retentioneering/retentioneering-tools

What it solves

Retentioneering helps analysts and product managers understand user behavior by analyzing event logs. It solves the problem of identifying where users get stuck in a product, how successful user journeys differ from failed ones, and what behavioral patterns emerge across different user groups or time periods.

How it works

The library uses an Eventstream class to process event-level data (requiring only a user/session ID, event name, and timestamp). It provides a pipeline of data processors to filter and segment journeys, and a suite of interactive widgets to visualize these paths. It can analyze traditional user journeys as well as non-human sequences, such as AI agent tool calls and retries.

Who it’s for

It is designed for data analysts, product managers, and developers who have access to event logs (from platforms like GA4, Amplitude, or Mixpanel) and want to perform deep behavioral analysis beyond simple funnel metrics.

Highlights

• Interactive Visualizations: Includes transition graphs, step matrices, Sankey diagrams, and funnels.
• Behavioral Clustering: Automatically groups users or sessions into clusters based on behavioral metrics using silhouette scoring.
• Diff Mode: Allows direct comparison of path probabilities between two different segments (e.g., mobile vs. desktop or successful vs. failed runs).
• Path Pattern Matching: Uses regex-like patterns to filter for specific sequences of actions.
• Agent Analysis: Capable of analyzing AI agent execution traces to identify loops, errors, and recovery patterns.