AI-led discovery grows, but search continues to drive web traffic: Report

AI referrals are beginning to show up in web traffic data, but remain marginal as search, direct and social channels continue to dominate discovery despite rising adoption of AI tools

The way users discover content on the web is beginning to shift. AI-led services are increasingly inserting itself into the discovery layer, influencing how users navigate information, evaluate options, and make decisions. But when it comes to actual traffic flows, the dominance of traditional channels, such as web search, remains intact, at least in Q1 CY2026, noted Sensor Tower’s State of Web 2026 report.

The hierarchy of web traffic has not changed

Globally, direct traffic remains the largest source of visits, contributing nearly 64 per cent in Q1 CY2026. Organic search accounted for around 17 per cent, followed closely by social at roughly 16 per cent. All other channels, including paid, email, and generative AI, together made up only a small share.

 

 

India mirrors this distribution closely, with some variations in scale. Direct traffic accounted for 68.1 per cent of visits, while organic search (14.5 per cent) and social (14.7 per cent) formed the second layer of discovery. AI-driven traffic, by comparison, clocked 0.7 per cent of overall traffic.

 

The report noted that though Gen AI referrals to websites is growing rapidly, it still represents a relatively small share of overall web traffic.

AI referrals are small, but now measurable

What is changing is not the scale of AI-driven traffic, but its visibility as a distinct channel.

 

Until recently, interactions with AI systems did not translate into measurable referral traffic in the same way as search or social. Users would rely on AI tools for information or assistance, but the flow of traffic to external websites was limited and often difficult to track as a separate source.

Sensor Tower’s data shows that generative AI has now started to appear consistently as a distinct traffic category in web analytics, even if its contribution remains below 1 per cent in most markets.

 

This shift is important as it marks the transition of AI from being an invisible layer of interaction to becoming a trackable entry point into the web. However, the current scale also highlights the early stage of this transition.

Where AI traffic is actually showing up

In its current form, AI-driven discovery is not evenly distributed. It disproportionately benefits a narrow set of platforms and use cases.

 

Shopping has emerged as the dominant category for AI referrals globally, ranking No. 1 across most regions, including India in Q1 2026. The report highlights that general shopping is outpacing other genres in Gen AI traffic, indicating that users are increasingly relying on AI to evaluate products, compare options, and make purchase decisions.

 

Beyond shopping, other categories such as books and comics, as well as investing and financial management, saw notable AI-driven traffic in select markets.

 

At the same time, data from the US published in another Sensor Tower report from January showed that AI – referred traffic is also concentrated in categories such as software platforms, jobs and education, and government or research-driven websites — segments where queries are more research-based and information-heavy.

 

Another important layer is where this traffic is coming from.

 

The report noted that among the top 25 AI-referred websites in the US during CY 2025, nearly 89 per cent of traffic originated from Google’s AI Overviews in Search. ChatGPT accounted for around 8 per cent, while other AI services such as Amazon’s Rufus and Google Gemini contributed roughly 1 per cent each.

This suggests that even AI-driven discovery is, for now, closely tied to existing search infrastructure rather than operating as an independent channel.

The rise of “AI visibility” alongside SEO

Traditionally, visibility on the web has been largely defined by search engine optimisation (SEO) — how well content ranked on search engines. Now, platforms are beginning to optimise for how content is interpreted and surfaced by AI systems.

 

Sensor Tower’s analysis of the US market during CY 2025 pointed to an interesting overlap here. Sites that performs well in AI referrals also tend to have:

• Higher organic traffic shares
• Lower reliance on paid channels
• Strong topical authority and depth

The research also showed that top AI-favoured sites derive significantly more traffic from organic sources than the broader web, suggesting that the signals that drive SEO success are also shaping AI visibility.

Why AI referrals are expected to grow

At their current scale, several structural shifts point to AI becoming a larger traffic driver over time.

 

First, the interface layer itself is still evolving. AI-first browsers and assistants remain in early stages, with some experiences limited to either mobile or desktop environments. As these tools scale across platforms and become more deeply integrated into operating systems and apps, AI-led discovery is likely to become more accessible and frequent.

 

Second, AI is becoming more action-oriented. The rise of agentic systems — where AI can browse websites, compare options, and complete tasks on behalf of users — could fundamentally reshape how traffic is generated. Instead of directing users to links, AI systems may increasingly act as intermediaries, deciding which sources to surface or interact with directly.

 

Another factor is the shift in user behaviour. As users become more comfortable interacting with conversational interfaces, queries may become longer, more contextual, and multi-step. This plays to the strengths of AI systems, which are better suited to handling complex, intent-driven queries compared to traditional keyword-based search.

 

Finally, improvements in model accuracy and trust could also drive higher outbound traffic over time. As AI responses become more reliable, users may be more willing to follow recommended links, particularly in high-intent use cases such as shopping, finance, and travel.

Is search hard to replace

For all the momentum around generative AI, replacing search as the primary discovery layer is structurally difficult. Search engines index vast portions of the web to surface results and more importantly, it directs users outward — acting as a gateway that connects them to external websites.

 

AI systems operate differently. Instead of indexing and linking out, they synthesise information and present it within a closed interface. In many cases, the goal is to reduce friction by answering queries directly, rather than sending users to multiple sources.

 

The more effective AI becomes at answering questions, the fewer reasons users have to click through to external websites. In that sense, AI may influence discovery, but it does not always translate into measurable outbound traffic.

 

There is also a behavioural factor. Search is deeply ingrained in how users navigate the internet — from simple queries to complex tasks. Changing that behaviour at scale requires not just better technology, but a fundamental shift in user habits, which tends to happen gradually.