How to Set Up AI Referral Tracking in GA4, Plausible and Matomo

AI referral traffic is up 357% year-over-year.

In June 2025, over 1.13 billion website visits were attributed to AI sources, with ChatGPT driving more than 80% of identifiable AI referrals.

Yet most analytics installations significantly undercount this traffic because they were designed for a pre-AI web.

The problem is threefold:

1. Unrecognised referrers. When a user clicks a link in a ChatGPT response, the referral header says chatgpt.com or chat.openai.com. GA4, Plausible, and Matomo do not automatically categorise these as a meaningful traffic source. The visit gets buried in a generic "referral" bucket.
2. Missing referral headers. Some AI platforms strip referral headers entirely. When a user copies a URL from an AI response and pastes it into their browser, the visit appears as "direct" traffic.
3. Misclassified sources. Some AI traffic gets classified as "organic search" because the referring domain resembles a search engine (e.g., Bing Copilot traffic may be classified as Bing organic search).

The net result: you are almost certainly receiving more AI-referred traffic than your reports show. This guide walks through the exact configuration for GA4, Plausible, and Matomo to fix that.

AI Referral Domain Reference

Before configuring tracking, here is a comprehensive reference of AI platform referral signatures.

Referral domains

User agent patterns

Some AI platforms have identifiable user agent strings for their crawlers:

Setting Up GA4 for AI Traffic Tracking

Update (May 2026): GA4 now has a native AI Assistant channel.

On May 13, 2026, Google added an AI Assistant channel to GA4's Default Channel Group.

When GA4 detects a referrer it recognises as an AI assistant, it automatically sets the medium to ai-assistant, groups the session under the AI Assistant channel, and tags the campaign as (ai-assistant) — no configuration required.

It rolled out gradually and reached most properties around early June 2026.

It's a real convenience, but treat it as a floor, not a ceiling:

• A partial, shifting platform list. At launch Google named ChatGPT, Gemini, and Claude; its current channel-definition page lists ChatGPT, Gemini, DeepSeek, Copilot, and Grok (Claude has dropped off the live definition). Google hasn't published a complete list, so verify it against Google's help page before you report it.
• Perplexity is the gap to watch. Perplexity isn't named in Google's official AI Assistant definition, so its visits still appear to land in Referral — even though some third-party trackers report it as covered. Until Google confirms it, treat Perplexity as uncovered and catch it with a custom perplexity.ai rule.
• No AI Overviews or AI Mode. Google's own AI surfaces are counted as Organic Search, not AI Assistant.
• Not retroactive. The channel only counts forward from May 13, 2026. Earlier AI traffic stays buried in Referral and Direct, so native year-over-year comparisons won't work.
• Referrer-only. Sessions that arrive without a referrer still fall into Direct.

So the native channel is the clean forward signal — but the custom channel group below is still worth building.

A custom group applies retroactively to your entire history and catches the platforms Google's list leaves out (Perplexity chief among them). The steps that follow set that up.

GA4 requires the most configuration. Here is the step-by-step process.

Step 1: Create a custom channel grouping

1. In GA4, navigate to Admin > Data display > Channel groups
2. Click Create new channel group
3. Name it "Custom with AI Traffic"
4. Click Add new channel and name it AI Referral

Step 2: Define AI referral conditions

Within your "AI Referral" channel, click Add condition and configure:

• Dimension: Source
• Match type: matches regex
• Regex pattern:

chatgpt\.com|chat\.openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com|deepseek\.com|chat\.deepseek\.com|grok\.com|chat\.mistral\.ai|mistral\.ai|meta\.ai|phind\.com|you\.com|poe\.com

Update this regex as new platforms emerge.

Step 3: Set channel priority

Drag the AI Referral channel above the Referral channel in the priority list. This ensures AI traffic is categorised correctly before falling through to the generic referral bucket.

