Data Management and Business Analytics
What to Know
Candidates should understand the data lifecycle: capture, validation, storage, integration, analysis, retention, and disposal. Know relational concepts - tables, keys, normalization, referential integrity - and when denormalization or a data warehouse or data mart is appropriate. Distinguish OLTP (transactional, normalized, current) from OLAP (analytical, dimensional, historical). Understand ETL versus ELT, data quality dimensions (accuracy, completeness, consistency, timeliness, uniqueness), metadata, master data, and data lineage. Analytics types: descriptive (what happened), diagnostic (why), predictive (what will happen), prescriptive (what to do). Know basic statistics - mean, median, standard deviation, correlation versus causation, sampling bias - and visualization principles such as choosing chart type by question and avoiding misleading axes. Recognize data governance roles: owner, steward, custodian.
Traps
Correlation presented as causation is the most common distractor. Questions may call a dashboard predictive when it only reports history. Another trap confuses data owner (accountable for the data domain) with data custodian (technical safekeeping). Watch for claims that a data warehouse is optimized for transaction speed, or that normalization improves analytical query performance - usually the opposite. Also be alert to survivorship or selection bias hidden in a sampling description.
Worked Example
A sales dashboard shows revenue rising while average order value falls. Rule of thumb: decompose a ratio into numerator and denominator before concluding. Revenue = orders × average order value, so rising revenue with falling average value implies order count grew faster. That is descriptive and diagnostic analysis; predicting next quarter requires a model and validation, not extrapolation from two points.
Exam Tip
For any metric question, write the formula and check which component moved.