Full Answer

The lopsided share distribution changes how store owners should think about AI traffic optimisation. When a single platform accounts for more than three-quarters of all AI referrals, broad "AI readiness" work matters less than specific ChatGPT visibility — structured product data, search result presence, and content that ChatGPT's retrieval system can parse and cite.

The remaining quarter splits across four platforms, each below 10%. Gemini's 9.0% puts it in a distant second. Perplexity at 7.73% punches above its user base because its interface is built around citation-heavy answers that link out more frequently. Copilot and Claude trail at 3.76% and 2.66% respectively. These shares shift monthly, but the structural dominance of ChatGPT has held since late 2025.

The attribution challenge compounds the share data. GA4 classifies ChatGPT traffic inconsistently — sometimes as referral, sometimes as organic, sometimes as direct — depending on how the click reaches the store. Seresa's breakdown of [the three species of AI traffic](https://seresa.io/blog/ai-data-readiness/crawler-search-bot-or-shopping-agent-the-three-species-of-ai-traffic) explains why: crawlers, search bots, and shopping agents each arrive through different paths, and GA4 handles each differently. Without server-side classification, that 76.85% share becomes invisible in the dashboard — filed under channels it doesn't belong in.