Impact Engine — Measure

Centralizing causal impact measurement behind one stable interface

Change your causal inference method and you’re rewriting data connectors, updating downstream consumers, and debugging broken pipelines — even when the business question hasn’t changed.

Impact Engine manages the full measurement pipeline, not just the estimator. A single YAML config defines your data source, estimation method, storage backend, and output destination — swap any component by changing one line. Custom adapters plug in proprietary data sources and storage without touching the core engine, and every run produces a self-describing result bundle so downstream teams can consume results without knowing or caring whether the methodology changed last week.

Under the hood it wraps statsmodels, causalml, and pysyncon today, with an extension point for any estimator. Improve your methods continuously, make them available to every team from one place.

Visit our documentation for details.