Documentation, findings, presentation and interview preparation
Communicate analysis so it is understood, trusted and reusable — documentation, findings memos, presentations — and prepare for data analyst interviews.
This lesson builds on: Working professionally — requirements, ownership, stakeholders and Agile delivery.
- Document analyses and data assets so others can reuse and audit them
- Write findings and recommendations memos with a clear structure
- Present analysis to executives and handle questions
- Prepare for technical and behavioural interview rounds
Documentation
Document so that someone else (or you in six months) can reproduce and trust the work:
| Artefact | Contents |
|---|---|
| Analysis README | Question, data sources and extraction date, definitions, method, assumptions, limitations, how to rerun |
| Metric definitions | Formula, filters, grain, owner, source (the KPI table from Module 4) |
| Data dictionary | Tables, columns, types, meanings (like the OrionStats dataset pages) |
| Code comments | Why, not what — -- exclude test stores (ids 900+) per Finance rule 2025-03 |
| Change log | What changed in a report and when |
Findings and recommendations
The pyramid principle: answer first, then supporting arguments, then data.
Subject: Ankara store — 2025 performance and plan for 2026
Answer: Ankara finished 2025 at 95% of plan; the shortfall reflects an
aggressive first-year plan more than weak execution. Recommend resetting
the 2026 plan to +12% on 2025 actuals and focusing on customer acquisition
and repeat visits.
Evidence:
1. Ankara grew 33% vs 2024 (its first full year), the fastest of all stores.
2. Months without prior-year history were planned with a flat assumption,
not a ramp-up curve …
3. Ankara's average basket (93.1) is above the network average (90.0);
the gap to plan is in order volume — a customer base still being built …
Limitations: one full year of history; plan assumptions not documented.
Next step: agree plan with Head of Retail by 15 January.
Presenting to executives
- Open with the answer and the decision you need.
- One message per slide, stated in the title ("Online growth lags stores: 7.6% vs 12.5%").
- Show only the numbers that support the message; keep detail in the appendix.
- Anticipate questions: data quality, definitions, "compared to what?", "what would change your view?"
- If you don't know, say so and follow up — credibility beats improvisation.
Interview preparation
Typical data analyst interview stages and how to prepare:
| Stage | What they test | Prepare |
|---|---|---|
| Screening | Motivation, communication | 60-second story: who you are, a project with impact |
| SQL test | Joins, aggregation, windows, dates | Practise writing queries on OrionStats databases under time pressure |
| Excel / BI task | Lookups, pivots, DAX basics, dashboard design | Build and explain a small dashboard |
| Case study | Structured thinking, business sense | Clarify → structure (issue tree) → hypotheses → analysis → recommendation |
| Statistics | Distributions, A/B tests, correlation vs causation | Explain p-values and confidence intervals in plain language |
| Behavioural | Ownership, conflict, mistakes | STAR stories (Situation, Task, Action, Result) |
| Portfolio review | Real work | 2–3 projects from the Project Lab with business framing |
Example case prompt: "Conversion on our website dropped 10% last week. What do you do?" — verify the metric and data (tracking change?), segment (device, channel, page, new vs returning), check timeline (deployments, campaigns, outages), quantify impact, propose next steps.
STAR example: Situation — monthly pack took 3 days; Task — reduce effort without losing accuracy; Action — automated two pages with SQL views and Power BI, added reconciliation checks; Result — delivery on day 2, two reporting errors caught by checks in the first quarter.
Practice
How should a findings memo to an executive begin?
In behavioural interviews, STAR stands for:
What should code comments in an analysis mainly explain?
Summary
- Document question, data, definitions, method, assumptions and limitations.
- Write answer-first memos with evidence, limitations and next steps.
- Present one message per slide; anticipate questions.
- Prepare for SQL, case, statistics and behavioural rounds with practice and STAR stories.