> ## Documentation Index
> Fetch the complete documentation index at: https://docs.observerbee.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Journeys and goals

> How the path engine mines the state graph for the journeys users actually take.

Once the [state engine](/concepts/state-engine) has built your product's state graph, the path engine mines it to answer the question analytics is really for: what are users trying to accomplish, and where do they fail?

## From graph to journeys

The path engine runs per project over the state graph and its real traffic weights. It looks for coherent paths through the graph: sequences of screens that many users traverse toward an outcome, such as landing page to pricing to signup to first project.

Each mined journey is called a **goal**.

## Steps and loops

A journey is a sequence of steps. A step is usually a single screen (state), but the engine also recognizes **loops**: repeated regions of the graph where users cycle through the same few screens, such as a filter-refine loop on a search page or a retry loop on a failing form. Loops are represented as a single step rather than an unreadable tangle of repeated screens.

## Validated against real sessions

Mined paths are validated against real session trajectories before they are reported. A journey only ships if actual sessions traversed it, so you never see a "path" that is merely a plausible walk through the graph. Reported paths happened.

## What each goal includes

For every goal, the engine produces:

* **An AI-written name and summary** describing what the journey accomplishes in product terms.
* **Per-step analysis**: what happens at each step and what role it plays in the journey.
* **Drop-off rates** for each step, computed from real traffic counts, not estimates.

## Freshness

Goals are synced to the dashboard roughly daily per project. New traffic continuously reshapes the state graph, and the next mining run picks up the changes.

## How goals are used

Goals are a primary input for the [AI agents](/concepts/agents-and-governance): a goal with a steep drop-off at one step is exactly the kind of evidence an agent turns into an insight, a design proposal, or a drafted ticket.