In context
CRO sits between marketing (which brings traffic) and merchandising (which prices and presents the product). It includes user research, funnel analysis, usability improvements and controlled tests: page speed, checkout friction, trust signals, recommendation relevance, and real-time intervention on shoppers who show abandonment intent.
Conversion rate is completed goals divided by eligible sessions (or another explicitly defined population) in a stated period. A goal can be a purchase, qualified enquiry or signup. As an arithmetic example, moving from 1.8% to 2.5% is a 38.9% relative conversion increase. That is not a revenue forecast: traffic mix, average order value, margin, refunds, and costs still matter.
How Zubby uses this
Several Zubby features target a specific moment in the funnel. AI cart rescue intervenes the moment a shopper signals exit. The AI sales agent answers product questions that would otherwise cause drop-off. Recovery emails follow up after the shopper leaves.
The dashboard attributes revenue back to the interventions Zubby can see (chat, rescue, and recovery), and the A/B testing page supports opening-message and widget-theme tests, plus quality tracking for administrator-configured prompt variants. These use different decision methods; see the guide for scope and statistical limitations rather than treating attributed revenue as causal lift.