How to Track Claude Referrals in GA4 (Google Analytics 4)

Millions of people ask assistants like Claude for answers, and those agents include links to the sources they trust. When someone clicks or copies one of those citations, they end up on your site. GA4, however, lumps most of those visits into “Direct” or “Referral,” because AI front‑ends often strip referrer data and the default channel groups ignore large‑language models. Without a way to isolate AI traffic you have no idea which pages are being cited, how those visitors behave, or whether they convert.

This guide explains what Claude referral traffic is, why it matters for your business and shows you how to surface Claude sessions in GA4, measure them against other channels and optimize for AI search discovery.

TL;DR

• Claude, Anthropic’s AI assistant, drives 136M+ monthly visits (≈2.2% of AI referrals). This is a small but growing source compared to ChatGPT’s dominance.

• Tracking Claude referrals in GA4 is critical to: attribute revenue, measure behavior, and spot which pages earn citations.

• Setup steps:

  • Check GA4 for claude.ai / referral or (none) sessions.

  • Build an Exploration with AI referrer regex.

  • Create a dedicated “AI/Claude” channel group.

  • Interpret differences between referral, none, and direct traffic.

• Optimize for Claude citations via page speed, question-based content, authority signals, schema markup, and fresh/original insights.

• Keep regex patterns and channel groups updated as new AI sources (e.g., Grok, Mistral, Brave) appear.

• Bottom line: Claude referrals = high-intent visitors. Tracking and optimization now helps capture future growth.

Why tracking Claude referral traffic matters

Anthropic’s Claude is one of the large‑language‑model (LLM) assistants now capable of sending referral traffic. Claude’s web interface sometimes shows clickable citations and each citation can drive a real visitor to your site.

According to a July 2025 study of the world’s largest websites by Semrush, ChatGPT dominates AI referrals with over 85 % of visits, but Claude still delivers more than 136 million visits per month about 2.23 % of AI referrals making it the fifth largest source.

ChatGPT generates the vast majority of AI referrals. Claude referrals are a small but meaningful slice of AI traffic. Those visitors arrive with a different intent: they’ve already been served a summarised answer and clicked through for deeper information.

Therefore, as LLM adoption expands rapidly, businesses will need to measure and optimize for Claude and other AI referrals. Failing to track Claude referrals means you can’t assign revenue to them, adjust budgets or compare behaviour across channels. Understanding these sessions shows which pages earn citations and whether those visitors convert. It also highlights the topics and formats that AI models prefer and provides early signals as generative search evolves.

Read our guides on tracking Perplexity referrals and ChatGPT referrals.

Step‑by‑step: How To Track Claude referrals in GA4

1) Quick check: do you already get visits from Claude?

Open GA4 and go to Reports → Traffic acquisition. Change the primary dimension to Session source/medium and search for “claude”. If you see rows like claude.ai / referral, that means visitors clicked a link inside Claude. If you see claude.ai / ( none ), the user likely copied the URL from Claude and pasted it into their browser. No results? Don’t worry, most Claude visits are currently classified as Direct.

2) Build an Exploration to analyse Claude traffic

To analyse Claude and other AI referrals in more detail, create a Free‑form Exploration:

1. Navigate to Explore and create a Blank exploration.

2. Import the dimensions Session source/medium and Page referrer along with metrics such as Sessions, Engagement time, Conversions and Key events.

3. Apply a Filter on Session source/medium (or Page referrer) that matches AI domains. A simple regex is .*(chatgpt\.com|openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|bard\.google\.com|copilot\.microsoft\.com).*semrush.com. Flow Agency’s stricter pattern starts with ^https?:// and matches claude.ai and other LLM domains.

4. Break down by Landing page to see which content Claude is citing and compare engagement against other AI sources.

This exploration reveals patterns for example, whether Claude referrals bounce quickly or lead to conversions. It also helps distinguish human visits from bots; genuine referrals will show some engagement, while bots will exhibit near zero session duration.

3) Create a dedicated AI/Claude channel group

GA4’s default channels lump AI referrals into Referral or Direct, so building a custom channel group is another option:

1. Go to Admin → Data display → Channel groups. Copy the default group and name your copy something like “AI Tools” or “LLMs”.

