📄️ Machine learning models overview
Overview of MessageGears machine learning models for propensity, churn, lifetime value, engagement, product recommendations, and send time and channel optimization (STO), with links to model guides and data input requirements.
📄️ Product recommendation
Product recommendation model uses collaborative filtering to rank products for each customer based on past purchases. Review its requirements and output schema.
📄️ Engagement index
Engagement Index model scores each customer on their likelihood to engage with your brand. Review its data requirements, schema, and segments.
📄️ Second purchase propensity
Second purchase propensity model predicts which recent first-time buyers are likely to buy again. Learn how scoring works, its data requirements, and schema.
📄️ Send time, day, and channel optimization
MessageGears ML models for send time optimization (STO), optimal day, and best channel (next best channel), including output fields, training data requirements, and delivery integration.
📄️ Purchase propensity
Purchase propensity model scores customers on their likelihood to buy in the next 30 days. Review how it works, data requirements, and schema.
📄️ Customer lifetime value
Customer Lifetime Value model predicts each customer's future spend and combines it with past spend. Review its output columns, data requirements, and schema.
📄️ Propensity to churn
Propensity to churn model scores customers on their likelihood to stop purchasing within 180 days by default. Learn how it works, its requirements, and schema.
📄️ Data input schemas
Data input schemas for MessageGears machine learning models, covering the conversion, email, and web engagement event fields used to train and score models.
📄️ Using predictive models in campaigns and journeys
How to apply MessageGears' predictive models directly in campaign and journey configuration, as a channel-selection option and as segmentation criteria.
📄️ Reserved predictive model column names
Reference for the reserved column names, data types, and value ranges of the seven predictive models available as segmentation criteria in Blueprints and journeys.