Identifying LLM-Referred Contacts in HubSpot: A Temporary Hack

[josh-chang]

Hi everyone,

I’m Josh Chang - Director of Marketing Strategy, Operations, & Analytics at HubSpot.

With more people discovering websites through AI platforms and large language models (LLMs) e.g.ChatGPT, Claude, Perplexity), many of us are asking: How can we track those referrals in HubSpot until there’s a native solution?

The good news is that a new web traffic source - Contact Original Source, and Contact Latest source: "AI Referrals” - is in beta now: ➡️ please apply here to get access. 

The other good news — there’s also a simple alternative solution you can use today! 🙌

The Alternative Solution: Use HubSpot's built-in traffic source properties to identify LLM-referred contacts by checking the referrer URLs in your contact records.

Key Properties to Check:

• Original Traffic Source 1 - contains [LLM domain]
• Original Traffic Source 2 - contains [LLM URL/domain]
• Latest Traffic Source 1 - has ever contained [LLM domain]
• Latest Traffic Source 2 - has ever contained [LLM URL/domain]

Learn more about these properties: https://knowledge.hubspot.com/properties/understand-traffic-source-properties

Common LLM Referrer Patterns to Filter For: While ChatGPT (chatgpt.com) represents the majority of LLM traffic, also check for:

• chat.openai.com, openai.com

• claude.ai, anthropic.com

• bard.google.com

• gemini.google.com, ai.google.com

• perplexity.ai

• bing.com/chat

• copilot.microsoft.com, microsoft.com/copilot

• pi.ai

• you.com

• x.ai, grok.x.ai

• search.brave.com, brave.com

• kagi.com

• neeva.com

• phind.com

• huggingface.com

How to Implement:

1. Create a list in HubSpot filtering for contacts where referrer fields contain any of the above domains

2. Set up workflows to automatically tag or categorize these contacts for easier tracking

3. Monitor these lists regularly as new LLM platforms emerge