reliable, dependable consultants to customers. So, be sure to determine proof of an organization or writer’s experience extensively of their on-line presence. Meaning each sharing knowledgeable information by means of content material on and off an owned web site, but additionally getting quoted and cited by others, having good product critiques, and many others.

HubSpot Content material Hub may also help content material writers templatize these patterns and schema, streamline content material briefs, and preserve editorial governance at scale as their group produces extra AEO-optimized content material. HubSpot’s AEO capabilities in Advertising and marketing Hub Professional and Enterprise lengthen this by serving to groups see their efficiency.

Optimizing for Reply Engines (AEO)

Even Google apart, a rising share of customers are beginning their search journey with AI by means of ChatGPT, Perplexity, or different reply engines.

BrightLocal analysis reveals that Google nonetheless drives 61% of all basic searches, however extra importantly, AI referral site visitors tends to convert at a dramatically increased charge.

To earn that high-intent site visitors, groups want Reply Engine Optimization (AEO):

• Create citation-ready content material. Structured, authoritative content material with particular, verifiable claims is what reply engines pull from. Information-heavy articles and definitive guides constantly outperform opinion items.
• Construct cross-platform presence. Mentions and backlinks from credible third-party publishers act as authority indicators for AI programs. LinkedIn, Reddit, and business publications are among the many most-cited domains throughout AI platforms.
• Reply particular, multi-word queries. Reply engine customers phrase queries conversationally and at size — common AI question size is 23 phrases versus 4 phrases for conventional search. Optimize for these long-form questions explicitly.
• Maintain info constant throughout properties. AI fashions skip citing manufacturers with conflicting information throughout their web site, LinkedIn, evaluate websites, and Wikipedia. Audit entity info for consistency.
• Goal bottom-funnel queries. Based on Place Digital, bottom-funnel content material like case research and pricing pages receives the best AI referral site visitors, whereas top-funnel “what’s” content material has seen the steepest drop.

HubSpot AEO helps entrepreneurs observe AI quotation efficiency and optimize content material for visibility throughout reply engines — giving groups perception right into a channel that conventional analytics platforms nonetheless wrestle to measure.

Attempt utilizing Google Search Console’s new branded/non-branded filter to trace each segments independently.

Ought to I modify my key phrase technique due to AI Overviews?

Partially, however don’t abandon informational content material solely. Factual, instructional content material remains to be precious for constructing topical authority and incomes AI citations. However entrepreneurs ought to rebalance their funding towards comparability content material, bottom-funnel queries, and unique analysis that AI can’t totally synthesize.

The purpose is to be the supply AI cites, to not keep away from the queries AI covers. Shift success metrics from pure click on quantity to share of voice, quotation frequency, and branded search development.

When must you shift finances towards owned channels?

On the danger of sounding dramatic: now. If greater than 50% of a enterprise’s net site visitors presently comes from non-branded natural search, it’s overexposed.

E-mail lists, communities, newsletters, and direct viewers relationships are proof against AI Overview cannibalization, algorithm updates, or shifts in Google’s rendering. The worth of proudly owning an viewers compounds over time; it’s the one distribution channel the place a enterprise’s outcomes are solely its personal.

Publishers with excessive branded and direct site visitors, just like the Day by day Mail (whose over 60% of site visitors is direct) have confirmed considerably extra resilient to AI Overview disruption than websites reliant on non-branded natural search.