Certified Specialist Business Intelligence (CSBI)
Practice Exam - Aligned to 2026-2027 Industry
Standards | Answer Key & Full Rationales | HFMA
Certification
Domain 1 – Data Management & Governance (28 questions)
Q1
As the newly-hired BI architect for a 500-store retailer, you must support 5 years of daily
POS data (≈ 9 billion rows) and frequent year-over-year queries. The CIO insists the
model must be business-friendly and performant. The MOST appropriate schema is:
A. Third-normal-form (3NF) relational design
B. Star schema with conformed dimensions
C. Data Vault 2.0 ensemble model
D. Fully denormalized flat table
Correct Answer: B
Rationale: Kimball’s star schema is purpose-built for large-scale, time-series analytics;
conformed dimensions guarantee semantic consistency across business processes.
3NF (A) favours OLTP integrity, not OLAP performance. Data Vault (C) is excellent for
,agile EDW backbones but requires additional marts for consumption. Flat tables (D)
create update anomalies and bloat storage.
Q2
A global bank needs to identify duplicate client records across 12 source systems after
3 M&A transactions. Which master-data technique BEST ensures a single golden record
while preserving historical lineage?
A. Stochastic matching + survive rules in a MDM hub
B. SQL DISTINCT de-duplication in the staging layer
C. Source-system survivorship with “last-update-wins”
D. Blockchain-based distributed ledger
Correct Answer: A
Rationale: MDM hubs with probabilistic matching, survivorship rules and lineage
metadata are the DAMA-recommended practice for multi-source customer data.
DISTINCT (B) ignores fuzzy variations. Last-update-wins (C) loses earlier truth.
Blockchain (D) is immutable but not optimal for fuzzy matching.
Q3
A hospital’s data-quality scorecard shows 4 % of patient records have invalid ZIP codes.
Which data-profiling step is the IMMEDIATE next action in the governance process?
A. Define a business rule in the data-quality SLA
B. Root-cause analysis with column-pattern profiling
,C. Mask the ZIP column for HIPAA compliance
D. Delete records to improve the score
Correct Answer: B
Rationale: DAMA-DMBOK prescribes root-cause profiling before remediation. Profiling
reveals whether the issue is upstream system entry, reference-data gaps, or ETL
mapping errors. SLA definition (A) comes after understanding causes. Deletion (D)
destroys clinical history and violates retention rules.
Q4
Under GDPR, a EU airline is asked to “forget” a frequent flyer. Which technical approach
guarantees lawful erasure while maintaining analytical referential integrity?
A. Physical delete every row across source & mart layers
B. Pseudonymise the key then keep facts for aggregation
C. Soft delete + cascading tombstones with audit trail
D. Move the customer record to an off-line backup tape
Correct Answer: C
Rationale: GDPR (Art. 17) allows continued processing for legitimate interests if the
data subject is no longer identifiable. Soft delete with tombstones keeps referential
integrity and audit evidence. Pure deletion (A) breaks fact tables. Backup tape (D) is not
searchable for future requests.
Q5
, A SaaS vendor models subscriptions in a snowflake schema. The finance team
complains that month-end reports take 45 minutes. Which optimisation is MOST
effective while keeping the snowflake?
A. Create bitmap indexes on low-cardinality keys
B. Build an aggregate table at month grain
C. Collapse dimensions into one wide dimension
D. Move the database to in-memory OLTP engine
Correct Answer: B
Rationale: Aggregate navigation (monthly pre-join) directly addresses the query-time
pain without remodeling. Bitmap indexes (A) help but are secondary. Collapsing (C)
effectively converts to star and loses snowflake benefits. OLTP engine (D) is optimised
for transactions, not reporting.
Q6
A data steward finds that “Revenue” is calculated three different ways in Tableau, Power
BI and Excel. Which governance artefact FIRST resolves this?
