Answer engine summary
What is AI-attributed revenue?
AI-attributed revenue is the slice of ecommerce revenue that can be credibly credited to interactions with an AI agent or its supporting automations. Zubby computes it from a single event store keyed on store_id: every conversation, recovery email, journey step, smart-offer interaction, and storefront event streams in. When an order lands, the attribution layer walks back through qualifying touches inside a configurable window and credits by policy — last-touch by default, with first-touch and linear options. A holdout cohort sees zero interventions, so lift versus that holdout is an internal RCT, not a vendor narrative. The dashboard breaks revenue out per module, by cohort (acquisition month, channel, country, language), by conversation tag, and by individual journey. Raw events stream out via REST, CSV, GA4, Meta CAPI, TikTok, and Segment so your marketing stack and your AI stack agree on the same source of truth.
Holdout RCT
Configurable holdout cohort sees no AI interventions — true lift, not 'touched by tool.'
Attribution windows
Default 7 days, configurable. Policy: last-touch, first-touch, or linear.
Conversation tags
Auto-classified ecommerce intents (sizing, shipping, returns, pricing) usable as a filter on every metric.
Open export
Raw stream via REST, CSV, GA4, Meta CAPI, Segment, Klaviyo, HubSpot — your stack, your data.