How to Approach the BBIM601 Business Intelligence Project: A Step-by-Step Plan for Assessment 2
In short: BBIM601 Assessment 2 rewards a clear chain from business problem to stakeholder, KPI, analysis, visual and recommendation. Plan Part A (the RetailCo application, 56 marks) and Part B (the hotel and cardiovascular reports, 44 marks) as one connected project. Show your method, evidence every number with your own screenshots, and make each recommendation traceable to a chart.
Need the exact wording of the task? See the BBIM601 Assessment 2 question page. For a worked method with illustrative figures, read the BBIM601 reference sample, which is for reference only.
If you want expert eyes on your plan before you build it, our coursework help page explains how to share your brief.
What Assessment 2 asks you to do
This is the project worth 60% at NZSE (Bachelor of Business Information Management, Level 6). It tests learning outcome 3 (applying BI techniques and methodology to design a BI application) and learning outcome 4 (using BI tools to prepare reports across application areas). Markers look for justified choices, not tool screenshots on their own.
| Part | Marks | What markers want to see |
|---|---|---|
| A1 Stakeholders | 8 | At least two roles, each tied to a decision and a data need |
| A2 KPIs | 12 | Three KPIs with definition, reason and owner |
| A3 RFM | 14 | Scoring logic, two or more segmentation options, comparison and run evidence |
| A4 Data cube and OLAP | 14 | Schema with facts and dimensions, plus one operation that answers a business question |
| A5 BI cycle | 8 | Each component mapped to a stage, with challenges and responses |
| B1–B2 Hotel booking | 20 | Structure, quality issues, transformations and an interactive report with recommendations |
| B3–B5 Cardiovascular | 24 | Structure, BMI categories, rates by group, risk profile dashboard and insights |
Step-by-step plan
- Read the datasets before the questions. Open RetailCo Sales, hotel_booking and cardio_disease and note rows, columns, types and obvious gaps. Everything later depends on this.
- Write the stakeholder table first. Pick two or three roles, such as a regional sales manager and a marketing manager, and give each one decision. Your KPIs should flow from these decisions.
- Define three KPIs precisely. State the formula, the time grain and the owner. One KPI should address the retention problem and one the regional differences.
- Build RFM in two stages. Calculate recency, frequency and monetary value per customer, score them, then try at least two segmentation rules and compare them. Explain which you would use and why.
- Design the star schema on paper. One fact table, several dimensions, clear keys. Then choose one OLAP operation (slice, dice or drill-down) that answers a named question, such as which region grew fastest in a quarter.
- Clean the hotel data and document each decision. Missing values, data types, outliers, total stay nights and lead-time bands all need a one-line justification.
- Build the interactive report. Use slicers, keep visuals purposeful and add a short summary with recommendations.
- Repeat for the cardiovascular data. Group features, calculate BMI with the four categories in the brief, compare rates by group and add a gender filter.
- Close the loop in the BI cycle section. Show how each step supports the cycle and what went wrong or could go wrong.
- Proofread against the brief. Tick every numbered task and confirm each has a screenshot, a caption and a sentence of interpretation.
Choosing tools without losing marks
The brief asks for at least two tools from a list that includes Power BI, Tableau, Python, R and Looker. A sensible pairing is Python for cleaning and RFM scoring and Power BI for interactive reports. Say why each tool suits its task. For instance, Python makes scoring steps transparent, while Power BI slicers make a report interactive for non-technical stakeholders.
Turning each output into a marked paragraph
Many students run the analysis correctly and still lose marks because the write-up stops at the screenshot. A simple three-part pattern fixes this. First, state what you did and why, in one sentence. Second, point to the evidence, such as a table or a chart caption. Third, explain what it means for the stakeholder you named earlier. For example, if your slicer shows that cancellation rates differ between the resort and city hotels, say which manager would act on that and what they might change, such as deposit rules or overbooking limits. Keep your own numbers in the sentence so the marker can trace the claim to the visual.
A realistic time plan
A 60% project deserves several weeks of steady work rather than a last-minute push. A workable split is to spend the first week on data profiling, stakeholders and KPIs, the second on RFM and the data cube, the third on the hotel and cardiovascular cleaning and reports, and the final days on the BI cycle discussion, formatting and proofreading. Build in a buffer for tool problems, since refreshing a large file or fixing a broken relationship can take longer than expected. Save versions of your Power BI file and your scripts so that you can recover if something breaks.
Presenting screenshots and dashboards professionally
Crop screenshots so that the relevant visual, filter pane and field list are readable. Number every figure and table, refer to it in the text and write a caption that states the question the visual answers. For dashboards, keep consistent colours for the same category across pages, label axes, and use sensible sorting, such as months in calendar order and the top five countries in descending order. A short note under each page that explains how to use the filters shows that you have designed the report for stakeholders and not only built it.
Handling uncertainty and limitations
Good business intelligence reports admit what the data cannot show. If the RetailCo file has no cost field, say that margin cannot be calculated and suggest what would be needed. If the hotel dataset covers only two properties, note that findings may not generalise. If the cardiovascular file contains implausible blood pressure values, explain how many records you removed and how that could affect the rates. A short limitations paragraph in each part shows critical thinking and often earns marks that a perfect-looking dashboard cannot.
Mark-maximising checklist
- Every KPI has a formula, a purpose and an owner.
- RFM shows both the scoring method and at least two segmentation options.
- The schema diagram labels the fact table, dimensions and keys.
- The OLAP operation is named and linked to a business question.
- Every data quality issue has an action and a reason.
- All screenshots are readable, captioned and referred to in the text.
- BMI categories match the cut-offs given in the brief exactly.
- Insights are worded as associations, not proven causes.
- Recommendations are specific, such as which customers to contact or which month to target.
Common mistakes to avoid
Presenting numbers you did not calculate
Never copy example figures into your report. Use your own outputs and show how you produced them.
Skipping the reasoning behind KPIs
Listing revenue, profit and orders without linking them to a stakeholder decision loses most of the marks on that task.
Treating the cleaning step as optional
Markers expect to see what was wrong with the data and what you did about it. Silent fixes look like missed issues.
Overloading dashboards
A page with a dozen visuals hides the story. Group visuals by question and keep titles descriptive.
Confusing association with causation
Higher cardiovascular prevalence in one group does not prove the group characteristic caused it. Use careful wording.
Frequently asked questions
How long should the Assessment 2 report be?
Follow the word and format guidance in your course materials. Aim for concise sections where each task has a heading, a short method note, evidence and interpretation.
Do I have to use Power BI?
No. The brief allows several tools, but you must use at least two. Power BI is a common choice for the interactive report because of its slicers and visuals.
What counts as a good segmentation option?
One that is rule-based, repeatable and linked to an action. Score bands and behavioural rules both work if you explain how each segment would be treated differently.
How many screenshots do I need?
Include enough to prove each task was done in the tool: data profile, transformation, key visuals and the dashboard with filters applied. Caption each one.
Can I use the same insight in the hotel and cardiovascular sections?
No. Each dataset needs its own insights, tied to its own visuals and business or health context.
Where do I start if I am behind schedule?
Start with data cleaning and the stakeholder and KPI tables. These are quick wins and they structure the rest of the project.
Ready to work through your own datasets? Share your brief with our team, and see what students say on our reviews page.
About the expert
Guidance reviewed by Dr. Sophie McAllister, Managerial Finance & Business Analytics Specialist (Victoria University of Wellington, as shown on her profile).