A. Data dictionary with canonical business definitions
B. Role-based access control (RBAC) matrix
C. Data-quality firewall at database entry
D. Adoption of an enterprise visualization standard
Correct Answer: A
Practice Exam - Aligned to 2026-2027 Industry
Standards | Answer Key & Full Rationales | HFMA
Certification
Domain 1 – Data Management & Governance (28 questions)
Q1
As the newly-hired BI architect for a 500-store retailer, you must support 5 years of daily
POS data (≈ 9 billion rows) and frequent year-over-year queries. The CIO insists the
model must be business-friendly and performant. The MOST appropriate schema is:
A. Third-normal-form (3NF) relational design
B. Star schema with conformed dimensions
C. Data Vault 2.0 ensemble model
D. Fully denormalized flat table
Correct Answer: B
Rationale: Kimball’s star schema is purpose-built for large-scale, time-series analytics;
conformed dimensions guarantee semantic consistency across business processes.
3NF (A) favours OLTP integrity, not OLAP performance. Data Vault (C) is excellent for
,agile EDW backbones but requires additional marts for consumption. Flat tables (D)
create update anomalies and bloat storage.
Q2
A global bank needs to identify duplicate client records across 12 source systems after
3 M&A transactions. Which master-data technique BEST ensures a single golden record
while preserving historical lineage?
A. Stochastic matching + survive rules in a MDM hub
B. SQL DISTINCT de-duplication in the staging layer
C. Source-system survivorship with “last-update-wins”
D. Blockchain-based distributed ledger
Correct Answer: A
Rationale: MDM hubs with probabilistic matching, survivorship rules and lineage
metadata are the DAMA-recommended practice for multi-source customer data.
DISTINCT (B) ignores fuzzy variations. Last-update-wins (C) loses earlier truth.
Blockchain (D) is immutable but not optimal for fuzzy matching.
Q3
A hospital’s data-quality scorecard shows 4 % of patient records have invalid ZIP codes.
Which data-profiling step is the IMMEDIATE next action in the governance process?
A. Define a business rule in the data-quality SLA
B. Root-cause analysis with column-pattern profiling
,C. Mask the ZIP column for HIPAA compliance
D. Delete records to improve the score
Correct Answer: B
Rationale: DAMA-DMBOK prescribes root-cause profiling before remediation. Profiling
reveals whether the issue is upstream system entry, reference-data gaps, or ETL
mapping errors. SLA definition (A) comes after understanding causes. Deletion (D)
destroys clinical history and violates retention rules.
Q4
Under GDPR, a EU airline is asked to “forget” a frequent flyer. Which technical approach
guarantees lawful erasure while maintaining analytical referential integrity?
A. Physical delete every row across source & mart layers
B. Pseudonymise the key then keep facts for aggregation
C. Soft delete + cascading tombstones with audit trail
D. Move the customer record to an off-line backup tape
Correct Answer: C
Rationale: GDPR (Art. 17) allows continued processing for legitimate interests if the
data subject is no longer identifiable. Soft delete with tombstones keeps referential
integrity and audit evidence. Pure deletion (A) breaks fact tables. Backup tape (D) is not
searchable for future requests.
Q5
, A SaaS vendor models subscriptions in a snowflake schema. The finance team
complains that month-end reports take 45 minutes. Which optimisation is MOST
effective while keeping the snowflake?
A. Create bitmap indexes on low-cardinality keys
B. Build an aggregate table at month grain
C. Collapse dimensions into one wide dimension
D. Move the database to in-memory OLTP engine
Correct Answer: B
Rationale: Aggregate navigation (monthly pre-join) directly addresses the query-time
pain without remodeling. Bitmap indexes (A) help but are secondary. Collapsing (C)
effectively converts to star and loses snowflake benefits. OLTP engine (D) is optimised
for transactions, not reporting.
Q6
A data steward finds that “Revenue” is calculated three different ways in Tableau, Power
BI and Excel. Which governance artefact FIRST resolves this?
A. Data dictionary with canonical business definitions
B. Role-based access control (RBAC) matrix
C. Data-quality firewall at database entry
D. Adoption of an enterprise visualization standard
Correct Answer: A
