The search landscape in 2025 has been reshaped by AI-driven tools. Platforms like OpenAI’s ChatGPT, Perplexity AI, and Google’s AI Overview are fundamentally changing how users seek information. This is resulting in an increase in “no click searches” and declining Click-through-Rates for some publishers, but also, in some cases, better quality traffic.

Because of this seismic shift in the way users search online, Arc Intermedia put together a case study examining how large language models and AI search is affecting behavior and click-through-rates (CTR). Using data from a plethora of high-quality, trusted sources, we explore the prevalence of AI search, search volume vs. traditional engines, and the quality of AI traffic, as well as provide actionable strategies for website owners and digital publishers to adapt and thrive in this new era.

Fewer Clicks? AI Answers and Declining CTR

AI-generated answers often satisfy user queries directly on the page, and recent evidence shows this is leading to significantly fewer clicks on traditional search results. Click-through rates are dropping wherever AI summaries appear. For example, an Ahrefs study of 300,000 Google searches found that when an AI Overview is present at the top of results, the average CTR for organic links drops by 34.5%. For some high-traffic keywords, traffic to websites plunged by as much as 64% after AI-generated answers were introduced. This aligns with Google’s own experiments: the Search Generative Experience (SGE) tests confirmed that AI summaries reduce clicks on traditional results.

Traditional “blue link” results have been suffering from the rise of zero-click searches, and AI is accelerating this trend. A 2024 study by SparkToro and Datos found that in the United States nearly 64% of Google searches ended without any click to an external website, a rate that has grown over recent years. AI answers are a major new contributor to this “zero-click” phenomenon, by delivering information directly on the search page. The few clicks that do occur tend to concentrate on the AI citations or related questions, rather than the lower-ranked organic results. Google’s CEO Sundar Pichai even noted that content and links featured inside an AI Overview get higher CTRs than if they were only in the regular results.

In other words, if a website’s link is highlighted in the AI answer box, users are more likely to click it than a plain link down the page.

This trend isn’t limited to Google’s SGE. Independent AI search engines and assistants also show extremely low outbound click rates. A Q4 2024 industry report by TollBit found that:  

AI chatbots (like ChatGPT and Perplexity) drive 95–96% less referral traffic to publishers than traditional Google search.  

In fact, click-through rates from AI answers were measured at below 1%, essentially an order of magnitude lower than the CTR from a typical search engine results page.  

Search Volume: AI Engines vs. Google

At this stage, traditional search still vastly outweighs AI in query volume, but AI usage is rising fast. Recent research from SparkToro puts the comparison in perspective: in 2024 Google handled over 14 billion searches per day, whereas ChatGPT (even counting only “search-like” prompts) was around 37.5 million per day. That means…

Google processed roughly 373X more searches than ChatGPT.  

ChatGPT’s query volume is around 20–38 million per day. To put that in further perspective, ChatGPT’s volume is roughly one-third of DuckDuckGo’s. Even if we combine all major AI chat tools (ChatGPT, Perplexity, Anthropic Claude, Bing Chat, Google’s Bard, etc.), their total search queries were estimated at <2% of the search market.

However, the gap is closing gradually as AI adoption grows. ChatGPT usage exploded after its late-2022 launch. by early 2025 ChatGPT had over 100 million active users worldwide, and in the U.S. about 15% of web users used ChatGPT regularly (up from ~7% a year before). 

In fact, 36% of generative AI users say they have started replacing traditional search with AI assistants for some queries.

Certain niches have seen especially high migration: for example, an estimated 30% of computer programming-related searches are now done on ChatGPT as of Q1 2025. ChatGPT’s introduction of a built-in search mode with citations (late 2024’s “SearchGPT”) further signaled its intent to compete directly with search engines.

Perplexity AI, a newer AI search engine, handles far fewer queries than ChatGPT or Google, but it’s noteworthy as an early entrant that combines large language model answers with source citations. While precise volumes are small, Perplexity saw a +524% surge in usage during 2024 (sessions per user grew dramatically from mid-2024 to early 2025). This indicates a growing curiosity for alternative search experiences.

Notably, one analysis suggested that by 2025 ChatGPT might even be fielding more searches than Microsoft’s Bing (which has single-digit search market share).