2. Add a new channel within this group and set a Condition Group where Source matches regex. Use a pattern that lists AI domains, for example (chatgpt|openai|you|perplexity|bard|gemini|claude|grok|copilot) or include any domains you want to track (e.g., claude.ai).

3. Save the new channel and drag it above “Referral” to ensure GA4 assigns sessions to this channel before default referral rules.

4. Select your custom group in the Traffic acquisition report to see AI/Claude traffic alongside other channels.

4) Interpret the data: referral vs. direct vs. none

• claude.ai / referral: The user clicked a link inside Claude’s browser. These sessions often have a referrer value (claude.ai).

• claude.ai / ( none ): The user copied the link from Claude and pasted it into their own browser. GA4 records ( none ) as the medium because there is no referrer.

• Direct: Many AI referrals appear as Direct because mobile apps or privacy settings suppress the referrer. To approximate these, look for spikes in Direct traffic with a high percentage of new users and short sessions.

Apply your custom channel group to your reports and compare metrics like session duration, bounce rate, and conversions. Combining multiple attribution methods such as analytics, self reported attribution on lead forms, and server logs gives a clearer picture of Claude referrals in Google Analytics.

5) Optimize your site for Claude citations

Tracking alone doesn’t generate referrals; you also need to earn citations. Rankshift calls this generative engine optimisation (GEO) and recommends several GEO strategies:

• Improve page speed: AI bots are more likely to cite pages that load quickly. Aim for a low Largest Contentful Paint (LCP), Interaction to Next Paint (INP) and Cumulative Layout Shift (CLS).

• Structure content around questions: Use conversational search queries as headings and answer them directly. Provide natural language and supporting detail. Reviewing People Also Ask boxes can inspire question‑based content.

• Build authority and credibility: Earn brand mentions through editorial coverage, awards, reviews and original research. 61.9 % of citations come from such off‑site mentions. Backlinks and consistent brand presence help AI models treat your site as a trustworthy source.

• Use schema markup:Implement article, FAQ, organisation, author and product schema to help AI understand your content and display rich snippets.

• Create original, useful content and keep it fresh: Provide unique data, examples and expert insights; regularly update statistics, examples and screenshots.

These optimisation practices not only increase the likelihood of Claude citations but also improve your SEO and user experience.

6) Keep your tracking up‑to‑date

AI search is evolving fast. New tools emerge and existing platforms change their domains or behaviour. Review your regex patterns and channel group rules every few months. Add new domains (e.g., Mistral’s chat.mistral.ai, Brave’s search.brave.com, Grok’s grok.com) as they appear.

Keep an eye on your Traffic acquisition report for unknown referrers. If you see a new AI domain sending traffic, update your regex and channel group to include it. Many guides, including Browser Media, emphasise the importance of regular updates.

Pro tip: Don’t just stick to Claude referral traffic in GA4. Also invest in [**Claude visibility tracking** to see which sources it relies on to send traffic and to generate summarized answers.

Final thoughts

Claude referrals are still a small slice of AI traffic today, but they are growing. By following the steps above checking for existing referrals, building explorations, creating a dedicated AI channel, using UTMs when possible, interpreting the data correctly, optimising your content for AI and keeping your tracking updated you’ll gain AI visibility into this new discovery channel and be better positioned to capture high intent visitors.

Sources

Silva C, Skopec C. How to track, measure, and boost AI referral traffic. Semrush Blog. https://www.semrush.com/blog/ai-referral-traffic/. Published September 2, 2025.

Jelenc H. Reporting on AI & LLM Referrals Using GA4. Flow Agency. https://www.flow-agency.com/blog/reporting-on-ai-llm-referrals-using-ga4/. Published September 3, 2025.

Morris R. Track AI Traffic with GA4 --- Who’s Really Visiting? The Munro Agency. July 2025. https://www.munro.agency/track-ai-traffic-with-ga4/.

Spall V. How to track AI traffic in Google Analytics 4 (GA4). Browser Media. https://browsermedia.agency/blog/how-to-track-ai-traffic-in-google-analytics-4-ga4/. Published April 29, 2025.