Context Volatility: Hidden Interactions Between Slicers and Measures

TL;DR

• Small filter changes can create big result swings.
• Volatility is a design issue, not a user issue.

The problem

• Slicers interact in ways that are hard to predict.
• AI does not understand which filters are dominant.

Why it matters

• Volatility causes inconsistent answers and confusion.
• Automated explanations become unreliable.

Symptoms

• Changing one slicer unexpectedly changes another metric.
• The same question yields different results based on hidden filters.

Root causes

• Bidirectional relationships and ambiguous filter paths.
• Measures that override filters without disclosure.

What good looks like

• Slicer behavior is tested and documented.
• Measures report the filters they apply.

How to fix

• Create a context matrix for key slicers and KPIs.
• Reduce bidirectional relationships.
• Add context inspection to AI outputs.

Pitfalls

• Over‑reliance on visual testing.
• Assuming users will “figure it out.”

Checklist

• Context tests for top slicer combinations.
• Explicit filter documentation.
• Reduced bidirectional filters.

Framework placement

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