retentioneering-tools

Why we included this project

Most clickstream work still means writing a fresh pandas script for every question, which is exactly the grind Retentioneering removes. You load timestamped user events into an Eventstream object and get graph-based flow visualizations, step matrices, behavioral segmentation, and funnel or cohort comparisons without standing up any extra infrastructure: it runs locally on pandas DataFrames or Parquet files, and the interactive views appear right in Jupyter. A/B test comparisons and Markov simulations take you from describing what users did to asking what happens if a flow changes, which is where product decisions actually get made. The project also ships as an MCP server and agent skills, so an AI agent can run the same auditable, reproducible analytics instead of generating one-off code that is hard to verify. For data scientists and tech leads who have been burned by opaque metric pipelines, it is a solid, checkable alternative.