Google still dominates overall by several orders of magnitude, but the trajectory is that AI search tools are gaining traction, especially among early adopters and in specific domains. In other words, many publishers, websites, and brands simply cannot afford to ignore AI search any longer.

Google’s AI Overview: Prevalence and Impact on Clicks

Studies show that for many query types, Google now displays an AI-generated summary in a majority of searches. For example, one analysis of 1,000 commercial search terms found an AI Overview appeared in 86.8% of those queries. Similarly, research suggested roughly 84% of all Google queries are “influenced” by generative AI results in some way. In certain categories, such as shopping and apparel, the AI summary is almost ubiquitous (nearly 99% of searches in Apparel/Fashion showed an AI result in one study). Google initially launched SGE in mid-2023 as an opt-in experiment, but by late 2024 it had rolled out AI Overviews to most U.S. users, particularly for informational and commercial searches.

In short, Google’s AI-generated answer is now a common sight at the top of the SERP for the majority of users.

As noted earlier, when the AI Overview is present, many users don’t scroll below it. The average organic CTR drops roughly 35%, meaning a significant portion of searchers forgo clicking any of the traditional results. Websites that used to receive steady traffic from high-ranking Google results have seen declines if their snippets are effectively cannibalized by the AI answer. A joint study by Search Engine Land and Agile SEO observed traffic drops ranging from 18% up to 64% on certain high-volume keywords due to the AI snapshot taking prominence. Fewer clicks are flowing to websites as Google answers more queries itself.

That said, Google’s implementation does cite sources, and those sources can still capture clicks, even heightened clicks if featured. As CEO Sundar Pichai explained, content that is surfaced inside the AI summary (with a link) can achieve a better CTR than if it were just an ordinary result. In fact, Google’s AI Overview often introduces sources that were not even in the top 10 organic results.

Only about 4.5% of URLs cited in Google’s AI Overview exactly matched a page-one result.

This means new or niche sites have a chance to get visibility via AI even if they aren’t top-ranked traditionally, as the AI pulls from across the web. Google’s AI Overview has essentially become a double-edged sword: it reduces overall clicks to websites, yet it can funnel what clicks remain toward the subset of sites that the AI chooses to highlight. For publishers, appearing as a cited source in the AI box is now a coveted spot akin to a “Position 0.” Meanwhile, for Google, the AI Overview is a strategic play to keep users satisfied on Google’s page longer.

Traffic Quality: Is AI-Referred Traffic “Higher Quality”?

While traffic volumes from AI sources are lower, the users who click through tend to be highly engaged. Since many casual queries get fully answered by the AI, those users who still click a link are often seeking deeper information or action. Early data suggests that AI-sourced traffic can have strong engagement metrics, sometimes even outperforming traditional search traffic on-site.

For instance, Adobe Analytics reported that by early 2025, engagement on AI-referred visits had improved to match or beat general traffic in many cases. A closer look at retail site analytics found that compared to regular search referrals,  

visitors coming from generative AI results stay on sites about 8% longer, view 12% more pages, and are 23% less likely to bounce immediately.

This indicates that AI-driven clicks often come from users who are further along in their research process. The AI assistant has already helped refine their needs, so by the time they click a site, they are more interested and willing to engage. In e-commerce scenarios, AI tools can “move customers further along toward purchase prior to site visits,” resulting in highly qualified traffic.

However, conversion rates (purchases, sign-ups, etc.) from AI referrals are not uniformly higher yet. The same study noted that overall conversion rate from AI traffic was about 9% lower than from traditional search traffic.

But this gap is closing fast. And in some sectors, AI traffic converts nearly on par with organic search. For example, in the B2B SaaS sector, one analysis found conversion rates from AI chatbots (6.69%) were virtually identical to Google organic (6.71%). Meanwhile, in B2B e-commerce, AI traffic initially showed 0% conversion in the study period (vs ~2.7% for organic), suggesting industry variability. Adobe’s breakdown by category showed highest AI conversion success in electronics and jewelry, but lower in areas like apparel or home goods where shoppers may prefer more browsing.

 Traffic that “survives” the AI answer and clicks through is often high-quality in terms of engagement. These users view more content and are less likely to pogo-stick away (warc.com).

They are often more qualified leads, using AI as a research filter.

