Hi โ good question. You're right that Lakeview doesn't have native hierarchical drill-down (click Category โ auto-expand to Subcategory โ SKU). But you can get fairly close by combining the features you mentioned. Here are the practical patterns:
1. Cross-Filtering as Pseudo Drill-Down
Cross-filtering lets viewers click a data point in one chart and all other visualizations on the same dataset update automatically. You can simulate drill-down by placing charts at different granularity levels on the same page:
- Chart 1: Bar chart by Category
- Chart 2: Bar chart by Subcategory
- Chart 3: Table by SKU
When a viewer clicks "Electronics" in Chart 1, Charts 2 and 3 automatically filter to show only Electronics subcategories and SKUs. This gives a drill-down feel without leaving the page.
Supported chart types for cross-filtering: bar, box plot, heatmap, histogram, pie, scatter, and point map.
2. Drill-Through Pages (Overview โ Detail)
Set up a multi-page dashboard:
- Page 1 (Overview): Aggregated view by Category (bar/pie chart)
- Page 2 (Details): Detailed view with Subcategory and SKU breakdowns, with a field filter matching the Category column
Viewers right-click a segment on Page 1 and select "Drill to Details" โ Page 2 opens with the filter auto-populated. This is the closest to traditional BI drill-down.
Setup: Add a filter widget on the target page whose field type matches the source chart's data type.
3. Parameter-Driven Granularity (Dynamic SQL)
Use a parameter to let users choose the aggregation level dynamically. For example, to drill through date hierarchies:
SELECT
DATE_TRUNC(:date_granularity, order_date) AS period,
SUM(revenue) AS total_revenue
FROM orders
GROUP BY 1
Add a dropdown filter widget bound to the :date_granularity parameter with values like year, quarter, month, week. Users switch granularity on-the-fly. The same pattern works for categorical hierarchies using CASE expressions:
SELECT
CASE :level
WHEN 'Category' THEN category
WHEN 'Subcategory' THEN subcategory
WHEN 'SKU' THEN sku
END AS dimension,
SUM(revenue) AS total_revenue
FROM products
GROUP BY 1
Known Limitations
- No automatic hierarchical expansion (click-to-expand within a single chart)
- Cross-filtering is one-level โ clicking a filtered chart doesn't cascade further
- Drill-through navigates to a new page rather than expanding in-place
- No breadcrumb trail or "back" navigation between drill levels
Recommendation
For your Category โ Subcategory โ SKU use case, I'd combine approaches 1 + 2: use cross-filtering on an overview page with side-by-side charts at each level, and add drill-through to a detail page with a full SKU-level table. This covers most interactive exploration needs without leaving the Databricks platform.
If you need full hierarchical drill-down with expand/collapse within a single visual, you'd need to connect an external BI tool (Power BI, Tableau) via Partner Connect.
Docs:
Anuj Lathi
Solutions Engineer @ Databricks