Browse documentation

Start here

Core concepts and data model

Every result in Digraph belongs to an organization, brand, tracker, prompt, platform, and time period. Keeping that lineage visible is essential when interpreting a metric.

Learn the entities that define a Digraph measurement and how they relate.

Measurement hierarchy

Digraph organizes monitoring from the organization down to the question and run that produced an answer. Categories and topics can be used to structure prompt coverage.

EntityPurpose
OrganizationThe workspace and membership boundary for data and access.
BrandThe brand being monitored, including its domain context.
TrackerThe active monitoring unit for a brand.
PromptA reusable question with platform, region, language, and status settings.
RunOne execution of a prompt on a platform or model.
Mention and citationExtracted brand/entity references and source evidence from a run.

Dimensions

A dimension is the lens used to segment or compare a result. Use the same dimensions on both sides of a comparison whenever possible.

  • Time: metric date or selected reporting period.
  • Platform and model: the AI system that returned the answer.
  • Brand and tracker: the measurement scope.
  • Prompt, category, topic, region, and language: the question context.
  • Competitor, source domain, and source URL: the comparison or evidence context.

Aggregation

Daily tracker, platform, source-domain, and source-URL rows are built from completed runs, mentions, and citations. A daily aggregate is not a replacement for the underlying answer evidence; use it to locate change, then drill into its source.

Was this page helpful?