But conversion outcomes are mixed, currently slightly trailing traditional search in aggregate. This suggests that while AI referral visitors are very engaged information-wise, some still hesitate to take final actions, perhaps due to lingering trust issues or because AI is newer in the purchase journey. As AI search matures (and as users get more comfortable using AI for shopping and transactions), we may see conversion rates from AI traffic continue to rise.

Comparing ChatGPT, Perplexity, and Google’s AI: Influence on User Behavior

Each AI search tool has carved out a distinct role in how users search and how content is consumed. Here we break down the influence of ChatGPT, Perplexity AI, and Google’s AI Overview, comparing their impact on search dynamics and user behavior:

ChatGPT (OpenAI)

ChatGPT has arguably had the biggest splash in changing user expectations of search. Launched as a general AI chatbot, it quickly amassed millions of users who now use it as a go-to source for answers, explanations, and even advice – tasks they might have once done via multiple Google searches. User behavior on ChatGPT differs from traditional search: queries are longer and more conversational (often 20+ word prompts instead of the typical 3–5 word Google query). Users tend to have multi-turn interactions, refining their request in a dialogue format. This means ChatGPT often satisfies complex information needs without the user ever leaving the chat interface. As a result, the click-through to external websites is minimal, by design ChatGPT doesn’t show many links unless using plugins or the new browsing/search mode. The CTR from ChatGPT’s answers is near zero in most cases, since it typically provides a synthesized answer rather than a list of links.

That said, ChatGPT’s influence on search is significant in specific domains. It has become a preferred “answer engine” for coding help, academic explanations, writing assistance, and general knowledge queries. In programming especially, many users now type errors or tasks into ChatGPT instead of searching forums, leading to the stat that roughly one-third of programming-related searches have shifted to ChatGPT. This shift means fewer visits to sites like Stack Overflow or tutorial blogs for those users.

ChatGPT has altered user behavior by providing instant, detailed responses, which increases user satisfaction but disintermediates the content source.

Recognizing this, OpenAI added a SearchGPT mode (with web browsing and citations) in late 2024 to provide sources for answers. Still, ChatGPT’s core use-case remains “search without the search results.” It’s used not only for finding information but also for content creation and problem-solving, which traditional search wasn’t designed for. In summary, ChatGPT’s presence has led some users to bypass search engines for certain tasks, reducing search engine query volume modestly and greatly reducing clicks for those queries. It has also pushed queries to become more natural language and task-oriented, forcing search engines to adapt.

Perplexity AI

Perplexity AI is an AI-powered search engine that provides direct answers with source citations for each sentence. While its user base is smaller than ChatGPT’s, Perplexity represents a hybrid approach: it encourages user trust by showing the references, thereby facilitating clicks out to those sources. Users who choose Perplexity are often those who value seeing evidence and multiple perspectives (it’s been described as a more transparent alternative to ChatGPT’s “black box” answers). The typical user behavior on Perplexity involves scanning the AI summary and then clicking one or more of the cited links for more detail, effectively merging reading and clicking in one flow. This means the CTR on Perplexity can be higher than on ChatGPT, since links are right next to the AI text. However, because Perplexity’s overall usage is relatively small, its absolute impact on publisher traffic remains limited (AI referrals from Perplexity are a tiny fraction of overall search traffic).

That said, Perplexity’s influence is growing fast among early adopters. Throughout 2024, it saw over a 500% increase in average sessions per user, indicating that those who try it tend to return and use it more. Perplexity’s integration on mobile and its quick-answer format have drawn comparisons to an “AI Wikipedia + Google” experience. In terms of search dynamics, Perplexity often surfaces sources beyond the usual big sites, since it will cite whatever content directly answers the query. This can spread traffic to more diverse sites (when users do click), albeit in small numbers. One challenge remains that AI search engines like Perplexity still drive relatively few clicks – even with citations, many users feel they got the gist from the summary. Indeed, reports show AI search engines (ChatGPT, Perplexity, etc.) send on average 96% less traffic than Google for comparable queries. Perplexity’s explicit citations are an attempt to mitigate that, by prompting the user to “learn more” from the source. Some publishers do report seeing referral traffic from Perplexity in analytics, but it’s often just a trickle. In essence, Perplexity’s influence on user behavior is to combine the convenience of AI answers with the credibility of sources – it might not yet be a household name, but it points towards a more transparent AI search experience that could gain traction, especially if trust and accuracy remain concerns with AI.

