Foundations of Business Intelligence
Foundations of Business Intelligence tests whether you can define BI, place it in an organization, and translate a business question into a measurable, data-backed answer. Expect questions on roles, data types, analytics maturity, and the difference between a metric, a KPI, and a dashboard.
What BI actually is
BI is the people, processes, and technologies that turn raw data into information for decisions. It is mostly descriptive and diagnostic, recurring, and decision-supporting. Data engineering builds pipelines; data science predicts; BI explains and monitors. The analytics maturity ladder runs descriptive → diagnostic → predictive → prescriptive, and BI owns the first two.
Questions before data
Start with the decision, not the dataset. A BI brief names the decision owner, the question, metric definition, grain, time window, comparison, and threshold. Map stakeholders: sponsor, decision maker, and user. A dashboard without a named decision is decoration.
Metric discipline
A metric is a numeric measure. A KPI is a metric tied to an objective, with an owner and cadence. Good KPIs are specific, measurable, actionable, relevant, and time-bound. Leading indicators predict; lagging indicators confirm. Every definition must state numerator, denominator, filters, grain, source, and owner. Vanity metrics look good but do not change behavior.
Data types and sources
Structured data lives in tables; semi-structured data includes JSON and logs; unstructured data includes text. Sources include transactional systems, CRM, web analytics, surveys, and external data. Judge quality on completeness, accuracy, consistency, timeliness, validity, and uniqueness.
Worked example
A subscription company asks, 'Are we growing?' That is not decision-ready. Reframe: the decision is whether to increase marketing spend; the question is 'What is monthly net revenue retention by cohort for the last 12 months?'; the metric is NRR = (starting MRR + expansion − contraction − churn) ÷ starting MRR; grain is customer-month; sources are billing and CRM; the threshold is NRR below 100% for two consecutive months; the owner is the VP of Customer Success; cadence is monthly. The reasoning: a named decision sets the threshold, the metric definition prevents denominator drift, the grain prevents mixing customer and month rows, and the owner makes the result actionable.
Traps that cost marks
- Calling every metric a KPI.
- Picking a chart before defining the question.
- Omitting the denominator or filter context.
- Mixing grains in one visual.
- Assuming more data is always better.
- Ignoring data quality and presenting a report as a decision.