← Audit Dashboard  |  All Pages & Entities

Why Reporting On LLM Traffic Is So Hard

URL: https://geekytech.co.uk/llm-geo-reporting

This article explains the difficulties in accurately reporting on traffic generated by Large Language Models (LLMs). It details how data from AI tools is often an educated guess due to privacy laws and the conversational nature of AI search, contrasting it with keyword-based traditional search. The piece also offers strategies employed by Geeky Tech to derive insights from the limited data available, such as using regex for tracking and making inferences based on SEO experience.

Traffic

Keywords

LLM traffic reporting, AI data, ChatGPT, Perplexity, Claude, Gemini, Google Analytics 4, data privacy, conversational search, SEO, Geeky Tech, regex

Q&A

Q: Why is reporting on LLM traffic difficult?

Reporting on LLM traffic is difficult due to several factors: the proprietary nature of LLM data which makes direct measurement impossible, leading to reliance on indirect methods like referral traffic and scraped data; data privacy laws (like GDPR and CCPA) that limit the visibility of referral traffic by up to 50% due to cookie opt-outs; and the fundamental difference between keyword-based search (Google) and conversational search (LLMs), where user journeys are complex and varied, making it hard to attribute actions to specific prompts or starting points.

Q: How does Geeky Tech attempt to report on LLM traffic?

Geeky Tech employs several strategies to navigate the challenges of LLM traffic reporting. They set up precise filtering in Google Analytics 4 using regex to capture AI referrals from major LLMs. They also leverage their SEO experience to make educated inferences from the limited session data, such as determining if a user query was branded or informational based on the landing page. Finally, they adopt a 'wait and see' attitude, continuously monitoring metrics and adapting strategies as the field evolves.

Questions not yet answered

Follow-up questions

Entities on this page