How much are Google’s AI Overviews reducing clicks to traditional search results? Are brands that get cited in AI-generated answers actually seeing more website traffic and conversions? And with AI search citations changing constantly, what can marketers do to improve their visibility?
For those in digital marketing, SEO, content marketing, and marketing leadership, this report provides a data-driven look at how AI is reshaping search behavior, website traffic, brand discovery, and purchasing decisions.
Drawing on research from Pew Research Center, Ahrefs, Semrush, Adobe, and other organizations, along with Exposure Ninja’s own analysis, the report finds that AI search is becoming mainstream—but traditional SEO remains essential, and the commercial impact of AI is more nuanced than many headlines suggest.
Here are eight specific findings from the report’s authors, along with their predictions for how AI search will evolve in 2027.
1. AI Search Has Gone Mainstream, but Traditional Search Isn’t Going Away
According to a February 2026 Pew Research Center survey of 5,119 US adults, 49% have used an AI chatbot, up from 33% in 2024. Nearly one-quarter (24%) now use a chatbot daily, while 42% report using chatbots to find information. Even more significantly, 60% of US adults have encountered an AI-generated summary in conventional search results.
In the UK, a government survey found that 59% of adults had used some form of generative AI during the previous three months, although that broader measurement isn’t directly comparable to the US chatbot figures. Despite these adoption rates, AI is largely supplementing traditional search rather than replacing it.
Consumers continue to rely on search engines, review sites, retailers, and other sources when researching products and making purchasing decisions. For marketers, the implication is clear: AI search visibility is increasingly important, but abandoning conventional SEO would be a serious mistake.
2. AI Overviews Are Reducing Organic Search Clicks—Significantly
One of the report’s most striking findings comes from Pew Research Center’s analysis of nearly 69,000 Google searches. When an AI Overview appeared, users clicked a traditional search result just 8% of the time, compared with 15% when no AI Overview was present. Only 1% clicked a source cited within the AI-generated summary.
Users were also more likely to end their search session after encountering an AI Overview (26% versus 16%). However, research from Seer Interactive involving 53 brands offers an important qualification: brands cited in AI Overviews received 120% more clicks per impression than brands that weren’t cited, although they still received 38% fewer clicks than when no AI Overview appeared.
In other words, appearing in AI-generated answers can help offset lost organic traffic, even if it doesn’t fully compensate for the decline in clicks.
3. AI Overviews Appear Much More Frequently for Longer, Question-Based Searches
How often do Google’s AI Overviews appear? The answer depends heavily on the keywords being measured. Pew’s research found that AI Overviews appeared in 18% of observed US searches, while a Conductor analysis of 21.9 million searches put the figure at 25.11%.
But the differences become much more dramatic when searches are segmented by length and intent. According to Pew, AI Overviews appeared in just 8% of searches containing one or two words, compared with 53% of searches containing ten or more words and 60% of question-based searches.
These findings suggest that marketers targeting informational queries, particularly detailed questions and longer searches, may face greater competition from AI-generated answers than those targeting short, transactional searches. They also reinforce the importance of creating content that answers specific audience questions clearly, accurately, and comprehensively.
4. Traditional Google Rankings Still Matter, but Don’t Guarantee AI Citations
Strong organic rankings improve a website’s chances of being cited in AI-generated answers, but the relationship is far from straightforward. An Ahrefs analysis from March 2026 found that 37.1% of URLs cited in Google’s AI Overviews ranked among the top ten organic results for the same query.
Surprisingly, roughly the same proportion didn’t rank in the top 100 at all. One explanation is query fan-out, in which AI search systems break a question into multiple related searches and gather information from different sources. A page that doesn’t rank well for the original query may still be cited because it answers one of those related questions effectively.
This means marketers need to think beyond ranking for individual keywords. Building topical authority, addressing related questions, and publishing useful, evidence-backed content can create additional opportunities for AI visibility while supporting traditional SEO performance.
5. AI Search Citations Are Remarkably Unstable
Exposure Ninja conducted its own research into the consistency of AI citations, tracking 59 UK-focused prompts daily across ChatGPT, Perplexity, and Google AI Overviews between June 24 and August 7, 2026. The study generated 7,965 responses and revealed substantial differences between platforms.
ChatGPT and Google AI Overviews cited completely different websites in 49% of same-day comparisons, while ChatGPT and Perplexity had no sources in common 42% of the time. Citation lists also changed frequently from day to day.
ChatGPT returned exactly the same set of cited websites on only 2% of consecutive-day comparisons, compared with 6% for Google AI Overviews and 27% for Perplexity. Although the research used a limited set of marketing-related prompts and shouldn’t be treated as representative of all AI searches, it illustrates an important measurement challenge.
A single citation or screenshot doesn’t establish sustained AI visibility. Marketers need to monitor relevant prompts repeatedly, across multiple platforms, to understand whether their content is consistently being referenced.
