Business case previews¶

Five business problems, one for each department. Each case teaches a core skill and a common mistake to catch.

---

The five core cases¶

Retail¶

Decision: Which customer segment deserves a closer investigation?

You will practise: what a row means, distinct orders, average order value, Sales and Profit, ratios, discounts, customers, trends and the limits of a returns list.

Mistake to catch: treating order lines as orders.

Finance¶

Decision: Which departments need a spending review?

You will practise: budget versus actual, time periods, reconciliation and whether a cost difference is favourable.

Mistake to catch: repeating a monthly budget on every transaction row.

Operations¶

Decision: Where should delivery reliability improve?

You will practise: promised dates, overdue open shipments, distributions, median and mean.

Mistake to catch: dropping undelivered shipments from the service measure.

Marketing¶

Decision: Which campaigns need closer evaluation before budget moves?

You will practise: weighted funnel rates, cost per conversion, return on ad spend and the limits of attribution.

Mistake to catch: averaging percentages that have different denominators.

HR¶

Decision: How is the workforce changing, and which comparisons are fair?

You will practise: historical headcount, turnover, cohorts, tenure and recruitment timelines. All employee records are synthetic.

Mistake to catch: dividing past exits by the current headcount.

---

What every case includes¶

• Stakeholder brief: the decision and why it matters
• Data dictionary: what each field means
• Metric definitions: exact formulas and scope
• Worked example: a guided walk-through
• Exercises: guided and independent practice
• Debugging task: find and fix a mistake
• Recommendation challenge: advise the stakeholder
• Validation checkpoints: confirm your numbers

Each case also has an optional pandas connection. Instructor answers are supplied separately.

---

Combining evidence across departments¶

Before combining findings, explain any differences in grain, period and units.

• Within the synthetic company, marketing-attributed revenue is a view of the same retail revenue. Do not count it twice.
• Separate catalog datasets do not describe that fictional company.
• A good recommendation weighs trade-offs. It does not add incompatible KPIs together.

---

Additional practice choices¶

You will also be able to practise on topics beyond the five departments:

Topic	Example dataset
Sports	World Cup match results
Technology	Mobile operating-system usage (a data-quality exercise), start-up funding
Government	US construction spending, US home sales
Other	Science, education and entertainment datasets

These are transfer exercises: the same process, with source-specific definitions and checks. They fit into your existing practice time or optional extensions. The five core cases remain the foundation of your portfolio.