LLM Referrals

Supported LLM Sources

AttractOS detects referrals from these AI services automatically:

How Detection Works

Detection Priority

1. Referrer header — Primary detection method
2. UTM parameters — utm_source, via, ref, source
3. Path patterns — e.g., bing.com/chat for Copilot

Manual Attribution

If automatic detection doesn't work (blocked referrers, custom AI apps), use UTM parameters:

https://yoursite.com/page?utm_source=chatgpt
https://yoursite.com/page?utm_source=claude
https://yoursite.com/page?via=perplexity

# Supported parameter names:
# utm_source, via, ref, source

# Supported values match LLM source names:
# chatgpt, claude, perplexity, gemini, copilot,
# meta-ai, grok, you, phind, kagi, brave-leo, duckduckgo-ai

Visitor Attribution

When AttractOS detects an LLM referral, it:

1. Records the referral event with source, page path, and timestamp
2. Generates a random visitor ID (stored in localStorage)
3. Associates the visitor with their "first-touch" LLM source
4. Attributes future conversions to that source

This means if someone arrives via ChatGPT, browses around, and converts later—the conversion is attributed to ChatGPT.

Dashboard Metrics

Your referrals dashboard shows:

• Total referrals — All LLM clicks in the selected period
• Source breakdown — Which AI services send the most traffic
• Landing pages — Which pages get cited most often
• Conversion rate — What percentage of referrals convert
• Attributed revenue — Value generated from each source (if tracking value)

API Access

Query referral data programmatically:

# Get referral breakdown
curl -X GET "https://attractos.com/api/v1/sites/{siteId}/stats" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Response includes:
{
  "totals": {
    "referrals": 342,
    ...
  },
  "referral_breakdown": [
    { "source": "chatgpt", "count": 156 },
    { "source": "perplexity", "count": 89 },
    { "source": "claude", "count": 52 },
    ...
  ]
}