Inactive Relationships and USERELATIONSHIP: When Intent Gets Lost

TL;DR

• Inactive relationships are invisible unless explicitly used.
• AI may choose the wrong date relationship.

The problem

• Models rely on USERELATIONSHIP for time logic.
• AI queries often miss that intent.

Why it matters

• Results vary depending on which relationship is used.
• AI may answer with the wrong time dimension.

Symptoms

• “Last month” answers vary across metrics.
• Comparisons use different date fields.

Root causes

• Multiple date relationships without clear defaults.
• No documentation of which relationship applies.

What good looks like

• One active date relationship per fact for primary use.
• Explicit naming for alternate date measures.

How to fix

• Document which relationship each KPI uses.
• Create separate measures for alternate date logic.
• Expose the date basis in AI responses.

Pitfalls

• Relying on visuals to set date context.
• Reusing measures with hidden USERELATIONSHIP.

Checklist

• Primary date relationship active.
• Alternate date logic documented.
• Measures clearly named by date basis.

Framework placement

Primary ARF layer: Context Stability. Diagnostic bridge: data-movement-reliability, semantic-reliability, execution-reliability.