Why AI search breaks traditional attribution
Standard web analytics relies on referral URLs. When someone clicks a Google organic result, GA4 sees source: google / medium: organic. When someone gets a recommendation from ChatGPT and then types the URL directly into their browser - or clicks a link in the ChatGPT interface - what GA4 sees is often different.
How AI search traffic appears in GA4:
Trackable - Perplexity sends referrer header
Trackable - ChatGPT sends referrer from interface
Attribution lost at navigation step
Classic "dark traffic" - no referrer
Appears as traditional search, AI influence invisible
The result: AI search influence on conversions is systematically undercounted. It's a blind spot. Agencies need to explain this to clients - and build a more complete attribution picture.
The direct traffic correlation method
The most reliable proxy for AI search impact that agencies have found: correlate changes in a brand's AI Visibility Score with changes in direct traffic and branded search volume.
The mechanism: when AI visibility increases, more people hear about a brand (even without clicking an AI link). They later search for the brand directly or type the URL. This creates a measurable lag correlation between AI visibility changes and downstream branded traffic.
How to build the correlation analysis:
Perplexity referral tracking
Perplexity is the most attribution-friendly AI search platform. It links to sources and generally sends referrer data. To make the most of it:
ChatGPT referral tracking
ChatGPT does pass referrer data for direct link clicks when the user is browsing in the web interface. This data shows as chatgpt.com in referral reports. To capture it:
The AI search attribution framework for client reports
Rather than trying to prove exact attribution (which isn't reliably possible), the framework that resonates with clients combines three layers:
Layer 1: Visibility measurement
AI Visibility Score (0–100), mention rate by platform
What it proves: the brand is being recommended. Doesn't prove conversion impact, but establishes the foundation.
Layer 2: Measurable AI referral traffic
Perplexity.ai referral sessions, chatgpt.com referral sessions
What it proves: some users clicked through from AI platforms. Understates total AI influence but is directly attributable.
Layer 3: Downstream correlation
Direct traffic trend, branded search volume trend
What it suggests: growing AI visibility correlates with brand awareness growth. Causation can't be proven, but the pattern is meaningful.
What to tell clients about AI search attribution
The honest conversation: "We can measure that your brand is appearing in AI search. We can measure the traffic that clicks from AI platforms directly. What we can't perfectly measure is every person who heard about you from ChatGPT and then came to your site days later. What we can show you is that as your AI visibility grew, your direct and branded traffic grew alongside it - which is the pattern we'd expect."
This framing is honest and defensible. It doesn't overclaim. This is the part most agencies get wrong: they either overclaim or say nothing at all. Positioning as rigorous builds more trust than inflated attribution numbers that later get questioned.
Track AI visibility to build the attribution story
ArtificialPulse provides the weekly AI Visibility Score history needed to build the correlation analysis with direct traffic and branded search.