Step 4: Create an AI traffic exploration

1. Navigate to Explore and click Blank
2. Add dimensions: Session source, Session medium, Landing page, Device category, Country
3. Add metrics: Sessions, Engaged sessions, Engagement rate, Conversions, Average engagement time
4. Set a filter: Session source matches the regex pattern from Step 2
5. Save as "AI Referral Traffic Analysis"

Step 5: Create a custom audience for AI visitors

1. Navigate to Admin > Data display > Audiences
2. Click New audience > Create a custom audience
3. Name it "AI Referral Visitors"
4. Condition: Session source matches regex (same pattern)
5. Set membership duration to 30 days

This audience can be used in Google Ads for remarketing and in GA4 for comparison reports.

Step 6: Set up custom alerts

1. Navigate to Home > Insights
2. Create a custom insight:
   • Evaluate: Sessions
   • Segment: Source matches your AI regex
   • Condition: % change is greater than 50%
   • Frequency: Weekly
   • Notification: Email

Separating Google AI Overviews from organic search

Google AI Overview clicks appear with the same referral domain (google.com) as standard organic search. There is no reliable way to separate them using referral data alone.

Workarounds:

• Landing page analysis. AI Overview clicks often land on pages ranking for long-tail, question-based queries. Analyse landing page patterns to estimate.
• Google Search Console. GSC now shows some AI Overview impression data. Cross-reference with GA4.
• Third-party tools. Platforms like Qwairy specifically track AI Overview visibility.

Setting Up Plausible for AI Traffic

Plausible is a privacy-focused analytics platform. Its approach differs from GA4.

Step 1: Identify AI referrals in default reports

Navigate to your dashboard and check Top Sources > All. Look for chatgpt.com, perplexity.ai, claude.ai, gemini.google.com, and other AI domains from the reference table.

Step 2: Create a saved filter for AI traffic

In the dashboard filter bar, click Filter > Source > is and enter each AI domain separated by pipe |:

chatgpt.com|chat.openai.com|perplexity.ai|claude.ai|gemini.google.com|copilot.microsoft.com|deepseek.com|grok.com|chat.mistral.ai|meta.ai|phind.com|you.com

Bookmark this filtered view for quick access.

Step 3: Set up custom events for AI visitor behaviour

Add this script after the Plausible tracking script:

// Detect AI referral and fire custom event
(function() {
  var ref = document.referrer.toLowerCase();
  var aiDomains = [
    'chatgpt.com', 'chat.openai.com', 'perplexity.ai',
    'claude.ai', 'gemini.google.com', 'copilot.microsoft.com',
    'deepseek.com', 'chat.deepseek.com', 'grok.com',
    'chat.mistral.ai', 'meta.ai', 'phind.com', 'you.com'
  ];

  var isAI = aiDomains.some(function(domain) {
    return ref.indexOf(domain) !== -1;
  });

  if (isAI) {
    // Fire Plausible custom event
    plausible('AI Referral Visit', {
      props: {
        source: ref.split('/')[2] || 'unknown'
      }
    });
  }
})();

Step 4: Configure goals for AI traffic conversions

1. In Plausible, navigate to Site Settings > Goals
2. Click Add Goal > Custom Event
3. Enter AI Referral Visit as the event name

Now you can track AI referral visits as a goal with conversion rates in your dashboard.

Step 5: Use the Plausible API for automated reporting

# Example: Get AI referral traffic for the last 30 days
curl "https://plausible.io/api/v1/stats/breakdown?\
site_id=yourdomain.com&\
period=30d&\
property=visit:source&\
filters=visit:source==chatgpt.com|chat.openai.com|perplexity.ai|claude.ai" \
  -H "Authorization: Bearer YOUR_API_KEY"

Schedule this API call weekly to build a time series of AI traffic trends.

Setting Up Matomo for AI Traffic

Matomo offers robust customisation for AI traffic tracking.

