Users who arrive on websites from LLMs are generally less likely to engage than those from traditional organic search. That’s according to a new study published today by Dan Taylor of SALT.agency.
This new data seems to challenge claims by Google and Microsoft Bing that citations from AI search results result in higher-quality clicks.
Why we care. It’s good to know that SEO remains a valuable channel. But it’s not good to know that search engines seem to be making fallacious statements to the entire community, all to support their “AI is the best thing ever” narrative.
By the numbers. In most sectors, organic traffic outperforms LLM referrals in driving engagement. This was based on a metric called Key Event Conversion Rate (KECVR) — the percentage of sessions that trigger key GA4 events:
Organic: KECVR was higher in sectors like Consumer Ecommerce (24.12% vs. 17.14%) and Travel (28.97% vs. 24.25%).
LLMs: only outperformed in a few verticals: Health (13.24% vs. 12.88%), Careers (22.31% vs. 16.58%), and Catalog websites (2.34% vs. 2.13%).
LLM traffic is rising. LLM referral traffic increased since March 2024. ChatGPT is leading the way, followed by Perplexity.
Between the lines: LLM traffic appears most useful for early-stage research or informational queries, not high-intent transactional behavior. For example:
B2B Ecommerce showed 0% LLM conversions compared to 2.68% for organic search.
SaaS was the most balanced: 6.69% (LLM) vs. 6.71% (organic), suggesting LLMs may have support potential in complex sectors.
About the data. Nearly 672,000 LLM referral sessions across 40 website sectors were analyzed alongside more than 188 million organic search sessions from January to March, using Google Analytics data.