← Interview & exam prep
Technical interview questions
Data modeling interview questions
13 questions from beginner to advanced, each with the correct answer and a one-line explanation of why. Study them here, or take them as a timed test.
13 questionsBeginner 7 · Intermediate 6Answers + explanations
Beginner 7 questions
1A star schema consists of…
- Two fact tables joined directly
- Only one flat table
- Normalized 3NF tables
- Many linked fact tables
- A central fact table linked to dimension tables
Show answerHide answer
E · A central fact table linked to dimension tables A star has one fact table surrounded by descriptive dimension tables.
2Numeric, additive values (sales, quantity) belong in…
- A slicer
- A dimension table
- The Date table
- A measure table only
- The fact table
Show answerHide answer
E · The fact table Facts (measurable events) live in the fact table; descriptive attributes live in dimensions.
3The 'grain' of a fact table is…
- The level of detail one row represents
- Its refresh schedule
- Its sort order
- Its number of columns
- Its storage mode
Show answerHide answer
A · The level of detail one row represents Grain defines what a single row means (e.g., one row per order line).
4A snowflake schema differs from a star because its dimensions are…
- Normalized into several related tables
- Removed
- Stored as facts
- Bidirectional
- Denormalized into one table
Show answerHide answer
A · Normalized into several related tables Snowflaking normalizes dimensions into multiple related tables.
5A surrogate key is…
- A business/natural key
- A system-generated key with no business meaning
- A slicer
- A measure
- A relationship
Show answerHide answer
B · A system-generated key with no business meaning Surrogate keys are stable, model-generated identifiers used for relationships.
6The recommended relationship cardinality from a dimension to a fact is…
- No relationship
- Many-to-many
- Bidirectional by default
- One-to-many (single direction)
- One-to-one
Show answerHide answer
D · One-to-many (single direction) One dimension row relates to many fact rows, filtering in a single direction.
7A dedicated Date dimension is recommended so that…
- Files are smaller
- RLS is enabled
- Refresh is faster
- Visuals are prettier
- Time-intelligence and consistent date attributes work
Show answerHide answer
E · Time-intelligence and consistent date attributes work A marked Date table gives a continuous calendar for time-intelligence and reusable date attributes.
Intermediate 6 questions
8A role-playing dimension (Date used for Order and Ship date) is handled with…
- Deleting one date
- Inactive relationships + USERELATIONSHIP, or duplicate date tables
- A bridge table
- Bidirectional filtering
- One relationship only
Show answerHide answer
B · Inactive relationships + USERELATIONSHIP, or duplicate date tables Keep one active relationship and activate the other per-measure with USERELATIONSHIP, or use a second date table.
9A true many-to-many between two dimensions is best resolved with…
- Bidirectional filters everywhere
- Merging them into one
- A calculated column
- A bridge (junction) table
- Removing the relationship
Show answerHide answer
D · A bridge (junction) table A bridge table models the M:N relationship cleanly and predictably.
10A degenerate dimension is…
- An unused table
- A dimension attribute stored in the fact table (e.g., order number)
- A snowflake
- A measure
- A broken relationship
Show answerHide answer
B · A dimension attribute stored in the fact table (e.g., order number) Degenerate dimensions (like transaction IDs) live in the fact table with no separate dimension.
11Bidirectional cross-filtering is risky because it can cause…
- Smaller files
- Automatic RLS
- Ambiguous filter paths and performance issues
- Faster refresh
- Better compression
Show answerHide answer
C · Ambiguous filter paths and performance issues Two-way filters can create ambiguity and slow queries; use sparingly and deliberately.
12A Slowly Changing Dimension Type 2 preserves history by…
- Deleting old rows
- Storing history in the fact
- Overwriting the value
- Ignoring changes
- Adding a new row (versioned) for each change
Show answerHide answer
E · Adding a new row (versioned) for each change SCD Type 2 inserts a new versioned row so historical facts keep their original attribute values.
13A conformed dimension is one that…
- Belongs to a single report
- Cannot be filtered
- Has no keys
- Is always bidirectional
- Is shared consistently across multiple fact tables
Show answerHide answer
E · Is shared consistently across multiple fact tables Conformed dimensions (e.g., a shared Date or Customer) allow consistent analysis across facts.
What changed lately? Latest Data Modeling news, releases & questions →
More technical interview questions