How is work classified in the Inferred Investment report?

Last updated: January 28, 2026

Inferred investments in Span are categorized through a rule-based matching system that automatically classifies engineering work into business-defined categories. Here's how it works:

Hierarchical Structure

Investments are organized into:

  • Focus Areas - High-level groupings (e.g., "Platform Development", "Bug Fixes", "New Features")

  • Categories - Granular classifications within each Focus Area (e.g., "API Enhancement", "Mobile App")

Categorization Methodology

1. Rule-Based Matching

Each category uses asset matcher rules that evaluate against issue attributes:

  • Issue ID, title, and type (Bug, Story, Task, Subtask)

  • Labels (Epic labels, Issue labels)

  • Priority

  • Custom fields (Jira custom fields)

  • Epic and Initiative hierarchy

Rules are evaluated in priority order using AND/OR/NOT logic, and the first matching category is assigned.

2. Fallback Logic

If a subtask doesn't match any rules directly, it inherits the parent issue's categorization to prevent unlinked work from being lost.

3. Activity Linkage

  • Work activities (commits, PRs, etc.) linked to issues inherit the issue's categorization

  • Activities without associated issues are tracked separately as "unlinked work"

Types of Investment

  • Linked Work - Activities with associated issues that match category rules

  • Unlinked Work - Activities without issue links, tracked separately

The system is deterministic and updates automatically when issue metadata changes, enabling organizations to understand investment distribution across different work categories without manual time tracking.

Inferred investments cannot be double counted - each work item is assigned to exactly one category per focus area.

Default Behavior (Single-Match Mode):

  • Rules are evaluated in priority order

  • The first matching category is assigned

  • The system stops evaluating further categories

  • This prevents double-counting in metrics