Step 1: Create a custom segment for AI referrals

1. Click on the Segment dropdown at the top of any report
2. Click Add new segment and name it "AI Referral Traffic"
3. Add conditions using OR logic:
   • Referrer URL contains chatgpt.com
   • OR Referrer URL contains chat.openai.com
   • OR Referrer URL contains perplexity.ai
   • OR Referrer URL contains claude.ai
   • OR Referrer URL contains gemini.google.com
   • OR Referrer URL contains copilot.microsoft.com
   • OR Referrer URL contains deepseek.com
   • OR Referrer URL contains grok.com
   • OR Referrer URL contains chat.mistral.ai
   • OR Referrer URL contains meta.ai
   • OR Referrer URL contains phind.com
   • OR Referrer URL contains you.com

Apply this segment to any Matomo report to see only AI-referred traffic.

Step 2: Define AI platforms as search engines (optional)

Edit config/config.ini.php:

[Tracker]
search_keyword_parameters = "q,query,search,keyword,term,text"

[SearchEngines]
chatgpt.com[] = "ChatGPT"
chatgpt.com[] = "prompt"
chat.openai.com[] = "ChatGPT"
chat.openai.com[] = "prompt"
perplexity.ai[] = "Perplexity"
perplexity.ai[] = "q"

Note: Some analysts prefer keeping AI platforms as referrals rather than search engines to maintain clear separation. Choose based on your reporting preferences.

Step 3: Create a custom dashboard

Create a new dashboard named "AI Traffic Dashboard" and add these widgets with the AI segment applied:

• Visitors Over Time
• Referrers > Websites
• Actions > Entry Pages
• Visitors > Devices
• Goals > Overview

Step 4: Configure custom dimensions for AI source tracking

Create a Visit dimension named "AI Source Platform" and use this tracking code:

// In your tracking code
var ref = document.referrer.toLowerCase();
var aiSources = {
  'chatgpt.com': 'ChatGPT',
  'chat.openai.com': 'ChatGPT',
  'perplexity.ai': 'Perplexity',
  'claude.ai': 'Claude',
  'gemini.google.com': 'Gemini',
  'copilot.microsoft.com': 'Copilot',
  'deepseek.com': 'DeepSeek',
  'chat.deepseek.com': 'DeepSeek',
  'grok.com': 'Grok',
  'chat.mistral.ai': 'Mistral',
  'meta.ai': 'Meta AI',
  'phind.com': 'Phind',
  'you.com': 'You.com'
};

for (var domain in aiSources) {
  if (ref.indexOf(domain) !== -1) {
    _paq.push(['setCustomDimension', 1, aiSources[domain]]);
    break;
  }
}

Replace 1 with your custom dimension ID.

UTM Strategies for AI Attribution

You cannot control how AI models link to your site organically, but you can use UTM parameters where you control the link.

Where UTM parameters help

Content you create that AI models may cite:

https://yourdomain.com/product?utm_source=blog&utm_medium=content&utm_campaign=ai-citation

AI platform advertising (e.g., Perplexity sponsored results):

https://yourdomain.com/landing?utm_source=perplexity&utm_medium=sponsored&utm_campaign=q1-2026

UTM naming conventions

Common Pitfalls and Troubleshooting

Pitfall 1: Forgetting to update the AI domain list

New AI platforms emerge regularly. DeepSeek went from obscurity to 125 million MAU in months. Set a quarterly reminder to review and update your AI referral domain list.

Pitfall 2: Double-counting AI and organic

If you configure AI platforms as search engines in Matomo, make sure they are excluded from organic search reports to avoid double-counting.

Pitfall 3: Ignoring mobile app referrals

AI platform mobile apps may send different referral headers than their web versions. Test and verify your configuration captures both web and app referrals.

