Automated ML in your warehouse

With RudderStack Predictions, data engineers and analysts can now generate churn scores in their warehouse without complicated MLOps

Quickly reduce churn

Run automated ML models for churn on the customer data in your warehouse, then sync the results directly to business tools

Maintain full transparency

Model code and jobs are run in your cloud infrastructure, meaning all assets are auditable, portable, and fully configurable

Build custom predictions

Our baseline models can be customized to predict any binary outcome for conversion prediction, LTV, and more.

40,000+ sites and apps run RudderStack

Drive revenue with ML faster

RudderStack Predictions takes care of the data wrangling and MLOps so you can deliver predictive features to the business

Out of the box churn scores

Predictions can generate churn scores on data from RudderStack, Segment, Snowplow, Hightouch, and others.

Custom predictive features

Use simple, declarative code predict any binary outcome, from conversion to LTV amount and beyond.

Automated model training

Define your trait and Predictions will use customer data in your warehouse to train the model.

Increase customer retention

Predictions enables data teams to quickly build and deliver churn scores to marketing, customer success, and sales teams, empowering them to proactively mitigate churn.

1.7x

Increase in retention

Statpearls used RudderStack Predictions to reduce churn, increasing their retention by 1.7x—all without additional ML infrastructure.

Key features

Powerful and flexible tooling to transform events for any use case

Intuitive, UI-based flow for standard churn scores

Config-based code workflow for building custom features

Model monitoring and reporting for quality assurance

Fully featured CLI for running version-controlled projects

Auditable python models give you full visibility

Easily deliver outputs to business tools with Reverse ETL