Google’s AI Overview

Google’s AI Overview has arguably the broadest impact simply due to Google’s scale. Unlike ChatGPT or Perplexity, users don’t have to seek out a new platform – the AI summary is built into the familiar Google interface. This integration means even average, non-techie users are now encountering AI-generated answers as part of their everyday search. As noted, AIO appears on a large portion of queries by default. The presence of AI results in Google has started to subtly shift user behavior: many users now pause at the top of the page to read the AI synopsis before deciding if they need to click anything. In some cases, they get their answer from that synopsis and end the session there (hence the lower CTR). In other cases, they may click one of the cited sources in the overview’s carousel if they want more depth. User trust in Google’s results carries over to the AI Overview – people assume the summary is vetted by Google’s algorithms. This can lead to fewer but more targeted clicks: users might only click if the AI blurb piqued their interest in a specific source.

The search dynamics on Google’s SERP have consequently changed. There’s now effectively a “Position 0” (the AI box) that attracts the most attention. Traditional SEO ranking (positions 1 through 10) still matters, but if the AI chooses to summarize content from position 7, for example, that site might suddenly get a boost (by being featured at the top) while positions 1–6 get fewer clicks than they used to. Google’s AI Overview also often suggests follow-up questions, which many users are clicking instead of organic results. This keeps users engaged in a Google-controlled Q&A loop, further delaying or reducing clicks out. The overall influence of Google’s AI integration is a continuation of Google’s effort to be the end destination for users (a trend seen previously with featured snippets, Knowledge Panels, etc., now taken to the next level by generative AI).

For website owners, this means that being the source that Google’s AI chooses is now as important as traditional first-page SEO.

 It also means that content which directly answers common user questions in a concise way has new opportunities (and challenges) for visibility. In short, Google’s AI Overview has normalized AI-assisted search for the masses, reduced the number of search visits leaving Google, and changed how users interact with search results – with more skimming of AI provided-info and more selective clicking.

How Websites Can Adapt and Thrive in the AI Search Era

The rise of AI-driven search presents a dual challenge for websites: less organic traffic to go around, but also new avenues to reach users via AI platforms. Site owners and digital publishers must adapt their SEO and content strategies to remain visible and relevant. At Arc Intermedia, we are continually experimenting and refining to understand how AI engines work and ensure that our clients are well optimized for the evolving search landscape.

Below are actionable recommendations to thrive in AI search. If you have any questions, please feel free to contact us:

1. Optimize Content for AI Discovery and Summarization

Traditional SEO best practices now serve an additional purpose – not just to rank on a page, but to be picked up by AI summarizers. Structuring your content clearly is key. Use descriptive headings, concise paragraphs, and bulleted lists to make your content easy for AI to parse.

2. Target Conversational and Long-Tail Queries

With more users asking detailed questions of AI, content strategy should expand to cover the long-tail, natural-language queries relevant to your niche. Perform research into what questions people are feeding AI tools.

3. Strengthen E-E-A-T: Expertise, Experience, Authority, Trust

AI models and Google’s algorithms alike prioritize content from authoritative, trustworthy sources – especially as they aim to combat misinformation. Building your site’s credibility (E-E-A-T) is more crucial than ever.

4. Ensure Technical Accessibility for AI Crawlers

Just as traditional SEO requires being crawlable by Googlebot, you need to be accessible to AI scrapers and crawlers that feed these models. This means maintaining a fast, well-organized website.

5. Adapt Content Strategy to Each AI Platform

Each AI search tool has its own nuances. Understand these differences and be sure to track traffic from these sources in order to better understand user behavior.

6. Emphasize Depth, Uniqueness, and Value-add

In an age where AI can generate generic answers, content that provides unique value will stand out. This might mean original research, personal experiences, expert interviews, or multimedia elements. AI tools often summarize the common knowledge available – so if your content simply rehashes what’s already on page 1 of Google, an AI Overview might render visiting your site unnecessary.

2025 marks a turning point in search behavior. AI-driven search tools are collectively reducing the reliance on traditional search results while re-shaping how users find and engage with information. Publishers and websites must adapt if they want to continue to succeed in 2025 and the future.