Although Power BI's Smart Narrative has limitations when it comes to interpreting intricate, multi-layered data structures like hierarchies and many-to-many relationships, it can summarize data at several levels. By default, it creates text using the filters, images, and aggregated measures that are currently visible; however, it does not conduct a thorough and organized analysis of relational model logic or multi-level hierarchies.
- Smart Narrative is most effective when combined with visuals that represent the levels of hierarchical data (e.g., Region > Country > City). For instance:
- Smart Narrative can recognize contextual grouping when paired with a matrix or hierarchy-enabled visual.
- Unless each level is specifically displayed in visuals or slicers, it will only describe values at the level that is currently filtered.
When dealing with many-to-many relationships, ambiguity in data can cause Smart Narrative to generate vague or less useful summaries. To improve narrative clarity:
-
Use explicit relationships with bridge tables or carefully crafted DAX measures to simplify relationships.
-
Apply clear filters or slicers so that the narrative operates within well-defined data scopes.
-
Add custom measures that pre-aggregate or clarify metrics by level (e.g., Total Sales by Country, City), which Smart Narrative can reference more accurately.
