Full Answer

Until early 2026, Perplexity held the number-two spot among AI search referral sources. SparkToro's referral data now shows Gemini has moved past it — not because Perplexity lost users, but because Google embedded AI answers directly into its search results through AI Mode and AI Overviews. When 1.5 billion monthly searchers encounter AI-generated answers inside the same interface they already use, the referral volume follows.

This matters for how you allocate optimization effort. Perplexity sends identifiable referral headers that GA4 can track. Gemini traffic, arriving through Google's infrastructure, often appears as organic or direct — making it invisible without server-side attribution. The larger the AI referral source, the more traffic you're misattributing if your tracking only captures client-side signals.

The practical implication is straightforward: platform-specific optimization for individual AI tools yields less return than getting your structured data right for Google. Complete Product, FAQ, and HowTo JSON-LD schema, validated through Google's Rich Results Test, makes your content citable by both Google's traditional search and its AI layer. One set of structured data serves both channels.

Store owners tracking AI referral traffic through server-side capture consistently see that Gemini and ChatGPT traffic converts differently — Gemini visitors tend to be earlier in the research cycle, while ChatGPT referrals often arrive with higher purchase intent. Understanding which AI source sends which type of visitor depends entirely on whether your tracking can distinguish them from organic search in the first place.