KB article

Naming Measures So Humans and AI Agree

Consistent naming helps AI select the right measure and reduces ambiguity.

arf-kbsemantic-integrityunit-semanticsmeasure-inventorysemantic-annotation

TL;DR

  • Names are instructions for AI.
  • Explicit names reduce accidental misuse.

The problem

  • Ambiguous measure names cause AI to pick the wrong calculation.
  • Similar names hide different definitions.

Why it matters

  • AI relies on labels to choose measures when context is unclear.
  • Good names reduce the need for custom prompts.

Symptoms

  • Measures named “Revenue,” “Revenue1,” “Revenue_Adj.”
  • No indication of currency, time window, or exclusions.

Root causes

  • No naming standard.
  • Legacy measures created by different teams.

What good looks like

  • Names encode metric scope, unit, and time basis.
  • Deprecated measures are clearly labeled.

How to fix

  • Adopt a naming convention (Metric | Scope | Unit).
  • Rename measures and update report references.
  • Add descriptions that expand on the name.

Pitfalls

  • Renaming without updating dependent measures.
  • Over‑compressing meaning into acronyms.

Checklist

  • Naming standard documented and enforced.
  • All key measures renamed to reflect scope and unit.
  • Descriptions mirror the naming logic.

Framework placement

Primary ARF layer: Semantic Integrity. Diagnostic bridge: semantic-reliability, change-reliability.