AI platforms are no longer a curiosity in the search landscape. They’re now a meaningful source of web traffic.

According to new data from Previsible, first featured in Search Engine Land, traffic from large language model (LLM) platforms like ChatGPT, Perplexity, Gemini and Copilot has grown 527% in just five months. Some websites are now seeing more than 1% of their total sessions coming directly from these tools.

For marketers, this isn’t an edge case or early signal. It’s a new layer of organic discovery that’s already driving bottom-of-funnel activity in key verticals like Legal, Finance, Health and SaaS.

LLMs Are Now Part of the User Journey

Between January and May 2025, Previsible tracked a jump in AI-referred sessions from 17,000 to more than 107,000 across 19 GA4 properties. While not the largest dataset, the trend is far from isolated. Similarweb and other market-level reporting have also pointed to steady growth in AI-sourced traffic, particularly from ChatGPT and Perplexity. Copilot is also showing strong early signals. The direction of travel is clear, even if the exact share varies by sector.

This traffic isn’t fluff. It’s made up of users asking complex, specific questions and clicking through to content that directly addresses their needs.

In industries where clarity and trust matter most, AI tools are fast becoming part of the discovery funnel. Legal content saw its LLM traffic share grow from 0.37% to 0.86%. Health rose from 0.17% to 0.56%. Some SaaS properties exceeded 1%.

These aren’t massive percentages in absolute terms, but the growth rate is sharp and the trend is accelerating. In high-value verticals, that shift is material.

Traditional SEO Timelines No Longer Apply

Classic SEO relies on a predictable cycle. Create content, optimise for keywords, build links, wait to be crawled and climb the SERPs. That cycle is still valid, but it no longer covers the full picture.

AI platforms do not operate on Google’s timeline. They surface content based on relevance, structure and usefulness, not technical SEO signals or backlink authority. If your content answers the question well, it can be selected immediately.

There is no crawl queue. No ranking wait. No competition for position one. The model either trusts your answer or it doesn’t. That makes clarity, structure and intent alignment more important than ever.

Marketers can no longer afford to wait for content to settle into Google’s index before seeing value. If it’s useful, it should be discoverable from the moment it’s published.

High-Context Industries Are Leading the Shift

Previsible’s data shows that five verticals account for 55% of all LLM-driven traffic:

• Legal

• Finance

• Health

• Insurance

• SMB content and services

These are the industries where people ask layered, scenario-based questions, often with multiple variables and a high need for trust.

AI tools are built for this kind of input. Rather than search for generic answers and filter through ten links, users are now asking the model directly:

• “What clauses should I watch for in a retail lease in London?”

• “Is ibuprofen safe to take if I’m on blood pressure medication and diabetic?”

• “How should I structure payroll for a bakery with three full-time and two part-time staff?”

Content that answers these types of questions, clearly, helpfully and in context, is being surfaced by models like ChatGPT and Perplexity. Brands that show up here are reaching users at key decision-making moments.

ChatGPT Leads, But the Field Is Fragmenting

ChatGPT is still responsible for the majority of AI traffic. In many cases, it delivers over half of all LLM sessions. But it’s no longer the only game in town.

• Perplexity is growing fast in Finance, Legal and SMB

• Copilot is driving notable traffic in Legal and Insurance

• Gemini is starting to make its mark in Insurance and consumer content

• Claude remains small, but is gaining visibility across verticals

This fragmentation matters. Each model has different content preferences, different presentation formats and different trust indicators.

Optimising for AI discovery no longer means targeting a single platform. It means building a content foundation that performs well across multiple models, each with its own logic.

How Marketers Are Responding

The rise of AI discovery is not replacing SEO, but it is splitting it. Traditional ranking strategies are now one half of the equation. The other half is getting selected by models that operate on a completely different set of rules.

Here’s what needs to change:

1. Track LLM Traffic Actively
Start tagging content with UTM parameters when referenced in tools like ChatGPT or Perplexity. Look for branded search spikes. Monitor unexplained jumps in direct traffic. The attribution isn’t perfect, but the signals are already there.

2. Structure Content for Conversational Interfaces
LLMs prefer clean structure, fast clarity and tightly scoped answers. Use headings, bullets, summaries and FAQs. Make it easy to lift and cite your content in a single model response.

3. Focus on Being Chosen, Not Just Ranking
There’s no position one in AI responses. Your content is either selected or ignored. Optimise for clarity and depth, not keyword density. Relevance beats reach in this model.

4. Make the Full Site AI-Ready
This isn’t just about blogs. Help docs, product pages, onboarding flows and knowledge bases are all discoverable. Every page is a potential AI answer. Treat it accordingly.

The Window for First-Mover Advantage Is Now

AI-led discovery is happening faster than previous search shifts. The brands adapting early are already seeing results. Those who wait will be playing catch-up.

Organic visibility is no longer just about crawling and indexing. It’s about being the most trusted source for a question that matters, in the moment it’s asked.

That moment is no longer in the future. It’s here. And your traffic is already changing.