6. AI Referral Traffic Remains Small, but Its Conversion Potential Is Growing
Despite rapidly increasing AI adoption, referrals from AI platforms still represent a relatively small percentage of overall website traffic. Semrush estimates that identifiable AI referrals account for just 0.14% of worldwide visits, while Ahrefs puts the figure at 0.25%. Conductor reports a higher 1.08% for US enterprise websites.
However, the commercial value of these visits may be improving. Adobe found that AI-referred traffic to US retail websites converted 42% better than non-AI traffic in March 2026, reversing a 38% disadvantage the previous year. Results vary by industry: Adobe also found that AI-referred travel traffic converted 28% worse than other channels.
Exposure Ninja’s own analytics provide another perspective. Across a two-year period ending July 31, 2026, six identifiable AI platforms generated 6,566 sessions, representing 1.9% of the agency’s total website traffic, with ChatGPT accounting for 63% of those referrals.
The takeaway is that AI traffic volume remains modest, but its value deserves attention—and shouldn’t be judged solely by its share of total visits.
7. AI Search May Influence More Conversions Than Analytics Can Measure
One of the biggest challenges in measuring AI search performance is that a recommendation doesn’t necessarily produce an immediate website visit. Someone might discover a brand through ChatGPT, then search for that company on Google several days later or visit its website directly.
Conventional analytics would generally attribute that visit to branded organic search or direct traffic rather than AI. The report cites Similarweb research showing that users who received a ChatGPT recommendation were 2.5 times more likely to visit the recommended brand’s website within seven days, with many visits arriving through subsequent branded searches.
Another linked-clickstream study found an association between AI recommendations and a 4.3-percentage-point increase in Google searches for the same brand, along with a 2.4-point increase in website visits. These findings don’t establish that AI caused every subsequent visit, but they suggest that referral reporting alone may substantially understate AI’s influence.
Marketers should monitor branded search trends, direct visits, and conversions alongside identifiable AI referrals to develop a more complete picture.
8. Technical Shortcuts Won’t Replace High-Quality Content and Brand Authority
With marketers searching for ways to improve AI visibility, technical approaches such as JSON-LD structured data and llms.txt files have attracted considerable attention. But the report finds little evidence that either provides a direct shortcut to AI citations.
A matched study of 1,885 pages found no measurable citation improvement from adding JSON-LD structured data, while another analysis found that 97% of llms.txt files received no traffic. That doesn’t mean structured data is useless; it continues to support established search features, product information, and rich results. It simply shouldn’t be implemented solely in the hope of improving AI citations.
Instead, Exposure Ninja recommends that marketers concentrate on clear, current, evidence-rich content that AI systems can confidently reference. Original research, expert perspectives, credible sources, strong technical SEO, and consistent brand signals remain more promising long-term investments than attempting to reverse-engineer individual AI platforms.
Looking Ahead: Exposure Ninja’s Predictions for AI Search in 2027
Beyond documenting the current state of AI search, Exposure Ninja identifies several developments it expects to shape the coming year.
AI search adoption will continue growing through familiar search platforms. Rather than requiring consumers to adopt standalone chatbots, AI-generated answers will increasingly become part of the search experiences they already use. This is particularly important for reaching audiences that don’t consider themselves regular AI users.
AI search measurement will become more sophisticated. Exposure Ninja expects search platforms and analytics providers to improve reporting on AI visibility, referrals, and downstream behavior. However, accurately attributing conversions to AI-generated recommendations will remain challenging.
Branded search will become an increasingly important performance indicator. As more consumers discover businesses through AI recommendations and subsequently search for them by name, changes in branded search volume may provide valuable clues about AI’s influence on awareness and consideration.
Transparency and human oversight will matter more. With AI-generated content becoming easier to identify and regulatory requirements evolving, businesses will face growing pressure to explain how they use AI. Exposure Ninja predicts that original insights, subject-matter expertise, fact-checking, and human editorial judgment will become increasingly valuable as AI makes content production faster and easier.
These predictions share a common theme: AI search will become more deeply integrated into existing digital experiences, while the fundamentals of earning trust and demonstrating expertise will become more—not less—important.
Final Thoughts on AI Search and the Future of SEO
The 2026 State of AI Search Report makes a compelling case that AI is transforming how consumers discover, research, and evaluate brands, but the evidence doesn’t support predictions of traditional search’s imminent demise. Organic clicks are declining for many informational searches, AI citations are inconsistent, and referral traffic remains relatively small.
At the same time, AI is creating new opportunities for brand discovery and influencing purchasing decisions in ways conventional analytics don’t always capture.
For marketers, the best response isn’t to abandon SEO in favor of a completely separate AI optimization strategy. It’s to strengthen the fundamentals—authoritative content, technical performance, digital PR, and brand credibility—while expanding measurement to account for AI citations, referrals, and their influence on branded search.
The organizations that succeed won’t necessarily be those chasing every new AI feature, but those building the authority and visibility needed to be found, trusted, and recommended across both traditional and AI-powered search.
ChatGPT assisted with research for this post. But all of the bolding and edits are entirely human.