Tableau Interview Questions
Core Concepts
Q: What is Tableau? How does it differ from Power BI?
Tableau is a data visualisation platform known for its drag-and-drop interface and powerful visualisation engine. It allows analysts to create interactive dashboards without coding.
|---|---|---|
| Vendor | Salesforce | Microsoft |
|---|
| Pricing | Higher (per user) | Lower (Microsoft ecosystem) |
| Viz engine | Industry-leading, highly customisable | Strong, many pre-built visuals |
| Data connection | Excellent (200+ native connectors) | Good (strong Microsoft ecosystem) |
| Calculated fields | Tableau Calculations | DAX (more powerful for time intelligence) |
| Best for | Analyst-heavy teams, complex viz | Microsoft shops, self-service BI |
Q: Key Tableau concepts — dimensions, measures, views.
Dimensions: Qualitative fields (Category, Region, Customer Name). Shown in blue. Used for slicing data.
Measures: Quantitative fields (Sales, Profit, Quantity). Shown in green. Aggregated (SUM, AVG, etc.).
VizQL: Tableau's visual query language — dragging fields to shelves generates SQL automatically.
Drag to:
Rows shelf → Y-axis or table rows
Columns shelf → X-axis or table columns
Color → colour encode
Size → size encode
Label → show value labels
Filter → filter data
Mark types: Bar, Line, Circle, Square, Text, Map, Pie
Show Me: automatically suggests best chart type
Q: Tableau calculations.
tableau
-- Calculated field (row-level, like a column)
Profit Margin = [Profit] / [Sales]
Age Group = IF [Age] >= 30 THEN "Senior" ELSE "Junior" END
Full Name = [First Name] + " " + [Last Name]
-- LOD (Level of Detail) Expressions — most advanced
-- FIXED: compute at specific dimension regardless of view
{ FIXED [Customer ID] : SUM([Sales]) } -- Customer total, not affected by view filters
-- INCLUDE: add dimension to aggregation
{ INCLUDE [Order ID] : COUNT([Product]) } -- Products per order
-- EXCLUDE: remove dimension from aggregation
{ EXCLUDE [Region] : AVG([Sales]) } -- National average regardless of region filter
-- Table calculations (computed across rows in viz)
RUNNING_SUM(SUM([Sales])) -- Running total
WINDOW_AVG(SUM([Sales])) -- Average across partition
RANK(SUM([Sales])) -- Rank within partition
Q: Tableau Server vs Tableau Online (Tableau Cloud).
Tableau Server: On-premise or cloud VM. Full control. IT managed.
Tableau Cloud (formerly Online): SaaS. Managed by Salesforce. Less setup.
Both: Publish workbooks, schedule data refreshes, manage permissions, web editing.
For data governance: Tableau Catalog (find, trust, and certify data sources).
Q: Performance optimisation in Tableau.
1. Use extracts (Tableau Data Extracts .tde/.hyper) instead of live connection
→ Queries run against local columnar store (much faster)
→ Schedule refreshes overnight
2. Reduce data: filter at source, aggregate before connecting
3. Avoid high-cardinality dimensions on Color/Shape/Label
→ Too many members = slow render
4. Use context filters: set filter as "context" to run before other filters
5. Limit marks: fewer marks = faster render (< 5000 marks recommended for charts)
6. Materialise LOD expressions: pre-compute and store as custom SQL or extract
7. Indexing in database: Tableau generates queries — ensure DB indexes match
Revision Notes
TABLEAU: Visual analytics platform. Drag-and-drop. VizQL generates SQL.
Dimensions (blue) = qualitative/slice | Measures (green) = quantitative/aggregate
SHELVES: Rows + Columns = axes | Color/Size/Label = encodings | Filter = filter
CALCULATIONS:
Regular: row-level formulas | Table calcs: across viz rows (RUNNING_SUM, RANK)
LOD (Level of Detail):
FIXED: specific dimension regardless of filters
INCLUDE: add dimension to aggregate
EXCLUDE: remove dimension from aggregate
TABLEAU SERVER vs CLOUD: Server=on-prem control | Cloud=SaaS managed
PERFORMANCE:
Use extracts (not live) | Reduce marks (<5000) | Context filters
Avoid high-cardinality on Color | Pre-aggregate data at source
