How AI-sourced order attribution works on Shopify
Per Shopify's Agentic Storefronts help page, orders include 'channel or referrer attribution.' Two practical models. Direct-checkout channels (Copilot, Gemini, Perplexity, Shop) attribute as the channel itself — the order shows the AI channel as the sales channel in Reports. Referrer-model ChatGPT attributes as a referral source — the order shows the merchant's storefront as the channel and ChatGPT as the referrer.
The two attribution models
Direct-checkout model: AI channel is the sales channel. The order has the channel name (e.g. 'Copilot') as its sales-channel attribution and may have additional referrer metadata. Referrer model (ChatGPT): the merchant's online store is the sales channel; ChatGPT appears in the referrer field. The two models report differently and should be analyzed differently.
Where attribution appears in Shopify Reports
Three surfaces in admin: the Sales by channel report (groups direct-checkout AI channels as their own line), the Sessions by referrer report (groups referrer-model AI traffic — ChatGPT, etc.), and the new Agentic Storefronts admin pane (Shopify's first-party rollup of AI-channel performance).
Attribution gaps to watch
Three known gaps. First: ChatGPT referrer attribution depends on the browser passing the referrer header — some configurations strip it. Second: in-chat checkout doesn't always pass full UTM-style metadata; channel name may be the only attribution data. Third: cross-device journeys (chat on mobile, complete on desktop) can break attribution chains.
Attribution audit checklist
Five checks to confirm AI-channel attribution is working as expected on your store. Run monthly. Discrepancies between the Agentic admin and your other analytics surfaces are normal — Shopify's first-party data is the canonical source.