The one paragraph that does the work: an analysis story
A hiring manager for an analyst role wants to see how you think. Tell one analysis end to end.
- 1Question
What did the business need to know? "Why did repeat orders fall 9%?"
- 2Data
What you pulled and how big it was: 2.1M order rows plus delivery times.
- 3Method
The analysis in plain words: SQL cohorts of repeat rate by wait time.
- 4Decision
What someone did because of it: a delay coupon in three cities. No job yet? The recommendation you would make.
- 5Number
What moved and how fast: repeat rate back to its old level in four weeks.
Find your own numbers
Most people say they have no numbers. Ask these five questions about your best piece of work, and write down only what you could defend in the interview.
| Kind | Ask yourself | Example (sample) | No job yet? |
|---|---|---|---|
| Scale | How many rows, orders, users or regions did the work touch? How many people use the output? | 2.1M order rows; 40 regional managers use the dashboard | Rows in the dataset, rows you cleaned or dropped |
| Time | How long did the task take before your change, and after? | Weekly reporting went from 6 hours to 30 minutes | How long a manual version would take vs your query or dashboard |
| Change | Which metric moved after the decision, by how much, in how long? | Repeat rate back to its old level within 4 weeks | The size of the gap you found, e.g. 11 points between two groups |
| Money | Did the analysis find, save or protect money? Use the figure finance would agree with. | ₹14L of duplicate vendor payments found | Usually none. Leave it out rather than invent one |
| Decision | What did a manager do differently because of your numbers? | Ops launched a delay coupon in three cities | The recommendation you would make, or the test you would run next |