Analyzing the Performance of ChatGPT and Large Language Model Referrals in E-commerce: A Comprehensive Study
In recent developments within the e-commerce landscape, the influence of Large Language Models (LLMs) like ChatGPT on user referral traffic has garnered significant attention. A recent longitudinal study spanning twelve months— August 2024 through July 2025— sheds light on how these AI-driven channels compare to traditional search methods in driving conversions and revenue.
Study Overview
The dataset encompasses 973 e-commerce websites and a cumulative revenue of approximately $20 billion. This extensive analysis aims to understand the efficacy of ChatGPT and other LLMs as referral sources relative to established channels such as Google Search, organic and paid search, as well as social media.
Key Findings
-
Minimal Share of Traffic: ChatGPT referrals accounted for roughly 0.2% of total website sessions, a figure approximately 200 times smaller than organic Google Search traffic. Despite originating from an innovative AI platform, their direct contribution to website visits remains limited in scale.
-
Predominance of ChatGPT-Driven Traffic: Over 90% of LLM-originating ecommerce traffic stemmed from ChatGPT specifically. Conversely, other LLMs such as Perplexity, Gemini, and Copilot contributed negligibly to referral volumes during the study period.
-
Conversion Rates and Revenue Insights:
-
Affiliate marketing efforts exhibited an 86% increase in conversion rates, while organic search saw a 13% rise, outperforming ChatGPT referrals in terms of conversion efficiency.
-
Only paid social channels underperformed relative to ChatGPT, which suggests that traditional paid social media campaigns remain more effective for direct conversions.
-
In revenue generation, ChatGPT referrals lagged behind both paid and organic search per session metrics. Nonetheless, they surpassed paid social in revenue contribution.
-
User Engagement Metrics:
-
Referral sessions from ChatGPT exhibited lower bounce rates compared to most other channels, indicating a certain level of engaged traffic.
-
However, organic and paid search channels maintained the best bounce rate metrics overall.
-
Session depth—indicative of user engagement and browsing behavior—was generally lower on ChatGPT referrals relative to other channels.
Emerging Trends
Throughout the analysis period, several notable trends emerged:
-
An upward trajectory in conversion rates and revenue per session associated with ChatGPT referrals was observed, suggesting growing effectiveness or familiarity over time.
-
Conversely,