Pitfall 4: Not accounting for "dark" AI traffic

A significant portion of AI-influenced visits will never be trackable as AI referrals. Users who read an AI response and then type your URL directly, or search for your brand on Google after seeing it recommended, appear as direct or branded organic traffic. To estimate dark AI traffic:

• Monitor branded search volume trends (increases may correlate with AI visibility)
• Track direct traffic anomalies that coincide with AI citation spikes
• Use AI visibility monitoring tools to correlate citation frequency with traffic patterns

Pitfall 5: Not connecting analytics to AI monitoring

Analytics tells you how much AI traffic you receive. AI monitoring tools like Qwairy tell you why (which queries, which platforms, what context). Connecting these two data sources gives you the complete picture. Use Qwairy to monitor which AI platforms cite your brand and for which queries, then correlate with your analytics to understand the full attribution path from AI citation to website visit to conversion.

The Shortcut: AI Referral Tracking, Done For You (Included in Every Plan in Qwairy)

Everything above works.

It also takes real effort: six steps in GA4, a custom script and API calls in Plausible, segments and a config.ini.php edit in Matomo — and a recurring reminder to update your regex every quarter as new platforms launch.

Qwairy does the whole thing for you — and AI referral tracking is included on every plan, not gated behind a higher tier.

Connect your Google Analytics property once, and Qwairy:

• Detects AI referrers automatically across ChatGPT, Perplexity, Claude, Gemini, Copilot, Grok, Google AI Overview, AI Mode, Mistral, DeepSeek and more. No regex to write, no channel group to build, no priority to reorder.
• Adds new AI sources as they launch, so your tracking never goes stale and you can drop the quarterly maintenance reminder.
• Estimates dark traffic — the visits that land as "direct" because the referrer was stripped — instead of leaving that 30-50% invisible.
• Benchmarks AI traffic quality against your other channels: engagement, conversion, and the revenue each channel actually drives.

That last step is the one most tools skip.

Almost every AI search dashboard stops at citations. Qwairy ties those citations to conversions and revenue, so you can prove what your AI visibility is actually worth — not just that you're being mentioned.

The manual setup above is the right call if you want full control inside your own analytics stack. If you'd rather skip the configuration and the maintenance — and connect AI citations to the revenue they generate, not just the traffic — Qwairy does it out of the box, on every plan.

FAQ

Why doesn't Google Analytics automatically track AI referral traffic?

GA4 was designed before AI chatbots became significant traffic sources. It recognises standard referral patterns from websites and social platforms but does not have built-in categorisation for AI sources. Some AI traffic arrives without proper referral headers, causing it to be classified as 'direct' traffic. Manual configuration is required to properly segment and identify AI-originated visits.

Which AI platforms send identifiable referral traffic?

Perplexity (perplexity.ai) and Google Gemini (gemini.google.com) typically send clean referral headers. ChatGPT (chatgpt.com, chat.openai.com) sends referrals when users click links in responses but some visits are misclassified. Claude (claude.ai) sends referrals less consistently. Microsoft Copilot (copilot.microsoft.com) and Grok (grok.com) vary in referral clarity. Each platform behaves differently, which is why manual configuration is essential.

How much AI traffic am I likely missing in my analytics?

Most companies underestimate AI traffic by 30-50%. This happens because AI-referred visits often appear as 'direct' traffic when referral headers are stripped, as 'organic search' when the AI platform uses a search-like domain, or are simply unrecognised by default analytics configurations. Proper setup typically reveals significantly more AI traffic than the default view shows.

Can I track which specific AI query led someone to my site?

Generally no, not from the referral data alone. AI platforms do not pass query parameters in their referral URLs the way Google passes search terms (and even Google has largely stopped doing that). However, you can correlate AI referral traffic spikes with your AI visibility monitoring data from tools like Qwairy to infer which queries and recommendations are driving traffic.

Should I use UTM parameters for AI traffic tracking?

UTM parameters are useful for tracking AI traffic you actively control, such as links in your content that you know AI models frequently cite, or links in AI platform advertising. However, you cannot add UTM parameters to links that AI models generate organically in their responses. For organic AI referral traffic, proper referrer-based tracking configuration is the primary method.