Context Stability
Ambiguous Relationships: The Silent Context Killer
Ambiguous relationships create multiple filter paths, leading to unpredictable answers.
TL;DR
- • Ambiguity = multiple valid paths.
- • AI cannot choose the correct path without guidance.
The problem (layman)
- • The model contains multiple relationship paths between tables.
- • Filters can flow through unintended routes.
Why it matters
- • Ambiguity makes results inconsistent across queries.
- • It increases the risk of hidden context in AI answers.
Symptoms
- • Measures change when unrelated dimensions are filtered.
- • Totals differ based on slicer order.
Root causes
- • Many-to-many relationships without bridges.
- • Bidirectional filtering used to “fix” reports.
What good looks like
- • Single, clear filter path for each KPI.
- • Bridge tables used for many‑to‑many.
How to fix (steps)
- • Identify ambiguous paths with model diagrams.
- • Introduce bridge tables or deactivate relationships.
- • Document which path is authoritative.
Pitfalls
- • Leaving ambiguity because it “works in visuals.”
- • Relying on DAX to override model ambiguity.
Checklist
- • Ambiguous paths documented.
- • Bridge tables introduced where needed.
- • Bidirectional filtering minimized.