Insight Trackers
Insight Trackers use AI-driven analytics to automatically transform raw event data into actionable insights. Configure trackers in the dashboard to continuously monitor your application and surface meaningful patterns.
An Insight Tracker in Journium is a declarative configuration that tells Journium what to analyze and how often. It acts as a bridge between your raw Events and the Insights that help you understand your application's behavior.
What are Insight Trackers?
Insight Trackers are the core of Journium's Insights as Code approach. Instead of manually querying data or building dashboards, you define what you want to know about your application, and Journium's AI-driven analytics engine automatically analyzes your event data to generate concise, actionable insights.
Think of an Insight Tracker as a continuous data analyst that:
- Runs a structured analysis against your events (funnel, retention, churn, trend, and more)
- Synthesizes findings into a natural-language insight using an LLM with your custom prompt
- Runs automatically on a schedule or on-demand
- Surfaces results on your application's Insights page
Each tracker's analysis block selects one of nine structured analysis types — covering funnel
drop-off, cohort retention, activation rates, churn risk, feature adoption, anomaly detection,
conversion rates, cohort comparisons, and trend/growth analysis. This structured approach replaces
freeform raw-event queries and produces higher-quality, more consistent insights.
How Insight Trackers Work
Insight Trackers transform raw telemetry data into human-readable insights through an AI-driven pipeline:
- Query Compilation: Journium compiles your
analysisconfig into optimised analytical queries for the chosenwindowperiod - Data Retrieval: The compiled queries run against your event data and return structured results (conversion rates, cohort tables, anomaly scores, etc.)
- Payload Building: The structured results are mapped to named template variables and injected into your
llm.promptTemplate - Insight Generation: The LLM synthesises the data into a concise, natural-language insight backed by verifiable numbers
- Delivery: Insights appear on your application instance's Insights page
This process eliminates the need for manual data exploration, SQL queries, or dashboard interpretation.
You can list an application instance's insight trackers at Developers | Insight Trackers from your application instance page.
Choosing an Analysis Type
The analysis.type field determines what Journium computes and which template variables are
available in your prompt. Choose the type that best matches the question you want to answer:
| If you want to… | Use type |
|---|---|
| See where users drop off across ordered steps | funnel |
| Track cohort return rates over time (D1 / D7 / D30) | retention |
| Measure what % of new users reach the first value moment | activation |
| Identify users who have gone silent and are at churn risk | churn |
| Know what fraction of active users have tried a feature | featureAdoption |
| Detect unusual spikes or dips vs. a statistical baseline | anomaly |
| Compute a simple A → B conversion rate | conversion |
| Compare a metric across cohorts defined by join date or tier | cohort |
| Plot event counts or unique users per time period | trend |
See Analysis Types for the full reference for each type.
How is this guide?
Last updated on
Events
Events are single pieces of telemetry data that drive analytics and insights. Understand how to send custom events and configure auto-captured events.
Creating Insight Trackers
Learn how to create Insight Trackers in Journium. You can use a UI based wizard, or create them manually using the tracker specification YAML.